University of Southern California Dissertations and Theses

Biosocial antecedents of schizophrenia-spectrum personality disorders: A longitudinal study

(USC Thesis Other)

Biosocial antecedents of schizophrenia-spectrum personality disorders: A longitudinal study

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content INFORMATION TO USERS This manuscript has been reproduced from the microfilm master. UMI films the text directly from the original or copy submitted. Thus, some thesis and dissertation copies are in typewriter face, while others may be from any type of computer printer. The quality of this reproduction is dependent upon the quality of the copy submitted. Broken or indistinct print, colored or poor quality illustrations and photographs, print bleedthrough, substandard margins, and improper alignment can adversely affect reproduction. In the unlikely event that the author did not send UMI a complete manuscript and there are missing pages, these will be noted. Also, if unauthorized copyright material had to be removed, a note will indicate the deletion. Oversize materials (e.g., maps, drawings, charts) are reproduced by sectioning the original, beginning at the upper left-hand comer and continuing from left to right in equal sections with small overlaps. Each original is also photographed in one exposure and is included in reduced form at the back of the book. Photographs included in the original manuscript have been reproduced xerographically in this copy. Higher quality 6” x 9” black and white photographic prints are available for any photographs or illustrations appearing in this copy for an additional charge. Contact UMI directly to order. UMI A Bell & Howell Information Company 300 North Zeeb Road, Ann Arbor MI 48106-1346 USA 313/761-4700 800/521-0600 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. BIOSOCIAL ANTECEDENTS OF SCHIZOPHRENIA-SPECTRUM PERSONALITY DISORDERS: A LONGITUDINAL STUDY by Deana S. Benishay A Dissertation Presented to the FACULTY OF THE GRADUATE SCHOOL UNIVERSITY OF SOUTHERN CALIFORNIA In Partial Fulfillment of the Requirements for the Degree DOCTOR OF PHILOSOPHY (Clinical Psychology) August 1996 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. UMI Number: 9705074 Copyright 1997 by Benishay, Deana S. All rights reserved. UMI Microform 9705074 Copyright 1996, by UMI Company. All rights reserved. This microform edition is protected against unauthorized copying under Title 17, United States Code. UMI 300 North Zeeb Road Ann Arbor, MI 48103 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. UNIVERSITY OF SOUTHERN CALIFORNIA THE GRADUATE SCHOOL UNIVERSITY PARK LOS ANGELES. CALIFORNIA 90089-4015 This dissertation, written by Deana S . B enishay under the direction of h i Dissertation Committee, and approved by all its members, has been presented to and accepted by The Graduate School, in partial fulfillment of re quirements for the degree of DOCTOR OF PHILOSOPHY Dean of Graduate Studies Date DISSERTATION COMMITTEE Chairperson Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Acknowledgments The efforts of Cyril Dalais, Brian Bell, Athenea Chiriaca, Kavi Beepat, and all the local members of the longitudinal study, the Mauritius Joint Child Health Project, are gratefully acknowledged. The author also extends gratitude to Adrian Raine D.Phil., John Brekke, Ph.D., Michael Dawson, Ph.D., Mitchell Earleywine, Ph.D., Samoff Mednick, Ph.D., Peter Venables, D.Phil. and Joseph Ventura, Ph.D. for their helpful comments. The author also gratefully acknowledges the comments and support of Todd Lencz, Ph.D. The data used in this study were collected while the project was supported by the Medical Research Council (UK), the Wellcome Trust (UK), and the Mauritian Government. The author was partially supported by NIMH grant MH10845-01. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table of Contents 1 . Acknowledgments................................................................................................. ii 2. List of Tables.......................................................................................................... iv 3. List of Figures....................................................................................................... v 4. List of Appendices.................................................................................................. vi 5. Abstract...................................................................................................................vii 6. Introduction............................................................................................................1 a. Aims............................................................................................ 1 b. Background and Significance....................................................... 10 7. Methods.................................................................................................................30 a. Subjects............................................................................ 30 b. Design.............................................................................. 31 c. Data Analysis................................................................... 38 8. Results....................................................................................................................47 a. Summary of Results......................................................... 76 9. Discussion............................................................................................................ 83 10. Literature Citations............................................................................................. 135 11. Appendices.......................................................................................................... 152 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. List of Tables Table 1 Significant findings for regression equations Y=b0 +btxt............................... 52 Table 2 Significant findings in additive model Y=b0 +b1 x1 + b-,x-.................................. 63 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. List of Figures Figure la Model 1................................................................................................... 3 Figure lb Model II................................................................................................. 4 Figure lc Model III............................................................................................... 4 Figure Id Model IV............................................................................................... 5 Figure le Model V................................................................................................ 6 Figure 2 Comparison of full scale IQ. VIO. and PIO for schizophrenism index and comparison groups................................................................................................ 50 Figure 3a-b Scatterplot of the relationship between age 11 full-scale 10 scaled scores and age 16 schizophrenism.......................................................................................... 52-53 Figure 4a-b Scatterplot of the relationship between age 1 1 verbal 10 scaled scores and age 16 schizophrenism................................................................................................ 54 Figure 5a-b Scatterplot of the relationship between age 11 performance IQ scaled scores and age 16 schizophrenism................................................................................... 55 Figure 6a-c Relationship between age 16 schizophrenism and ZPIO. the standardized residual of PIO regressed on VIO......................................................................... 59-60 Figure 7 Three regression lines showing the relationship between age 1 1 ZPIO and age 16 schizophrenism at three different levels of the Family Stress Index fFSD............. 73 Figure 8 “Fully dimensional” model of SSPD as described bv Claridge & Beech (19953......................................................................................................... 101 Figure 9a-b Venn diagrams represent variance of schizophrenism accounted for bv 10 measures and SES in multiple regression..............................................................110-111 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. List of Appendices Appendix 1 Age 3 cognitive measure............................ Appendix 2 Family Stress Index................................... Appendix 3 Survey of Attitudes and Experiences......... Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Abstract vu This study uses a longitudinal design on general population subjects to examine the role of intellectual deficits in the development of schizophrenia-spectrum personality disorders (SSPDs). Predictions were based on Meehl’s model (1962, 1990) of a neurocognitive genetic defect, as well as four variant models suggesting possible additive and interactive effects of stress. The neuro-cognitive measures utilized in the study were collected on 1795 Mauritian children (non-patient sample) and include a cognitive measure based on the Boehm Test of Basic Concepts-Preschool Version (age 3), as well as subtests of the WTSC-R administered at age 11 that were scaled and made into indices of performance, verbal and total IQ. Psychosocial variables include a family stressor index (FSI) based on data at ages 3 and 11, and a socioeconomic status index (age 3) based on the Hollingshead rating scale (Hollingshead and Redlich, 1958). Analyses of variance (ANOVAs) and multiple regression analyses were used to compare these variables' relationships to self- report measures at age 16 (Survey of Attitudes and Experiences, Venables, et al., 1990) and clinical (age 20-22) measures of SSPD (PDE; Loranger, 1988), which provide both categorical and dimensional indices. Findings generally supported predictions by Meehl’s hypothesis: cognitive deficits in childhood related to higher levels of SSPD. However, this relationship was significant only when SSPD was measured by the schizophrenism scale at age 16, and only with IQ variables collected at age 11. Specifically, ANOVAs showed that subjects scoring high on Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. viii schizophrenism had lower cognitive test scores at age 11 (total IQ, performance IQ and the discrepancy score between verbal and performance IQ), and multiple regression analyses also revealed a strong relationship between lower IQ at age 11 (total IQ, verbal IQ, performance IQ, and the discrepancy score between VIQ and PIQ) and higher schizophrenism. Results also demonstrated that sex did not moderate these interactions. Tests of the second hypothesis also generally favor Meehl’s model, suggesting that in most cases, the psychosocial stressors measured in the present study do not provide additive, moderating, or mediating effects for the relationship between cognitive deficits and later SSPD. However, results revealed two notable exceptions: (1) SES was shown to have a significant additive effect with the verbal-performance IQ discrepancy score (ZPIQ index) in the prediction of schizophrenism, and (2) findings revealed an interaction effect such that at the lowest Family Stress Index scores, ZPIQ had the strongest predictive value on later schizophrenism, and at levels of high family stress, ZPIQ had virtually no predictive value. These findings indicate that the VIQ-PIQ differential score (ZPIQ) may be a better measure of the genetic vulnerability to SSPD, and possibly schizophrenia. The use of a primarily non-hospitalized and unmedicated sample that is balanced for gender and is ethnically diverse allowed the results obtained from this study to have implications for the early identification of risk factors for schizotypy, and, by implication, schizophrenia, which would allow for prediction and intervention. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Introduction Aims The two main aims of this study serve as a replication and an extension of previous research on vulnerability markers for schizophrenia-spectrum disorders. These aims are based on Meehl’s theory (1962, 1990) which asserts that a single genetic vulnerability, termed “schizotaxia,” results in an “integrative neural defect,” or an early neuro- developmental abnormality of the central nervous system (CNS); Meehl hypothesized that such a defect underlies schizophrenia, schizotypal personality disorder (SPD), and, more broadly, schizophrenia-spectrum personality disorders (SSPDs; Webb and Levinson, 1991; Kendler & Gruenberg, 1982; Gunderson et al., 1983). Due to this shared theoretical relationship to schizophrenia, SPD and SSPD will be used interchangeably throughout this paper. However, the component traits of SSPD, including “positive” symptoms such as schizophrenism, and “negative” symptoms such as anhedonia, will be examined and discussed separately to explore potential heterogeneity within the diagnostic category. Proponents of Meehl’s theory (Comblatt & Keilp, 1994; Holzman et al., 1995, Lenzenweger et al., 1991) further suggest that certain cognitive deficits, such as reduced intellectual functioning, can be considered markers of the genetic vulnerability to schizophrenia. In fact, schizophrenics have consistently shown reduced neuro-cognitive test scores as compared to matched controls (e.g., Aylward, et al., 1984; Bilder et al., 1992). However, there has been some controversy concerning whether the deficit predates schizophrenia symptoms (e.g. Bilder et al., 1992; Erlenmeyer-Kimling & Comblatt, 1987). Some have suggested that the deficit is neurodegenerative in origin (Payne, 1961; Hamlin, Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1969; Bilder et al., 1992; Bilder et al., 1991); others argue that it is neurodevelopmental and predates the onset of psychosis (e.g., Bilder et al., 1992; Erlenmeyer-Kimling et al., 1991; Lane & Albee, 1968). Other studies have shown evidence of cognitive deficits in SSPDs (Lencz et al., 199S). Based on these findings, the present study’s first main aim is to explore the association between decreased intellectual functioning in childhood and schizophrenia-spectrum personality disorders (SSPDs) in adolescence and adulthood. The second main aim of this study is to examine the role psychosocial stress plays in the relationship between intellectual test performance and later personality functioning. Because of the general paucity of literature on biosocial models of SSPD, this study aims to explore hypotheses suggested by five competing biosocial models. The first model (Model I; see Figure la) predicts that psychosocial stress does not provide an additive effect or an interaction with cognitive performance in the development of SSPD. This prediction is based on Meehl’s conceptualization of the role of the “integrative neural defect” in the development of schizophrenia. By this model, individuals with SSPDs would show decreased intellectual functioning and other signs of a neuro-cognitive deficit (e.g., IQ), but lack the psychosocial stressor(s) that cause(s) the development of psychosis. Models II (see Figure lb) and HI (see Figure lc) are based on an alternate diathesis-stress model proposed by Fowles (1992) and Gottesman & Shields (1982). Their theory promotes a multifactorial polygenic etiology for schizophrenia, in which an individual becomes schizophrenic after a certain threshold of genetic diathesis and environmental stress is attained. Similarly, an individual who manifests an SSPD can also show signs of the diathesis and the environmental stress. Thus, the development of SSPD Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. may parallel the development of schizophrenia; individuals may have to reach a certain threshold of vulnerability and stress to become schizotypal. While Models II and HI are both based on this diathesis-stress model, Model II predicts that environmental stress will provide only an additive effect to the development of SSPD, whereas Model III predicts that stress will serve as a moderator such that higher stress and lower IQ interact to predict SSPD. FIGURE la - MODEL T GENE -» NEURO. DEFECT-* SSPD~> SCHIZOPHRENIA / / ENVIRON. POLYGENIC STRESSORS POTENTIATORS Figure 1. Graphical depiction of five models of the relationship between genes, cognitive deficits, environmental stressors, schizophrenia spectrum personaity disorder (SSPD), and schizophrenia. Figure la. Meehl’s (1962, 1990) model, in which the genetic effect of a single major locus (the schizogene) leads directly to a “neurointegrative defect,” which may be manifest phenotypically in intellectual deficits in childhood. This defect leads directly to SSPD, regardless of environmental circumstances. In the presence of various environmental stressors and other genes which act as potentiators, the gene is expressed in adulthood as schizophrenia. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4 FIG U R E 1h — M O D E L II GENE -> NEURO. DEFECT— » SSPD-» SCHIZOPHRENIA / / / / I (OTHER | ENVIRON. ENVIRON. 1 GENES) j STRESSORS STRESSORS Figure lb. In Model II, SSPD does not flow directly from genetic effects alone. According to this model, one (or more) aberrant genes lead to a neurodevelopmental abnormality, which again may be reflected in lower IQ and which contributes to the risk for subsequent SSPD and schizophrenia. As compared to Model I, however, this model posits that SSPD is not solely related to genetic effects, and that environemental stressors will have an additive effect in contributing to the prediction of SSPD. Note also that multiple gene effects are permissible. This model is consistent with the polygenic (“multiple hit”) theory of schizophrenia as posited by Fowles (1992). FIG U R E !c - M O D EL III STRESS HIGH GENE -+ NEURO. DEFECT -» SSPD-> SCHIZOPHRENIA LOW GENE -» NEURO. DEFECT III SSPD~>SCHIZOPHRENIA Figure 1c. Model III posits a moderator effect of stress on the relationship between intellectual deficits and schizotypy, such that both high stress and low IQ are needed to produce SSPD and schizophrenia. By this model, the schizophrenogenic gene(s) will produce a neurointegrative defect regardless of environmental circumstance, but will lead to SSPD only in the presence of stress. Thus, IQ deficits would only predict SSPD in individuals who have encountered significant stressors, while a low stress environment serves as a protective factor against SSPD and schizophrenia. This model is analogous to the diathesis-stress model of schizophrenia, applied at the level of SSPD. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Model IV (see Figure Id) also predicts an interactive effect between psychosocial stress and IQ, but, based on findings by Raine (1987), proposes that the association between cognitive deficits and higher rates of SSPD is stronger in lower psychosocial stress conditions. This model asserts that the etiology of SSPD is multidetermined by environmental and genetic factors, and that these influences can combine at different levels to create SSPD. Thus, a child who develops SSPD in a lower stress environment must have a stronger genetic vulnerability, and hence lower IQ scores, than children who become schizotypals in a high stress environment. FIG U R E Id - M O D E !, IV STRESS HIGH GENE — » NEURO. DEFECT III SSPD~» SCHIZOPHRENIA ENVIRON. STRESSORS LOW GENE — » NEURO. DEFECT-»SSPD-»SCHIZOPHRENIA Figure Id. According to the model proposed by Raine (1987), an interactive effect is posited such that low IQ (reflective of genetic risk) is predictive of SSPD only in individuals with low environmental stress. This model suggests that the high stress environment produces too many other sources of psychopathology, which might obscure the relationship between genetic risk, neurocognitive abnormalities, and psychopathology. The fifth model (Model V; Figure le) asserts that intellectual deficits are not an indicator of later SSPD, but are rather a mediator of the relationship between psychosocial Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 6 stress and personality functioning. This model is based on findings in the literature that children from homes of lower socioeconomic status (SES) and higher family stress often show lower IQS (e.g., Hollingshead and Redlich, 1958). As such, this model proposes that high levels of stressors lead to cognitive deficits, which, in turn, lead to the development of SSPD. In other words, low IQ is a predictor of subsequent SSPD only insofar as it is a reflection of stressors, such as low SES. The mediator model thus tests the possibility that any relationship between intellectual deficits and personality may be spurious, and accounted for by a third variable. F IG U R E 1e - M O D EL V LOW SES / HIGH STRESS -> LOW IO~* SSPD | GENE j Figure le. Model V depicts a mediator effect for stress. According to this model, intellectual deficits result from a high stress or low SES environmental background, rather than being a reflection of genetic abnormalities. (Genetics are depicted as being uninvolved in the prediction of SSPD). If this model is correct, the prediction of SSPD by IQ will be accounted for by stress and/or SES. The hypotheses derived from the two main aims are outlined below in more detail: Hypothesis la: High rates of schizotypal symptomatology at age 16 or at age 20- 22 will relate to lower intellectual/cognitive test scores at ages 3 and 11. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Two secondary hypotheses are also related to research about lower IQ and schizophrenia. First, a meta-analysis of research on lower IQ in schizophrenics (Aylward et al., 1984) demonstrated that premorbid IQ deficits are more typical for male than for female schizophrenics. Second, some studies have shown that schizophrenics tend to show lower Performance than Verbal IQS (e.g. Goldberg et al., 1988; Parker & Davidson, 1963). These findings lead to the following predictions: Hypothesis lb: Males with higher scores on the SSPD scales will show greater deficits on tests of cognitive functioning at ages 3 and 11 than will females. Hypothesis lc: Subjects with SSPD will show more Performance than Verbal intellectual functioning deficits at age 11. The second main hypothesis (Models I-V; Figures la-le) tests the five competing models examining the role of environmental stress in the relationship between early cognitive and later personality functioning. Model I (Meehl, 1962, 1990) predicts that schizotypals share the vulnerability but not the stressors involved in schizophrenia, leading to the following hypothesis: Hypothesis 2a (Model D: A biosocial relationship is predicted in which the schizotypal subjects will show lower scores on intellectual tests but will not have significantly different ratings on psychosocial stress compared to matched controls. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 8 Alternately, based on Fowles (1992) additive diathesis-stress model of schizophrenia, Model II predicts that schizotypals would show signs of a genetic vulnerability and would need to experience certain negative psychosocial stressors in order to develop SSPD, leading to the following hypothesis: Hypothesis 2b (Model II): An additive relationship is predicted in which both lower scores on intellectual tests and higher ratings on psychosocial stress would relate to higher levels of SSPD. Model IE, analogous to (but at odds with) the diathesis-stress model of schizophrenia (Meehl, 1962) suggests that the development of SSPD itself is predicated on an interaction between the genetic vulnerability and environmental stress, so that psychosocial stress moderates the relationship between cognitive performance and SSPD. Hypothesis 2c (Model IIP: A biosocial interaction is predicted such that lower intelligence test scores interact with higher environmental stress index scores to predict higher rates of SSPD. Model IV is based on findings by Raine’s (1987) study which showed that schizotypals from intact homes had significantly poorer Verbal IQ scores than those from broken homes. This led to the following hypothesis: Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 9 Hypothesis 2d ("Model IV): A biosocial interaction is predicted in which SSPD will be related to lower scores on intellectual tests and lower ratings on environmental stress. Finally, lower IQ has been proposed as a mediator of psychosocial stress and higher rates of psychopathology (e.g. Hollingshead and Redlich, 1958). This leads to Model V: Hypothesis 2e fModel V O : A biosocial model is predicted in which intellectual deficits act as a single, dominant mediator of the relationship between psychosocial stress and SSPD, such that the predictive value of IQ for SSPD will be zero when environmental stress is accounted for. These eight hypotheses were tested using longitudinal data collected in a birth cohort of 1795 subjects from the island of Mauritius. Beginning in 1972, subjects were tested with cognitive functioning measures at age 3 (based on an early version of the Boehm Test of Basic Concepts-Preschool Version; Boehm, 1989), as well as Verbal and Performance IQ measures (WISC-R, Wechsler, 1974) at age 11. The subjects were also assessed for parental socioeconomic status (Hollingshead and Redlich, 1958) at age 3 and family stress (e.g., divorce, parental death) at ages 3 and 11. SSPD was measured by self- report questionnaire (Venables et al., 1990) at age 16 (n=760), and DSM-III-R (American Psychiatric Association, 1987) defined SSPD was diagnosed using the Personality Disorders Examination (PDE, Loranger, 1988) at age 20-22 (n=541). Analyses of Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 10 variance (ANOVAs) and multiple regression analyses were used to examine antecedents of SSPD. To the author's knowledge, the study is the first of its kind to explore both the longitudinal relationship between intellectual functioning and SSPD, as well as the biosocial interaction between intelligence, environmental stress (family stress and SES), and SSPD. It is felt that the longitudinal design and large sample size afford a unique opportunity to test these biosocial effects, and can further clarify and explain the etiology of schizophrenia-spectrum disorders. Background and Significance This section will provide (1) a review of the literature on the relationship between intellectual functioning and schizophrenia, (2) a review of related studies on schizotypy and schizophrenia-spectrum personality disorders (SSPDs), (3) an outline of the five competing models on the development of SSPD in relation to schizophrenia, (4) a review of the literature on social/environmental stressors, with a focus on schizophrenia and schizotypy, and, (5) a synthesis of the aforementioned literature in relation to the main aims of this study. Finally, the significance of the study will be discussed. Intellectual functioning and schizophrenia Meehl (1990) offers an analogy for central nervous system (CNS) functioning of the schizophrenic individual; he compares it to a complex machine fulfilling many functions, one that operates on many pulleys and gears. However, because of a small defect in the teeth of all the gears (e.g. they are a Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 11 millimeter too short), the machine frequently malfunctions in what seems to be an unpredictable manner. Similarly, according to Meehl, the schizophrenic's CNS has a "gears" problem, one that manifests either neuroanatomically or neurochemically at the synaptic level. Meehl (1962, 1990) proposed that this underlying "integrative neural defect" could reveal itself in a number of different ways. It has been suggested that one such manifestation is decreased cognitive and intellectual functioning (Comblatt & Erlenmeyer-Kimling, 1984; Zubin & Steinhauer, 1981; Nuechterlein & Dawson, 1984) in the form of attentionai deficits (Comblatt et al., 1992; Nuechterlein & Dawson, 1984), as well as in lower IQ scores (Aylward et al., 1984). Whereas a number of studies have explored attentionai functioning in schizophrenic and schizotypal subjects, as well as intellectual functioning in schizophrenics (e.g., Crawford et al., 1992, Worland and Hesselbrock, 1980), only eight studies (Asamow & Ben-Meir, 1988; Fenton & McGlashan, 1989; Poreh et al., 1995; Lencz et al., 1993; Trestman, et al., 1995; Kelley & Coursey, 1992; Raine, 1987; Bergman & Walker, 1995) have utilized SSPD samples to examine the status of IQ deficits, per se, as a vulnerability marker. These will be reviewed in a subsequent section of this introduction. Studies have shown an association between schizophrenia and decreased IQ levels in chronic patients as compared to first-time hospital admitted schizophrenics, suggesting a neurodegenerative process (e.g., Bilder et al., 1992; Goldberg et al., 1988). It is now widely accepted that cognitive functioning often diminishes after the first psychotic break, and, in chronic cases, this neurodegeneration continues (Bilder et al., 1992). However, it remains unclear whether a neurodevelopmental process is also at play, whether IQ deficits Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 12 are evident before the onset of psychosis, even in childhood. If IQ deficits are present premorbidly in schizophrenics, they may serve as an indicator of the integrative neural defect. Findings on premorbid schizophrenia, and on relatives of schizophrenics, while distinct from studies directly researching SSPD, can shed light on whether these IQ deficits are reflective of a genetic vulnerability marker. Several retrospective studies of premorbid functioning in schizophrenics have shown that pre-schizophrenic children have lower IQS as compared to controls (Crawford et al., 1992; Watt & Lubensky, 1976; Offord, 1974; Offord & Cross, 1971; Roff et al., 1976). These studies are complemented by a number of high-risk and longitudinal high-risk studies (e.g., Schreiber et al., 1992; Watt et al., 1982; Griffith et al., 1980; Mednick & Schulsinger, 1968; Worland et al., 1982; Bergman & Walker, 1995, Walker et al., 1981), in which children of schizophrenic parents were tested on IQ measures and compared to matched controls. These studies' findings have consistently demonstrated that lower IQ is characteristic of the biological offspring of schizophrenics, and therefore may be a manifestation of a vulnerability to schizophrenia. Additionally, while findings are mixed, fairly consistent evidence points to differences in types of IQ; schizophrenics show more deficits in Performance IQ in comparison to Verbal IQ (e.g., Goldberg et al., 1988; Parker & Davidson, 1963). Sex differences have also been noted; male schizophrenics show greater IQ deficits than females (Aylward et al., 1984), mirroring other findings of earlier age of onset (e.g., Lewine, 1980) and more severe symptomatology (Goldstein et al., 1989) for males. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 13 The aforementioned high-risk studies provide some support for the hypothesis that decreased intellectual functioning is neurodevelopmental in origin, and may be the manifestation of a diathesis for schizophrenia. However, the generalizability of high-risk studies is limited by the fact that only approximately 10% of schizophrenics have a schizophrenic parent. Because of this limitation, researchers (e.g., Erlenmeyer-Kimling & Comblatt, 1984), as well as panels advising on an NIMH plan for schizophrenia have recommended broadening the scope of longitudinal research by studying general- population children (see Significance section below). For example, Holzman et al. (1988) suggested that research on schizotypal personality may elucidate genetic factors in vulnerability to schizophrenia. The study builds upon previous research by (1) examining the relationship between intellectual functioning and schizotypy in the general population, thereby avoiding confounding factors such as family risk, acute psychotic symptomatology, and long-term neuroleptic and institutional treatment, and offers (2) a longitudinal assessment of the development of SSPD, and (3) an examination of biosocial interactions between cognitive functioning and proposed environmental stressors, specifically, family stress and socioeconomic status. Intellectual functioning and schizotypy Whereas findings clearly show a relationship between lower IQ scores and schizophrenia, the relationship between intelligence test performance and schizotypy is more elusive; to date, only eight studies have examined this relationship, with mixed results (Asamow & Ben-Meir, 1988; Fenton Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 14 & McGlashan, 1989; Poreh et al., 1995; Lencz et al., 1993; Trestman, et al., 1995; Kelley & Coursey, 1992; Raine, 1987; Bergman & Walker, 1995). In a study of children with SSPDs, Asamow and Ben-Meir (1988) showed that schizotypal and schizophrenic children had significantly lower IQ scores as compared to children with depressive symptomatology. The fact that this study examined children is notable in light of the present study’s focus on childhood predictors of late adolescent SSPD. Fenton & McGlashan (1989) showed similar findings in a 15-year longitudinal study of adults at a psychiatric hospital (Chestnut Lodge). The investigators examined predictive variables of later schizophrenic psychosis on a group that had previously been hospitalized for a diagnosis of SPD. IQ was a significant predictive variable; schizotypals who developed psychosis had lower premorbid IQ scores than patients who did not become schizophrenic. The findings demonstrating a clear relationship between lower IQ and SSPD were not supported by several other studies. Lencz et al. (1993) found no significant difference in IQ sub-scales between an undergraduate schizotypal group and controls. Poreh et al. (1995) also found that psychosis-prone undergraduate subjects showed no significant difference compared to matched controls on the Vocabulary and Block Design subtests of the WAIS-R. Similarly, Kelley and Coursey (1992) did not find differences in IQ sub-test levels between undergraduates with schizotypal symptoms and controls. Trestman et al. (1995) also found no significant group differences when comparing schizotypal subjects, subjects with other personality disorders, and normal volunteers on the WAIS-R block Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 15 design and vocabulary sub-scales. In fact, WAIS-R vocabulary test performance showed a significant negative correlation with schizotypal symptoms. In light of these studies' disparate findings, it is important to note that several of these studies (Lencz, et al., 1993; Kelley & Coursey, 1992; Poreh et al., 1995) utilized undergraduate samples. Schizotypals who are attending a university may score higher on IQ tests as a reflection of their higher cognitive functioning, thus restricting the range of scores. In order to overcome this possible restriction of range, the present study examines an unselected sample. To the author’s knowledge, only two studies have examined the biosocial interaction between negative home environment, IQ, and SSPD. First, Bergman & Walker (1995) studied children from a high-risk longitudinal sample over a I year period. They explored the relationship between environment (maltreating parents) and child behavior by testing whether cognitive ability was mediating or moderating that relationship. They concluded that cognitive ability acted not as a mediator, but rather as a moderator, such that lower IQ scores on the WISC-R interacted with higher levels of parental maltreatment to predict schizoid behavior in the children. These results are consistent with Model III proposed in the present study; lower IQ and higher psychosocial stress interact to predict SSPD symptoms. It is important to note, however, that this study did not use a SSPD sample, but rather, children of schizophrenics, none of whom were diagnosed with SSPD. Additionally, cognitive ability was conceptualized as the moderator/mediator of the relationship between family stress and schizoid behavior; in the present study, intellectual deficit is viewed as the putative indicator of later SSPD, whereas environmental stress is Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 16 tested as a moderator/mediator. Even so, the study by Bergman & Walker (1995) provides more evidence that cognitive ability and psychosocial stress interact to predict SSPD traits. Second, Raine (1987) examined a measure of intact/broken homes as an environmental stressor variable, and found that schizotypals from intact homes had significantly poorer Verbal IQ scores than those from broken homes. This study showed that schizotypals are marked by a combination of low IQ and the lack of family distress, a result that supports Model IV. However, Raine (1987) utilized subjects from a criminal population as part of a larger study on psychopathy. The study seeks to increase generalizability of findings by testing for similar interactions in the general population. Models of SSPD development In addition to exploring the relationship between intellectual functioning and SSPD, the study examines five models of the development of schizotypal personality. Models are based on either Meehl's (1962, 1990) theory of the genetic etiology of schizophrenia, or on other theories, contradictory to Meehl’s, of the development of schizophrenia. Meehl’s diathesis-stress model of schizophrenia proposes that SPD is a psycho-behavioral expression of the same genetic vulnerability that underlies schizophrenia. Only when the schizotypal individual is impacted by an environmental stressor does schizophrenia develop; when the individual with SSPD is not affected by a significant stressor, symptoms do not develop into florid psychosis. Thus, schizotypal individuals are thought to manifest the same "integrative neural defect" that characterizes schizophrenia, without sharing significant environmental stressors. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 17 Alternately, Model II hypothesizes that SSPD is a disorder with its own vulnerability and stress factors which, in an additive fashion, can predict SSPD. This model is based on Fowles’ (1992) diathesis-stress model of schizophrenia, which asserts that there is a multifactorial, polygenic etiology of schizophrenia (Gottesman & Shields, 1982); a number of genes and environmental stressors (e.g. family, psychosocial, biological) contribute in an additive fashion to the overall liability to schizophrenia, and phenotypic schizophrenia arises when a certain threshold is reached. Thus, the etiological pathway to schizophrenia is continuous, and individuals with SSPD would show signs of the genetic vulnerability (e.g. cognitive deficits), and would experience environmental stressors, before reaching the threshold for schizophrenia. As such, Model II predicts an additive effect of lower IQ and higher psychosocial stress (lower SES, higher family stress) in predicting later SSPD. Models in and IV are also based on Fowles’ (1992) diathesis-stress model of schizophrenia, albeit with one important difference; the genetic vulnerability to schizophrenia and environmental stress do not only have an additive effect, instead, they interact to predict later schizophrenia. In other words, environmental stress moderates the relationship between the genetic vulnerability and later SSPD. However, Model III and Model IV have different predictions about this interaction. Model III hypothesizes an etiology of SSPD patterned like Meehl’s etiological model of schizophrenia: higher levels of environmental stress interact with the genetic vulnerability to create SSPD. Thus, in the present study, Model m predicts an interaction between higher psychosocial stress (lower SES, higher family stress) and lower IQ scores in predicting later SSPD. In contrast to R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 18 Model H’s additive hypothesis, therefore, the prediction of an interaction between environment and gene implies a higher threshold for the development of SSPD. Model IV offers an alternate biosocial approach which predicts that environmental stress interacts with the genetic diathesis to predict later SSPD only at low levels of stress. In the present study, this model hypothesizes that lower psychosocial stress (higher SES, lower FSI) will interact with intellectual deficits to predict SSPD. This model is based on the findings described above (Raine, 1987), which indicated that schizotypals from “intact homes” showed lower Verbal IQS than did schizotypals from “broken homes”. Other studies have also reflected a similar pattern in which better environment interacting with a genetic vulnerability factor could predict later psychopathology (Dohrenwend et al., 1992). These findings may be explained if the etiology of SSPD is multifactorial, similar to the etiology described in the multifactorial, polygenic diathesis-stress model of schizophrenia (Fowles, 1992; Gottesman and Shields, 1982). In such a model, an individual would need to experience a certain amount of stress (such as genetic vulnerability, psychosocial stressors, prenatal insult, etc.) to develop SSPD. Thus, children who are raised in an environment with high levels of psychosocial stressors may not need as strong of a genetic loading to develop SSPD. Similarly, to develop SSPD in a low- stress environment, the child might need a stronger genetic “push,” and thus would manifest more signs of the integrative neural defect, such as cognitive deficits. The fifth model assumes a relationship between psychosocial stress and later psychopathology in the form of SSPD, and is based on many studies (e.g. Hollingshead and Redlich, 1958; Myers and Roberts, 1959) showing higher family stress and lower SES R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 19 predicting schizophrenia and other psychopathology. This model predicts that intellectual deficits will act as a single, dominant mediator in the relationship between psychosocial stress and SSPD. In other words, the relationship between IQ and schizotypy can be accounted for by their shared relationship to psychosocial stress. Psychosocial stressors The effects of environmental stressors on the development of schizophrenia have been studied in various ways, with focus on the biological (e.g., perinatal complications; Schulsinger et al., 1987), and the social (e.g. Green, 1990). In the present study, two measures of “psychosocial” stress have been collected; specifically, a Family Stress Index that quantifies stressors in the subject’s life at ages 3 and 11, and a Socioeconomic Status Scale (SES) that quantifies parental SES at subject age 3, according to the Hollingshead Rating Scale (Hollingshead and Redlich, 1958). A number of studies have examined the effect of social and family stressors on schizophrenia such as institutionalization of the child (Mednick et al., 1987; Walker et al., 1981), quality of home and neighborhood (Mednick et al., 1987; Walker et al., 1981), paternal absence (Mednick et al., 1987; Mednick & Schulsinger, 1968; Walker et al., 1981), and parental maltreatment (Bergman & Walker, 1995). Mednick et al. (1987), using the Copenhagen high-risk sample, showed that the mothers of children who developed SPD were significantly more stable (as determined by a social stability scale) than the mothers whose children became schizophrenic. Additionally, the children who had SPD but did not develop schizophrenia had more contact with their fathers than did children who became schizophrenic. Walker et al. (1981) used the same sample and also R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 20 found that the schizophrenic group of children had experienced significantly more paternal absence during the second year of life, as well as childhood institutionalization. However, the quality of home or neighborhood did not differ for the two groups. With the exception of these high-risk studies and the studies described previously (Bergman & Walker, 1995; Raine, 1987), researchers have not fully explored the relationship of family/social/environmental variables to intelligence and SSPD. However, studies on more specific signs of family instability, such as expressed emotion or communication deviance, have demonstrated significant interaction effects in the etiology of schizophrenia (e.g., Green, 1990; Nuechterlein & Dawson, 1984). Additionally, family instability has been researched in relation to other psychopathologies and personality disorders. For instance, the loss of a parent before the age of 11 when combined with a threatening life event or long term difficulty will make a child three times more likely to manifest adult depression (Rutter, 1990). Furthermore, in a study examining risk factors for criminality, Raine et al. (1996) performed a cluster analysis of biological and psychosocial risk factors, and revealed that when early neuromotor deficits (signs of the diathesis) and negative family factors such as early maternal rejection, family conflict and family instability clustered together, subjects were more likely to be criminal and violent as compared to those defined by only biological factors or only psychosocial stress factors. These findings demonstrate that family stress may significantly interact with biological factors in the development of psychopathology, and support a model analogous to Model III in the present study. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 21 Research has also shown a relationship between psychopathology and SES (Myers & Roberts, 1959; Myers & Bean, 1968; Hollingshead and Redlich, 1958; Dohrenwend, et al., 1992). Lower social class has been shown to relate to schizophrenia (Dohrenwend & Dohrenwend, 1969; Gift et al., 1988; Holzer et al., 1986), leading to the theory of “social causation,” (Hollingshead & Redlich, 1958) which relates lower social status to higher stress and greater adversity, in turn, leading to psychopathology. A contrasting theory is one of “social selection,” or the tendency of individuals with the diathesis to drift downward into lower social classes, or fail to move upward (Wender et al., 1973; Dohrenwend et al., 1992). Longitudinal studies of children who suffered birth complications or early developmental delays compared those from high SES and low SES homes (Drillien, 1964; Davie et al., 1972; Rubin & Balow, 1979). Findings indicated that children from lower SES homes tended to manifest signs of cognitive and neuromotor deficits throughout childhood, whereas children from higher SES homes were more likely to improve, and, by middle childhood, were not distinguishable from a normal sample. It is possible that, in these cases, SES moderated the relationship between genetic/biological vulnerabilities and later neurocognitive functioning such that lower SES interacted with vulnerability to predict poor functioning. Similarly, as in Model E Q proposed in the present study, SES may moderate the relationship between the vulnerability factor (cognitive deficits) and later SSPD, such that lower SES interacts with lower IQ to predict schizotypy. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. Taxon vs. Dimension A discussion of the categorical versus the dimensional nature of schizotypy is important at this juncture, as findings from the two main hypotheses may shed light on this issue. Another important assumption in Meehl’s model of schizophrenia posits that schizotypy is a discrete category of illness, a taxon, and that schizotypy does not lie along a continuum with normal personality. Examinations of this hypothesis have attempted to determine the latent liability distribution (e.g. dichotomous vs. continuous) of sets of putative indicator variables for schizotypy by running taxometric analyses on a variety of samples. A number of these studies (e.g. Lenzenweger & Korfine, 1992; Golden & Meehl, 1979; Tyrka et al., 1995), some using the MAX-COV-HIT-MAX method, have revealed a dichotomous distribution with a low base rate (5-10%), lending support to Meehl’s notion of schizotypy as a taxon. This element of Meehl’s theory would predict that group differences as measured by ANOVAs would be more likely to elucidate the relationship between the latent liability (cognitive deficits) and SSPD, as compared to correlation or regression analysis. This is because the dimensionalized ratings used in the regression analysis would represent “noise” or error variance, obscuring the true taxon (Gangestad & Snyder, 1985). It is important to note, however, that if schizotypy does represent a taxon, it is still difficult to determine whether one “schizogene” is responsible; a polygenic model with a threshold effect or a genetic heterogeneity model with independent genes could also account for these findings (Tyrka, et al., 1995). Claridge & Beech (1995), proponents of a dimensional model, stress that dimensionality is not necessarily dependent on either a single gene or polygenic hypothesis. Further, they expand the range of the concept by contrasting two different R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. views of the dimensionality of schizophrenia. First, a “quasi-dimensional” theory is one supported by the psychiatric framework which focuses on variations of traits within the illness, but does not include personality traits in the healthy domain. Thus, clinically diagnosable symptoms define the starting point of this continuum severity within the taxon. On the other hand, a “fully dimensional” model includes in its scope “healthy” and “normal” behavior not usually rated symptomatic by clinicians, but nonetheless considered part of the continuum (Claridge, 1987). A dimensional view of schizotypy would promote the use of correlational or multiple regression techniques so as to test the full effect of the relationship between the indicator (e.g. cognitive deficits) and SSPD (Gangestad & Snyder, 1985). Thus, this view would support the use of dimensional measures of SSPD (e.g. SAE; Venables et al., 1990), rather than a measure based on the fulfillment of a certain number of diagnostic criteria (e.g. PDE; Loranger, 1988). Also important when examining schizotypy along a continuum is understanding the multidimensional nature of SSPD and schizophrenia (Claridge & Beech, 1995). Two factor (e.g. Crow, 1980; 1985) and three factor models (Liddle, 1987; Arndt et al., 1991) of schizophrenia and schizotypy (e.g., Raine, et al., 1994; c.f. review by Venables, 1995) have been developed to account for heterogeneity of symptoms. Common to all of these models is a distinction between "positive" and "negative" symptoms; “positive” symptoms denote the presence of abnormal experiences such as delusions and hallucinations in the case of schizophrenia and perceptual aberrations and magical ideation for schizotypal personality. “Negative” implies an absence or deficiency in functioning, as manifested in symptoms such as withdrawn behavior, poverty of speech, interpersonal deficits, little R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 24 interest in people and activities (social anhedonia), little pleasure in life (physical anhedonia), and flat or blunted affect (Crow, 1980; Strauss et al., 1974). Researchers have differed over the fit of the two-factor model onto schizotypy. Based on family studies of the genetics of schizophrenia, Gunderson et al. (1983) proposed that negative schizotypy is the genetic core of schizophrenia, and that positive schizotypy is distinct at the genetic and pathophysiological level. In this model, positive symptoms are a reaction to negative schizotypy. Research by Siever et al. (1984, 1994) shows eye tracking dysfunction relating to negative and not positive symptomatology. However, Meehl (1975, 1989) holds that only more “positive” symptoms such as cognitive slippage are manifestations of the integrative neural defect; “negative” symptoms such as anhedonia are dimensional traits that serve as polygenic potentiators of the schizogene. Most studies of neurocognitive abnormalities in SPD use positive symptoms to measure schizotypal personality in their subjects (Lencz et al., 1995), so inferences about these abnormalities are somewhat limited. Several researchers have used the two-factor approach in conjunction with the polygenic model (Kendler et al., 1989; Siever, 1995). For example, Kendler et al. (1989) showed, in a study of monozygotic twins, positive and negative schizotypy were independently heritable, and have different patterns of neuropsychological and psychophysiological correlates. Similarly, Siever (1995) proposes a model in which two separate genes and pathophysiologies underlie positive and negative schizotypy. By this model, positive schizotypy is related to ventricular enlargement and dopaminergic over activity in the sub-cortical areas, whereas negative schizotypy is associated with atrophy R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 25 and hypo-dopaminergic activity in the cortical areas as well as functional abnormalities of the pre-ffontal cortex as evidenced by eye-tracking dysfunction. This model has been supported by many findings in which eye tracking dysfunction (Siever et al., 1984, 1994), and P3 attenuation relate to “negative” symptoms, whereas increased cerebrospinal homavanillic acid (HVA; Siever et al., 1991 ) was associated with “positive” symptoms. These findings indicate that "negative" symptoms are associated strongly with both structural and functional pre-ffontal deficits. It might be expected, then, that cognitive deficits might be expected to be stronger indicators of the genetic diathesis when predicting “negative” rather then positive symptoms. However, it is important to note that in these studies, negative symptoms refer to social isolation and constricted affect; the present study’s anhedonia scale does not measure constricted affect extensively, and also rates physical anhedonia. Also, the hypothesis of independent heritability is based, in sum, on a relatively small number of studies on biological correlates to schizotypy, and in each area, they tend to be done by the same laboratory (e.g. Siever et al., 1984, 1994 on eye tracking). The group comparison and multiple regression analyses in the present study will help to clarify some of the issues around whether schizotypy is best conceptualized as a taxon or a dimension. Synthesis of findings To date, a number of studies have shown the commonality between neuro-cognitive deficits in schizophrenia and in SSPDs (addressed e.g. in Lencz et al., 1995; Siever et al., 1993). These deficits are also found in relatives of R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 26 schizophrenics, and are presumed to have a genetic basis (Erlenmeyer-Kimling & Comblatt, 1987). One such presumed deficit, lower IQ, has been examined in schizophrenia and family high-risk studies (Aylward et al., 1984). Findings lend support to the hypothesis that low IQ is a genetic vulnerability marker for schizophrenia. However, research has not yet clarified the relationship between SSPD and intelligence, which could provide further support for the presumption that IQ is a vulnerability marker. The first main aim of the study is to test the association between IQ and SSPD (Hypothesis la). If schizotypals, like schizophrenics, are marked by low intellectual functioning, results will support the hypothesis that IQ is part of the shared "integrative neural defect." However, if IQ deficits are not found in schizotypals, important etiological information will still be obtained. For instance, other hypotheses may be implied; IQ may relate to environmental factors such as prenatal/perinatal insult (Aylward et al., 1984), or to state factors, like decreased motivation (Green et al., 1992). Subsidiary hypotheses to this first main prediction are also derived from the literature on schizophrenia and IQ (e.g. Aylward et al., 1984; Goldberg et al., 1988; Parker & Davidson, 1963), which suggests that IQ deficits were greater in schizotypal males than females (Hypothesis lb), and that schizotypals will show greater deficits in Performance as compared to Verbal IQ (Hypothesis Ic). The second main aim of the study is to further explore not only the role of IQ in the etiology of SSPD, but also to examine the additive, moderating, and mediating effects of environmental stress on SSPD development. The results obtained, regardless of the direction of the effect, will provide important information about the development of R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 27 schizotypy. A result that shows that negative home environment does not play a role in SSPD development will suggest that SSPD is only a manifestation of the "integrative neural defect" shared with schizophrenia (Model I). However, it is important to note that family stress and SES, while important variables in psychopathology research, are not the only possible environmental factors affecting SSPD development, and so support for Model I must be interpreted with caution. Results that reflect an additive relationship between psychosocial stress and SSPD development will support Model II. Alternately, an interaction effect between higher stress and lower IQ would support Model IQ. If an interaction effect between psychosocial stress and intellectual performance reveals a relationship between lower stress and lower IQ, results would support Model IV. These last three models imply that schizotypy is not just the expression of the genetic vulnerability to schizophrenia; instead, SSPD may be the outcome of many interacting factors, genetic and environmental. If any of the preceding four models are partially supported, the possibility still remains that the relationship between cognitive deficits and SSPD is spurious, and that the deficits act as a single, dominant mediator in the relationship between psychosocial stress and SSPD. As such, the fifth model ensures that any significant relationship emerging between IQ and SSPD is not spurious. Several conditions must hold to determine possible mediation (Baron & Kenny, 1986): (1) The psychosocial stressor (family stress, SES) and the cognitive measures must be significantly related to each other; (2) IQ must significantly predict SSPD, and (3) when psychosocial stress is partialled out, intellectual functioning does not predict later SSPD. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 28 Summary The present study tests hypotheses related to the development of schizotypal personality. First, the study explores the relationship between intellectual deficits and schizotypy. Then, it compares five competing biosocial models of SSPD development by examining the additive, moderating and mediating effects of family stress on the relationship between IQ and schizotypy. Predictions were tested using analyses of variance and multiple regression analyses. The longitudinal nature of the data base, its ethnic diversity, and its size make it possible to more fully investigate subjects from a general population on the basis of SSPD traits, and to study possible childhood correlates to later schizotypy. Significance of the study The significance of the proposed research can be set in a number of contexts: 1. A study of schizotypal personality addresses recommendations which have been made by researchers (e.g. Erlenmeyer-Kimling & Comblatt, 1984) and several advisory groups on a NIMH national research plan for schizophrenia (Andreasen et al., 1988; Eaves et al., 1988; Holzman et al., 1988). These researchers have argued for expanding the range of longitudinal populations studied to include general populations, and for examining whether certain bio-behavioral variables affected by schizophrenia are associated with schizophrenia-spectrum disorders. The study here will in part meet these recommendations. 2. To the author’s knowledge, this study is the first to look at intellectual functioning and SSPD in an unselected sample drawn from a general population. Results R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 29 that reveal differences in IQ levels between schizotypal individuals and controls would suggest IQ as part of the "integrative neural defect," thus increasing understanding of IQ's role in schizophrenia. On the other hand, findings showing no differences may imply that IQ is not a manifestation of the genetic predisposition, but instead may reflect the results of environmental stress or secondary vulnerability. If this is the case, high IQ may even represent a possible protective factor against schizophrenia; i.e. an individual with the genetic liability (schizotypy) who has a high IQ would be less likely to develop schizophrenia. Such information can ultimately prove useful in prevention of schizophrenia. 3. The study's use of a biosocial model of SSPD, across developmental phases, provides a multi-modal and integrative view of this spectrum of personality disorders. This allows for a teasing-out of the specific variables involved in the etiology of SSPD. By testing five competing models about the impact of psychosocial stress on SSPD development, the study will reveal significant etiological information regardless of which model is supported. 4. To the author’s knowledge, no published studies have examined DSM-III-R (American Psychiatric Association, 1987) diagnosed SSPD longitudinally in the general (non-high-risk) population. The study is a prospective one, and the longitudinal design not only provides important etiological data, but also can suggest cause-and-effect relationships, although, admittedly, causality cannot be established by this design alone. 5. The roughly equal mix of males and females in the sample allows for a comprehensive analysis of sex differences in the etiology of SSPD. Accordingly, the study R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 30 will test a hypothesis, derived from the schizophrenia literature (Aylward et al., 1984), about stronger effects for males in the relationship between lower IQ and schizotypy. Experimental Design and Methods Subjects A birth cohort composed of 1795 three-year-old children bom in 1969 and assessed initially between September 1972 and August 1973 were examined in this study. The subjects in the sample were assessed at ages 3, 11, 16 and 22-23 on a variety of measures. As mentioned above, cognitive measures at ages 3 and 11, personality measures at ages 16 and 22-23, socioeconomic status at age 3 and a family stressor index at ages 3 and 11 were used to explore biosocial interactions and their prediction of later SSPD. The female to male ratio of this sample is 48.1 : 51.9 (females=856).The sample forms almost the entire population of that age in two towns (Quatre Bomes and Vocoas) on the island of Mauritius; the towns were chosen because they contained a racial mix very similar to the racial distribution of the island as a whole. In this sample, 60.4% of the children were of Indian origin (Moslems, Hindus, and Tamils), 22.6% were from the "General Population" (the official term for a Creole group with mainly African and some Caucasian origins), and a smaller proportion (17%) were of Chinese, English, or French descent. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. Cognitive measures Cognitive testing at age 3. Subjects were assessed with a test designed by Brian Sutton-Smith based on Piagetian principles and the Boehm Test of Basic Concepts- Preschool Version (Boehm, 1989). It utilized Mauritian items as materials to test the child's understanding of 26 basic relational concepts such as size, direction, position in space, quantity, and time. The score on this scale serves as a global intelligence measure for age 3. Scores on these tasks were available on 1381 of the subjects, and ranged from 0-81 (mean=36.73; sd=T2.43). For the purposes of group comparisons, this measure was z-transformed, and high and low groups were formed using subjects who scored 1.5 standard deviations above (n=62) and below (n=97) the mean. Subjects who were tested on this measure were compared to those who were not, and did not significantly differ on sex (X^.003, p>.95), but did on ethnicity (X2 =17.87, p< 001). Frequency distributions of ethnicity on those who received the measure were: Indians, 64.6%; General Population, 21.1%; Other, 14.2%. Frequencies for those who did not receive the test were: Indians, 57.3%; General Population, 31.3%; Other, 11.4%. Consequently, ethnicity was added as a covariate in all analyses of variance involving intellectual testing at age 3. A list of the scale’s items are included in Appendix I. Cognitive testing at age II. The following sub-tests of the Wechsler Intelligence Scale for Children-Revised (WISC-R; Wechsler, 1974) were administered to 941 subjects,: Similarities, Digit Span, Picture Completion, Block Design, Object Assembly, Coding, and Mazes. Comprehensive reliability and validity data are available on this test R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. (Sattler, 1982), and raw scores were scaled according to age and norms listed in the WISC-R manual (Wechsler). Standardized scores were averaged into three indices for purposes of data analysis: (I) a Verbal IQ score consisting of the Similarities and Digit Span sub-tests of the WISC-R (range 1-12.5, mean=5.73, SD=2.10), (2) a Performance IQ measure consisting of the Picture Completion, Block Design, Object Assembly, Coding and Mazes sub-tests of the WISC-R (range 1-12.4, mean=5.55, SD=2.02), and (3) an estimated full-scale IQ score for age 11 (range 1-12.3, mean=5.60, SD=1.86). Subjects who were tested on this measure did not significantly differ on sex Q 0=. 148, p>.70) from those who were not tested. However, there was a significant difference on ethnicity (X*=l 1.06, p<005); frequency distributions for those who were administered the measure were: Indians, 65.7%; General Population, 20.2%; Other, 14.1%. Distributions for those who did not take the test were: Indians, 59.9%, General Population, 27.1%, Other, 13.0%. Consequently, ethnicity was added as a covariate in all analyses of variance involving IQ at age 11. Social Variables Family stress index (FSI). This measure (FSI) is an index comprised of the following variables collected on 1241 subjects at ages 3 and 11: father’s employment status (full-time, part-time or unemployed); and family conditions (orphaned, adopted into two-parent family, adopted into a one-parent family, death of mother, death of father, father or mother absent from home, and parental divorce). These variables were chosen based on the literature (e.g. Mednick et al., 1987; Walker et al., 1981; Samerofif et al.. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 1987) regarding predictors of schizophrenia in high-risk sample; parental absence from the home and institutionalization were found to be significant predictors of later schizophrenia. Father’s employment status was included based on findings regarding the relationship of lower SES to schizophrenia (e.g. Hollingshead & Redlich, 1958), and on studies that found differences in SES between their high-risk and control groups (e.g. Rutschmann, et al., 1986; Worland et al., 1982). The variables were weighted by two experts in the field of social factors in schizophrenia: Drs. John Brekke and Joseph Ventura. They weighted each variable on a scale of 0-10, with the score o f“0" reflecting no social/familial stress, and a score of “ 10" reflecting a highest degree of stress. The weightings of the two experts were averaged and an index was formed with each variable reflecting the weight assigned. Items on this scale are included in Appendix 2. Those subjects administered these items were compared to those who were not and did not differ on sex (X^.025, p>.87), but did significantly differ on ethnicity (X2 =49.41, p<00l); frequency distributions for those who measured on this index were: Indians, 61.6%; General Population, 21.1%; Other, 17.3%, whereas frequency distributions for those who were not were: Indians, 65.8%; General Population, 29.0%; Other, 5.2%. Consequently, ethnicity was entered as a covariate in all subsequent analyses involving this variable. Socioeconomic Status (SES). This measure was collected on 1772 subjects at age 3. It was formed based on the Hollingshead two-factor scale (Hollingshead, 1965) which R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 34 combines parents’ education and employment status in an index of social position. The scale ranks professions and businesses into different categories based upon value and size. Specifically, the seven employment positions are: (1) executives and proprietors of large concerns (50 or more employees) and major professionals (including academics); (2) managers and proprietors of medium concerns (20-49 employees) and minor professionals, including high-level clerical, nurse, midwife; (3) administrative personnel of large concerns, owners of small independent business (less than 20 employees), and semiprofessionals; (4) owners of little businesses, clerical and sales workers, and technicians; (5) skilled workers such as factory worker; (6) semiskilled workers such as outdoor laborers; (7) unskilled workers, part-time workers and unemployed. This employment scale is weighted with a factor weight of 7. The educational scale has a factor weight of 4 and is divided into seven positions: (1) graduate professional training; (2) standard college or university education; (3) partial college training; (4) high school graduation; (5) partial high school; (6) junior high school; (7) less than seven years of school (Meyers and Bean, 1968). After the educational and employment ratings were weighted and added together, a continuum of scores ranging from 15 to 77 (mean=62.96, SD=9.32) was created. Lower scores on the continuum translate into a higher SES status. The decision to use this scale as a separate construct from the FSI was made on a conceptual and statistical basis. First, the Hollingshead scale is a well-known, well- validated measure, and findings can be more clearly understood in the context of the literature. Additionally, whereas SES has been hypothesized to be a causative variable in R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 35 the development of schizophrenia, it has also been theorized to be caused by the psychopathology. Second, while the FSI and the SES measures had a significant correlation (p< 001), the Pearson correlation coefficient was only = . 187, and thus does not speak strongly to collinearity. Subjects who were rated on this scale were compared to those who weren’t on sex and ethnicity, and no significant differences were found. Personality assessment Measurement of SSPD at age 16. A self-report questionnaire designed to identify traits related to SSPD was administered to 760 members of the cohort at age 16. The scale, called the “Survey of Attitudes and Experiences,” (SAE, Venables, et al., 1990), was specifically designed for the Mauritius study; it is as short as is possible while retaining reliability, and innocuous as possible as it was to be given to a general population for survey purposes. The scale contains two dimensionalized sub-scales and measures "schizophrenism," analogous to the cognitive-perceptual aberrations of SPD, and "anhedonia", which is related to the social-interpersonal deficits in SPD. The scale, its use, and its validation are described in detail in Venables et al. (1990); the average internal reliabilities (Cronbach’s alpha) of the subscales in several test samples were 0.73 for schizophrenism and 0.67 for anhedonia. The scale’s factor structure has been replicated on Westerners as well as Mauritians (Venables and Bailes, 1994). A list of the items on this scale is included in Appendix 3. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 36 An index and a control group for each scale (anhedonia, schizophrenism) were created on the basis of standardized (z-transformed) scores; index groups were comprised of subjects scoring 1.5 SDS above the mean, and control groups of subjects 1.5 SDS below the mean. The sample sizes of the index and control anhedonia groups are 54 and 29, respectively; the schizophrenism groups are 51 and 43, respectively. The anhedonic index group had two subjects in common with the schizophrenism index group; the comparison groups had one subject in common. Those who were administered this measure (n=760) did not significantly differ from those who were not (n=1035) on sex (X2 =.362, p<55); however, there was a significant difference in ethnicity (X2 =271.7, p<001). Frequency distributions for ethnicity for those who were administered the SAE were: Indians, 55.2%; General Population, 15.5%; Other, 29.3%; distributions for those who were not administered the measure were: Indians, 68.5%; General Population, 29.4%; Other, 2.1%. Consequently, ethnicity was included as a covariate in all analyses of variance including this variable. Diagnosis of SSPD at age 20-22. Subjects were interviewed by a clinical psychologist (Kavi Beepat, Ph.D) using the Personality Disorders Examination (PDE; Loranger, 1988). Each pathological symptom is rated on a 3-point scale: not present (0), subthreshold (1), and present (2). Presence of schizophrenia-spectrum personality disorders (SSPDs) was evaluated by scores of “2" on DSM-III-R (American Psychiatric Association, 1987) criteria in three disorders: (1) schizotypal personality disorder (SPD), (2) paranoid personality disorder (PPD) and (3) schizoid personality disorder (SZPD). R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 37 Ratings on this examination were also transformed into dimensional scores by summing the scores for all SSPD items. A total of 541 subjects were administered the PDE and differed significantly from those who weren’t (n=1254) on sex (X2 (1)=16.92, p< 001); frequency distributions for those who were administered the PDE were: males, 59.4%; females, 40.6%, whereas distributions for those who were not administered the PDE were, males, 48.8%; females, 51.2%. Those who were administered the PDE also differed from those who didn’t on ethnicity (XZ (2)=103.27, p<001; PDE: Indians, 55.9%; General Population, 17.7%; Other, 26.4%; No PDE: Indians, 65.9%; General Population, 26.0%, Other, 8.2%). Consequently, both sex and ethnicity were used as covariates in subsequent ANOVAs involving this variable. Only 12 subjects met criteria for any SSPD (SPD=1, PPD=8, SZPD=3). In order to attain sufficient sample size for multivariate analyses, subjects were included in the index group who were one symptom below threshold for DSM-III-R (American Psychiatric Association, 1987) diagnosis (e.g. 4 out of 9 symptoms for SPD; Thaker, et al., 1993). This increased the size of the index group to 38, which corresponds to the 6% incidence rate reported in prior studies of schizotypy (Baron & Risch, 1987; Raine, 1991). A control group was formed consisting of 38 subjects, matched one-to-one on sex and ethnicity with the index group. None of the subjects in the control group reached threshold for more than one SSPD symptom, and all but one had a total score of I or 2 on the dimensional SSPD index. One subject had a total score of 6 on the SSPD index, but was included in order to obtain a perfectly matched sample. As a form of construct validation, the SSPD group and controls were compared on their SAE scores. The SSPD group R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. scored significantly higher than controls on schizophrenism (t=3.01, p=.004) but not anhedonia (t=.25, p>.79). Subjects were also administered the Present State Examination (PSE, Wing et al. 1974), scored using the CATEGO scoring program. One subject diagnosed with schizophrenia was excluded from the present study. Data Analysis Normality. Prior to tests of the main hypotheses, the distributions of the dependent measures were examined for deviations from normality. The family stress index had high skew (>2.9) and kurtosis (>8.87), as did the socioeconomic status variable (skew < -2.05, kurtosis >6.33). However, square root and natural log transformations of these variables also showed significant skew and kurtosis. Thus, the original variables were used for analyses. Distributions of all other variables did not deviate from normality. Hypothesis Testing. The primary data analyses will consist of analyses of variance (ANOVAs) and multiple regression analyses to examine the relationships between IQ, environmental stress variables, and SSPD. Effect sizes will be computed for significant findings for ANOVAs (f) and regressions (r), and categories of Cohen (1988) of small, medium and large will be employed. The hypotheses listed in the aims section were tested as follows: R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 39 Hypothesis la 1. SSPD group comparison (ANOVAs). In order to test Hypothesis 1 , intellectual performance of subjects with a diagnosis of SSPD at age 20-22 (n=38) were compared with an equal number of control subjects matched one-for-one on gender and ethnicity. It was predicted that those diagnosed with SSPD would score significantly lower on the cognitive indices than the control group. Four analyses of variance (ANOVAs) were performed comparing these groups on the global intelligence measure at age 3, as well as verbal, performance, and “full-scale” IQ scores at age 11. Ethnicity was included as a covariate as was described previously. To control for experiment-wise error, a Bonferroni correction of .05/4, or 0.0125, was used to maintain overall alpha=.05. 2. Anhedonia and schizophrenism group comparison (ANOVAs). These ANOVAs were then repeated first with index (n=51) and control (n=43) subjects on the SAE schizophrenism scale, and then with index (n=54) and control (n=29) groups created from the SAE anhedonia scale. Again, a Bonferroni correction of 0.0125 was used to control for Type I error. 3. Regression equations. Linear regression analyses were also performed to investigate the predictive value of intellectual performance on SSPD. Specifically, the following regression equation was entered: Y=b0+b1 x1 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 40 in which X ! is the IQ measure (IQ at age 3, total, performance or verbal IQ at age 11), and Y is the dimensionalized SSPD score (PDE score, schizophrenism, or anhedonia). Before these analyses were performed, the sample was split into two groups based on subject number (even or odd). The regression model described above was tested on one sub sample; then, these analyses were repeated on the other sub-sample for purposes of cross- validation. Significant replicated results at p<05 will be reported. Use of cross-validation in a second sample is used to control Type I error. B. Hypothesis lb 1. SSPD group comparison (ANOVAs). A series of 4 ANOVAs were used to test the hypothesis that males with a high degree of schizotypal traits would show greater IQ deficits than schizotypal females. SSPD at age 20-22, ethnicity, and sex served as the independent variables, and the four IQ indices the dependent variables for each of the four ANOVAs. Hypothesis la predicted that a significant diagnosis by sex interaction would be found, with male schizotypals showing the lowest IQS. Again, a Bonferroni correction of 0.0125 was applied to alpha to control for experiment-wise Type I error. As noted above, the sample size for each group was 38. 2. Anhedonia and schizophrenism group comparisons. These ANOVAs were then repeated using the high and low groups from the SAE anhedonia and schizophrenism scales, with sex as the second independent variable to examine diagnosis by sex interactions. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 41 3. Regression Analyses. Additionally, regression analyses were run to determine the effect of sex on the relationship between IQ and schizotypy as measured dimensionally. The equation, Y= b„ +btX ! + b2 x2 + bjX jX j, was run four times for each personality variable, in which x l is the IQ variable, x2 is sex (dummy coded), xtx2 is the interaction term, and Y is the personality variable. Before these analyses were performed, the sample was split into two groups based on subject number (even or odd). The regression model described above was tested on one sub sample; then, these analyses were repeated on the other sub-sample for purposes of cross- validation. Any significant change in R2 was reported, b-weights entered into the equation, and the new regression equation at each level of x2 (0 or 1) so as to establish the role of the moderator on the slope of b^ C. Hypothesis lc 1. SSPD group comparisons. In order to test the prediction that schizotypals will show a greater VIQ-PIQ discrepancy than controls, and that this effect will be stronger in males than females, a VIQ-PIQ index was created. The Performance IQ (PIQ) measure from age 1 1 was regressed onto the Verbal IQ (VIQ) measure, and the residual was used as the index. It will be referred to as ZPIQ. A two by two factorial ANOVA, similar to the one described in #2 above, was used to test the prediction that subjects with SSPD would score lower on this residual; that is, the PIQ would be lower than that normally predicted R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 42 by VIQ. Again, these ANOVAs were run first using SSPD (PDE) index and control groups, and then schizophrenism and anhedonia index and control groups. 2. Regression analyses for each of the three personality measures were run to test this prediction as well. The following equation was entered: Y= bo+biX,, in which xt is the ZPIQ index and Y is the personality variable. Additionally, the effect of sex will be tested in the following equation: Y= b0 +b,x, + b2 x2 + b3 x,x2 in which x, is the ZPIQ index, x2 is sex (dummy coded), and x,x2 is the interaction term, and Y is the personality variable. D. Supplementary analyses: Taxon vs. Dimension In order to determine whether findings lend support to the notion of SSPD as a taxon or as a dimension, several supplementary analyses were performed on equations showing significant results. First, the regression equations will be re-run without the index subjects from the ANOVA. The change in R2 from the full regression equation to this second equation will be analyzed for significance using Fisher’s r to Z transformation. If the R2 is significantly reduced from the full equation to the first alternative equation, the taxon hypothesis will be supported; alternatively, if no significant change occurs, the dimension hypothesis is supported. Second, for purposes of control, the original regression equations will be re-run without the comparison subjects from the ANOVA. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 43 Some attenuation is expected when subjects are deleted from the regression analysis, so the difference in R2 between the equations minus the index group and the equations minus the comparison group should be non-significant. Hypotheses 2a fModel D. 2b fModel ID. 2c fModel HD. and 2d fModel IV) These models were compared using a series of multiple regression analyses that examined the additive and interactional predictive strength of the family stress index on the development of SSPD. Each of the following equations was then repeated using the SES measure. Significant additive effects were further analyzed with an analysis of covariance (ANCOVA) so that effect sizes of the group comparison model and regression model could be compared. ANCOVAs could not be carried out for significant interaction effects; cell sizes were too small for analyses. 1. Additive effects (SSPD as dependent variable). First, as described in #A3 above, each of the four IQ indices (global intelligence age 3, VIQ, PIQ, and total IQ at 11) were entered as a predictor variable into a separate regression equation, Y= b0+blx1 , where X t is the IQ measure, and Y is the dimensionalized PDE score. Second, the additive effects of IQ and social stress were examined using the following equation: Y=b0+b,x,+ b 2X2 where x, is the IQ measure and x2 the environmental stress index (FSI or SES) score. Third, the significance in the change of R2 between equation I and equation 2 was tested. The delta R2 represents the unique variance contributed by the environmental stress R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 44 variable (Family Stress Index, Hollingshead SES rating). This third analysis tests the four competing models of SSPD development, where Model I (Hypothesis 2a) theorizes that family stress will not be a significant predictor of SSPD, Model II (Hypothesis 2b) predicts that increased family stress (and lower SES) will show an additive effect to the predictive value of lower IQ on later SSPD, and Model HI (Hypothesis 2c) hypothesizes that decreased family stress (and higher SES) will predict later SSPD. For these regression analyses, a total sample of 541 subjects was examined. Before these analyses were performed, the sample was split into two groups based on subject number (even or odd). The hypotheses described above were tested on one sub-sample; then, the analyses were repeated on the other sub-sample for purposes of cross-validation. Reported results include only replicated findings at p<05. 2. Interaction effects (SSPD as dependent variable). Hypotheses 2a, 2b, 2c and 2d were further examined using a regression analysis in which the potential interaction effects between family stress and IQ were examined. First, as mentioned in #1 above, the additive effects of IQ and family stress were examined using the following equation: Y= b0 + b[X, + b2 x2 where x, is the IQ measure and x2 the environmental stress score (FSI or SES). Second, the interaction effects of IQ and family stress were examined using the following equation: Y= b0 +bIx1 + b2 x2 + b3 x,x2 where Y is the dimensionalized PDE measure, x, is the IQ measure, x2 the environmental stress index score (FSI or SES), and x,x2 the interaction between the two. Third, the R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. significance in the change of R2 between equation 1 and equation 2 were tested. The delta R2 represents the unique variance contributed by the interaction variable (Jaccard et al. 1990; Cohen & Cohen, 1983). In order to further explore the nature of the interaction effect, equations with significant R2 change were re-run with the two independent variables transformed. This transformation, as recommended by Jaccard et al. (1990), “centers” the variables, such that the mean is subtracted from each score so that the new mean=0. This serves to reduce multicollinearity of the interaction term with the other independent variables, thus making the slopes more interpretable. At this point, a new interaction term was computed with the “centered” variables (note that this does not change the b weight for the interaction term or its significance; Jaccard et al., 1990). These transformed variables entered into the regression equation Y= b0 + b3 x, + b2 x2 + b3 x,x2 Following the guidelines of Jaccard et al. (1990), the resultant regression equation is solved at three levels of the moderator (x2): low, medium and high. This procedure yields three equations of the form Y= b0 + b ^ . These three regression lines are then plotted, and the significance of the different b,s can then be tested, thereby showing the relationship between x, and Y at different levels of the moderator.1 As mentioned above, the sample was split into two groups based on subject number (even or odd) for purposes of cross-validation. Reported results include only replicated findings. 'This analysis was accomplished using BASIC code taken from Jaccard et al. (1990), p. 83-88. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 46 The preceding regression analyses examine both the interaction of environmental stress (FSI, SES) and IQ and the additive effects of the two possible predictors. Hypothesis 2a (Model I) predicts that there will not be significantly more SSPD in unstable families. More specifically, Hypothesis 2a predicts that both the main effect of family stress and the interaction effect will be non-significant, showing that environmental stress is not a moderator of the relationship between IQ and SSPD. Hypothesis 2b (Model H) predicts that there will be an additive effect, with both IQ and psychosocial stress playing an independent role in predicting SSPD. Hypothesis 2c predicts that there will be a significant interaction, with low IQ and high family stress combining to result in SSPD. In other words, low IQ would only be related to SSPD for subjects from unstable homes (two-hit model). Hypothesis 2d predicts that if the main or interaction effect of environmental stress is significant, lower family stress (and lower IQ) would be related to increased schizotypy. 3. Additive and interaction effects (anhedonia and schizophrenism as dependent variables). The preceding series of regression analyses testing Models I, II, and HI (Hypothesis 2) was repeated using each SAE variable (schizophrenism and anhedonia) as the dependent variable. The 760 subjects who completed the questionnaire were randomly split into two groups, and then compared for the purposes of cross- validation. The benefits of using the SAE are twofold: (1) the questionnaires provide a broader range of scores compared to the dimensionalized PDE, and (2) factors of R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 47 schizotypal personality roughly correspond to positive and negative features in schizophrenia can be examined for differential results. Hypothesis 2e fModel VI Regression equations: mediator test. Regression analyses will be used to test IQ as a single, dominant mediator of the relationship between environmental stress (FSI, SES) and SSPD (Model V). This prediction asserts that environmental stress leads to lower cognitive test scores and to SSPD, and that the relationship between IQ and SSPD can be entirely accounted for by their shared relationship with environmental stress. In order to establish mediation, the following equations must be tested (Baron & Kenny, 1986): (1)Y= bo+fyx,, where IQ (mediator) is Y , and the stressor variable (independent variable) is xt; this equation tests the predictive value of psychosocial stress on IQ, (2) Y= b0+b,xt, where the SSPD variable is regressed onto the IQ variable, which examines the predictive value of IQ on personality, and (3) Y= bo+b^ft- b2 x2 , where xt is the stress variable (mediator), x2 IQ (TV), and SSPD is Y. If significant effects are evident in the first two equations, and the R2 change for the mediator term in the third equation is non significant, environmental stress can be established as a mediator in the relationship between IQ and SSPD (Baron and Kenny, 1986). Results Hypothesis la: The prediction of lower childhood intellectual test scores relating to higher levels of SSPD was supported; lower IQ scores at age 1 1 (total IQ, VIQ, PIQ) R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 48 related to higher scores of schizophrenism at age 16. Effect sizes2 of the ANOVAs (medium to large range) were larger than those of the regression equations (small to medium range). No significant relationships were revealed in which cognitive performance at age 3 played a role. Additionally, this finding was specific to the age 16 schizophrenism scale; SSPD measured at age 20-22 and anhedonia at age 16 did not show a significant relationship to IQ. ANOVA and multiple regression results are reported below; significance for the ANOVAs is reported at the .0125 level and significance for the regression analyses is reported only when replicated in both sub-samples utilized in the cross-validation technique. Figure 2 depicts the results from the ANOVA, and Figures 3 a, 3 b, and 3 c, as well as Table I, illustrate significant results for the regression equations in the examination of Hypothesis la. ANOVAs on SSPD index and control groups SSPD from the PDE at age 20-22: The ANOVAs comparing matched index and control groups on the four IQ indices: cognitive testing at age 3 (n’s for high and low SSPD group = 30 and 25, respectively), total IQ at age 11 (n=32, n=27), VIQ at age 1 1 (n=32, n=27), and PIQ at age 11 (n=32, n=27) were not significant (all F<44, all p>.50). Categories of “small”, “medium” and “large” effect sizes from Cohen (1988) were employed to compare f (for ANOVAs) and r (for regressions). R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 4 9 Anhedonia scale from the SAE at age 16: The ANOVAs comparing index and control groups on age 3 IQ (n=48, n=l9) and the three IQ indices at age 11 (n=50, n=24). The four IQ indices were non-significant (all F< 41, all p>.52). Schizophrenism scale from the SAE at age 16: The ANOVAs comparing index (n=36) and control (n=34) groups showed no significant relationship between schizo phrenism and IQ at age 3 (F<1.5, p>.23), or between schizophrenism (index group, n=42; control, n=38) and age 11 VIQ (F<3.9, p>.05). However, a significant group effect was revealed for the schizophrenism groups and total IQ at age 11 (F=10.54, p=.002); as shown in Figure 2, the index group performed more poorly (mean=5.27, SD=1.68) than did the control group (mean=6.83, SD=1.51). The effect size of that analysis was in the medium-large range (f=.378). Additionally, as shown in Figure 2, a significant group effect was found for PIQ at age 11 (F=10.15, p=.002), with the index group again showing lower scores (mean=5.25, SD=1.80) than the control group (mean=6.79, SD=1.84). The effect size was also in the medium-large range (f=.371). A non-significant effect was found for verbal IQ at age 11 (F=3.81, p=.055), but the index group showed lower scores (mean=5.32, SD=2.09) on VIQ than did the controls (mean=6.93, SD=1.66), and the effect size was in the medium range (f=.227). Regression analyses. Y= b0 +btx, SSPD age 20-22: Regression equations testing the predictive value of IQ on SSPD at age 20-22 resulted in no significant replicated findings (all R2 < .004, all F< 1.1, all p>.29). R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 5 0 Anhedonia at age 16: No significant findings were revealed (all R2 < 002, all F< 28, all p>.60) when anhedonia was regressed onto the IQ variables. FIGURE 2 7 6 5 4 3 2 1 0 Low Schizophrenism High Schizophrenism | Age 11 Total IQ (F=10.54, p= 002) ! g Age 11 VIQ (F=3.81, p= 055) u Age 11 PIQ (F=10.15, p=.002) Figure 2. Comparison of total IQ, VIQ, and PIQ subtest scaled scores (age 11) of subjects with high scores on the age 16 schizophrenism scale (n=34) to those scoring low on schizophrenism (n=36). High schizophrenism subjects have lower IQ scores, with this effect achieving statistical significance (p< 0125) for Total IQ and PIQ scaled scores. VIQ scaled scores showed increased variance, leading to statistical nonsignificance. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 51 Schizophrenism at age 16: When IQ at age 3 was tested for predictive value of later schizophrenism, it did not emerge as a significant predictor (R2 <.011, F<3.03, p>.08). However, for both sub-samples, (even-numbered subjects, n=345; and odd- numbered subjects, n=355), lower total IQ at age 11 significantly predicted higher schizophrenism at age 16 (evens: R^.071, F=26.15, p<00l, B=-.423; odds: R^.051, F=18.82, p<001, B=-.338). The effect size was in the small to medium range (odds: r=.224, evens: r=.251). Refer to Figure 3a-b for a scatterplot of the relationship between total IQ and schizophrenism. Additionally, as depicted in Figure 4a-b, lower verbal IQ significantly predicted higher schizophrenism (evens: R^.043, F=15.55, p<001, B=-.292; odds: R^.020, F=7.12, p=.008, B=-.177). The effect size for this equation was also in the small to medium range (odd: r=. 138; even: r=.205). Finally (Figure), lower performance IQ emerged as a significant predictor of higher schizophrenism (evens, R2 =.066, F=24.11 , p<001, B=-.376; odds, R^.053, F=19.89, p<00l, B=-.322). The effect size was again in the small to medium range (odd: r=.230; even: r=.239). These significant results are summarized in Table I. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 5 2 Group Y TABLE 1 Xi R2 F P b, J - b0 even schizophrenism Total IQ - age 11 .071 26.15 <001 -.423 9.41 odd schizophrenism Total IQ - age 11 .051 18.82 <.001 -.338 8.34 even schizophrenism Verbal IQ - age 11 .043 15.55 <001 -.292 8.85 odd schizophrenism Verbal IQ - age 11 .020 7.12 =.008 -.177 7.91 even schizophrenism Perf. IQ - age 11 .066 24.11 <.001 -.376 9.11 odd schizophrenism Perf. IQ - age 11 .053 19.89 <001 -.322 8.74 Table 1 . Significant regression weights lb,! for the equation Y=b0 + b,xt , where Y= schizophrenism as measured by the SAE at age 16 and xt = the IQ index listed in column three of the table. Note that the intercept (b„) is provided for completeness only; it is not directly interpretable since the IQ variables (the respective x/s) cannot equal 0. FIGURE3a Even-numbered cases 141 12- g 10- e £ i s w 0 2 6 4 8 10 12 scaled full-scale IQ age 11 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 53 FIGURE 3b Odd-numbered cases 1 4 " 1 2 * e e 8 0 2 6 10 12 4 scaled full-scale IQ age 11 Figure 3a-b. Scatterplot of the relationship between age 11 full-scale IQ scaled scores and age 16 schizophrenism for even-numbered cases (Figure 3 a, n=345) and odd-numbered cases (Figure 3b, n=355). Open circles represent a single case at a given x,y coordinate. “Sunbursts” represent multiple cases at a given coordinate, with the number of spokes depicting the number of subjects at that point. Regression line depicts significant linear relationship between the two variables, such that lower IQ is related to high schizophrenism. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 54 IS ' 1 4 * 1 2 - E 1 0 ' i9 ■ 6 < n s ■ IS ' 14" 1 2 - FIGURE 4a Even-numbered cases < | > < j > o o A o o 4c 4 4 ° o 4 4r 4c 4^ o 4 o 4 9 4 ^ 4 - . ■ 4 ~ ° 4 ° • 4 4 - 4 4c 4~ 4^ 4c ..................................+ 4 4 < j > 4 . 4< 4< 4 ^ 9^ 4 o o ,-A^ 9 4 c 4 4 ^ ° o 4 4 ~4~ 4 c 4 4 4 o 4^ 4 c ~4“ -4 - 4 o 4 4< 4 4 o o o .A ^ T — 8 10 'T ' 12 Scaled VIQ age 11 FIGURE 4b Odd-numbered cases o c j > 4 4 4 4 o 4 4 4 4 ~ ° 4. 4c 4 - 4 4 . 4 4 4 4 - 4 ^ 4c 4 -4~ ° j # 4 - 4s- 4s- ~ 4 - ° 4 . 4 _ 4 c 4 4 4 c 4 f # 4 4c 4c ° ^ 4 ^ 4 4^ 4~ 4 o 4 4 4 : | ^ o o o 4 4 4 4 4 - o A. 4 4^ 4^ 14 4 10 — » - - 1 2 14 Scaled VIQ age 11 Figure 4. As in Figure 3 above, for Verbal IQ scaled scores. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. :1 - 0.043 :* = 0.020 55 FIG U R E 5a Even-numbared cases 1 6 ' 1 4 " 1 2 - E 10 * 6 * 4 - 2- 0 4 2 6 8 10 12 Scaled PIQ age 11 FIGURE 5b Odd-numbered cases 16" 14- o Q Q 12- o < j > o .A. A . o o o o A a , 4 A , o E l 0* o ^ ^ O ( |> o < & o o ■ $ } £ ■ A o # ■ 4 - ~4 - o o is o -p ic o o <fl 8 ■ < j> ^ ' A ' # ^ ^ o Ac o o 4 ■ 4 o < j> A r A , A . 4 o o o o -<£— A ^ . * o o 2- O < J > ( j> o A 0 ■ o < |> A , $ o W 1 c m 2 V 1 4 6 ■ 8 10 1 Scaled PIQ age 11 Figure 5. As in figure 3 above, for Performance IQ scaled scores. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. R2 - 0.066 R2 = 0.053 56 Hypothesis lb: The hypothesis that males would show stronger relationships between lower IQ scores and higher SSPD was not supported by the findings. SSPD age 20-22: All diagnostic group by sex interactions for cognitive test at age 3, total IQ at age 11, VIQ and PIQ at age 11 were non-significant (all F< 33, all p>.57). Anhedonia at age 16: All personality group by sex interactions were non significant (all F< 2.2, all p>. 15). Schizophrenism at age 16: All personality group by sex interactions were non significant (all F<3.5, all p>.06). Regression equations Y= b „ b2 x2 +b3 x1 x2 : All regression equations with a sex*IQ variable interaction term showed non-significant R2 change (all F<2.19, all p>. 14) When compared to regression equations without the interaction term. Thus, sex is not a significant moderator of the relationship between cognitive performance and SSPD. Hypothesis lc: This hypothesis, which predicted that subjects with higher levels of SSPD would show a greater differential between PIQ and VIQ scores, such that their PIQ was deficient was supported by the results from the ANOVA (effect size=medium) and the regression equation (effect size=small to medium), when schizophrenism was the R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. SI dependent variable. Sex effects were also explored on this model, but did not approach significance. See Figures 6a, 6b and 6c for a graph and scatterplot of these results. ANOVAs on SSPD index and control groups SSPD at age 20-22: The ANOVA comparing matched index (n=32) and control groups (n=27) on the ZPIQ index (F<47, all p>.49). No group by sex effects were evident (F<33, p>.57). Anhedonia at age 16: The ANOVA comparing index (n=50) and control groups (n=24) on the ZPIQ index were not significant (F<1.0, p>.31). No group by sex effects emerged (F<86, p>.35). Schizophrenism scale on the SAE at age 16: The ANOVA showed a trend towards a significant group effect (F=4.98, p=.029) such that the index group (n=42; mean=-.040, SD=.957) had significantly lower scores on the ZPIQ residual score as compared to the control group (n=38; mean=.347, SD=1.I2). The effect size for this analysis was in the medium range (f=.259). Results are illustrated in Figure 6a. Regression analyses. Y= b0+b1 x1 SSPD age 20-22: Regression equations testing the predictive value of the ZPIQ index on SSPD at age 20-22 resulted in no significant findings (R2 <006, F<1.66, p>. 20). Sex did not emerge as a moderator in this equation. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 58 Anhedonia at age 16: No significant findings were revealed (all R^.OOl, all F< 22, all p>.64) when anhedonia was regressed onto the index. Sex did not emerge as a moderator in this equation. Schizophrenism at age 16: When the index was tested for predictive value of later schizophrenism, it emerged as a significant predictor with a small to medium effect size (even: r=.141; odd: r=.!76). Subjects with higher levels of SSPD showed a greater discrepancy between their verbal and performance IQ scores (evens: R2 =.027, F=9.60, p=.002, B=-.489, n=345; odds: R^.030, F=10.99, p=.001, B=-.477, n=355). This is illustrated in Figures 6b and 6c. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 5 9 FIGURE 6a Low Schizophrenism High Schizophrenism Age 11 ZPIQ (residual of PIQ regressed on VIQ) Figure 6. Relationship between age 16 schizophrenism and ZPIQ, the standardized residual of PIQ regressed on VIQ. Figure 6a. Group comparison (ANOVA) of subjects with high vs. low schizophrenism scores at age 16. High schizophrenism group shows a trend toward significantly lower ZPIQ scores (F=4.98, p=.029). Figure 6b. c. Scatterplot depicting results of regression analysis for Y= b0 +b,x,, where Y=schizophrenism and xt = ZPIQ for even-numbered cases (Figure 6b) and odd-numbered cases (Figure 6c). R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 60 FIG URE 6b Even-numbered cases 1 6 ' 1 4 " 1 2 - 4 « -1 1 2 -2 0 3 ZPIQ (stand, res. of scaled PIQ regressed on VIQ) FIGURE 6c Odd-numbered cases 1 6 ' 1 4 - 1 2 * E 1 0 * a I , S 8 ■ 2 * -3 ■ 1 1 -2 0 2 3 ZPIQ (stand, res. of scaled PIQ regressed on VIQ) R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. R2 = 0.027 R2 = 0.030 61 Supplementary analyses: Taxon vs. Dimension Regression equations with significant results were further analyzed; specifically, those in which Y=schizophrenism, and x,=total IQ, VIQ, PIQ, and ZPIQ. When the schizophrenism index group was removed from the sample, and the regression equations re-run, the change in R2 was not statistically significant (as measured by Fisher’s r to Z transformation), lending support to the dimensional hypothesis (all Z< 51, all p>61). When the schizophrenism comparison group from the ANOVA was removed from the original regression equation, change in R2 was also not significant (all Z< 87, all p>.36). The difference in R2 between the equations without index group and without comparison group were also non-significant (all Z<76, all p>.44) Hypotheses 2a. 2b. 2c. and 2d: Additive effects: Significant results for the equation, Y= b0 +btx,+ b2 x2 are listed below. Again, in each equation, x, is the IQ variable (IQ at age 3, total IQ at age 11, VIQ and PIQ at age 11, ZPIQ index); x2 is the environmental stressor variable (FSI or SES), and Y the SSPD variable (SSPD age 20-22, anhedonia and schizophrenism at age 16). It is important to note that the SES variable is scored so that higher scores = lower SES. Hence, a negative slope would indicate a positive linear relationship. Listed below are the squared multiple correlation, value of the F-test, significance of the regression equation, and significant t-tests for individual slopes. Additionally, level of the F test and the p-value of the R2 change for the stressor variable are included. As detailed below, additive effects R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 62 (R2 change) emerged for the ZPIQ residual and SES when schizophrenism was the dependent measure. A follow-up ANCOVA was performed in order to determine whether this additive relationship was also evident in a categorical model. Significant results are listed in Table 2. 1. SSPD at age 20-22 (Y) and FSI (xT): Y= b„+b,Xi + b2 x2: No significant replicated results emerged for any of the IQ variables (x,) when FSI was x2(all R2 <015, all F<2.05, all p>.13). R2 change: No significant replicated results were found for the additive effects of family stress to any of the IQ measures (all F<2.6, all p>. 11). 2. SSPD m and SES fx,): Y= b0 +b,Xt + b2 x2 : No significant replicated results emerged for the full regression equation (all R2 <010, all F<1.47, all p>.23). R2 change: R2 change did not show significant replicated results (all F<2.24, all P> 13) 3. Anhedonia at age 16 HO and FSI fx,) Y= b0 +b,x, + b2 x2 : No significant replicated results emerged for the full model (all R2 <010, all F<1.46, all p>.23). R2 change: R2 change did not show significant replicated results (all F<2.36, all p>12). R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. TABLE 2 group x, x 2 R2 F P b, t P b2 t P c4 < 1 F P even ZPIQ - 1 1 SES .033 5.79 =.003 -.321 -1.92 = 055 .037 2.90 = 037 .013 4.40 .037 odd ZPIQ - 1 1 SES .064 12.13 <001 -.407 -2.84 =.005 .052 3.55 <.001 .033 12.62 <001 Table 2. Regression models in which SES has a significant additive role in the prediction of age 16 schizophrenism. Regression equations take the form Y=b„ + b,x, + b2 x2 , where Y=schizophrenism at age 16, x,=IQ variable listed in the column, and x2 = SES. R2 in column four represents the amount of variance accounted for by the full model, while AR2 represents the change in R2 when SES is added to the model. ON u > 64 4. Anhedonia fYl and SES fx -/> Y= b0 +btx1 + b2 x2: No significant replicated results emerged for the full equation (all R2 < 007, all F<1.07, all p> 35). R2 change: R2 change did not show significant replicated results (all F<1.87, all P> 17). 5. Schizophrenism at age 16 (Y) and FSI a. Y= b0 +b,xt + b2 x2 : R2 for the full model in which xt = total IQ at age 11, revealed significant prediction of schizophrenism at age 16. On one subsample (even), R^.061, F=10.59, p<001, n=327; these results were replicated in the cross-validation sample (odd; R2 =.054, F=9.43, p<00l, n=334). However, in both samples, t-tests on the slopes for the independent variables revealed that only the total IQ at age 11 variable achieved significance (even: B=-.398, t=-4.60, p<.001; odd: B=-.328, t=-4.13, p<001), while the FSI variable did not (even: t=.366, p=.714; odd: t=.961, p=.337). R2 change: R2 change for the stress variable was not significant, again indicating only a significant relationship between total IQ at age 11 and schizophrenism ( F< 93, p>33). b. Y= b0 +b1 x1 + b2 x2 : The full regression model in which X ; =VTQ significantly predicted higher schizophrenism at age 16. On one subsample (even), R^.040, F=6.74, p=.001, n=328; these results were replicated in the cross-validation sample (odd: R^.023, R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. F=3.84, p=.023, n=335;). However, in both samples, t-values revealed that only the VIQ variable achieved significance (even: B=-.284, t=-3.67, p< 001; odd: B=-. 165, t=-2.43, p=.016), whereas the FSI variable did not (even: t=.440, p=.66; even: t=.983, p=.326). R2 change: R2 change for the FSI variable was not significant, leading to the conclusion that family stress does not contribute in an additive fashion to the predictive relationship between verbal IQ at age 11 and schizophrenism ( F< 97, p>32). c. Y= b0 +b,x, + b2 x2 : The model in which xt=PIQ significantly predicted higher schizophrenism at age 16. On one subsample (even), R^.055, F=9.53, p<001, n=327; these results were replicated in the cross-validation sample (odd: R^.057, F=10.07, p<001, n=334;). However, in both samples, t-values revealed that only the PIQ variable achieved significance (even: B=-.348, t=-4.37, p=.008; odd: B=-.313, t=-4.28, p<.001) R2 change: R2 change was again not significant (F<1.2, p>.27). 6. Schizophrenism fY) and SES fb ^x -Z l a. Y= bo+bjX, + b2 x2 : The model in which X[=IQ at age 3, significantly predicted higher schizophrenism at age 16 (evens: RH025, F=3.63, p=.028, n=288; odds: R^.049, F=7.81, p<.001, n=306). The t-values reveal that IQ (even: B=.047, t=2.45, p=.015; odd: B=.051, t=3.54, p<001), and not SES show significance (even: t= -1.19, p=.236; odd: t=- 1.83, p=.068). R2 change: R2 change is non-significant, F= . 125, p=. 104 (odd); this result is replicated in the cross validation sample (even: F=.601, p=.471). Thus, it seems that SES R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 6 6 does contribute significant prediction to later schizophrenism when IQ at age 3 is controlled. b. Y= b0+btx, + b2 x2 : This regression model, in which xt=total IQ at age 11, predicted higher schizophrenism (even: R^.065, F=11.84, p< 001, n=343). This result was replicated in the cross validation sample (odd: R^.070, fi=13.54, p< 001, n=354). In both samples, only the t-value for the IQ score was significant (even: B=-.367, t=3.94, p<.00l; odd: B=-.261, t=-3.19, p=.002), whereas the t-values for SES had different levels of significance in the two subsamples (even: t=.366, p=.714; odd: t=2.75, p=.006). R2 change: When SES was added to the regression model, R2 change was not significant in one sub-sample (odd: F=.6l3, p=.434), but was in the other sample (even: F=7.58, p=.006). c. Y= b„ +bi*, + b2 x2 : Lower age 1 1 VIQ scores and lower SES predicted higher schizophrenism (even: R^.050, F=8.91, p<001, n=344). This result was replicated in the cross validation sample (odd: R^.050, F=9.25, p<.00l, n=355). The t-values for VIQ were significant only the even subsample (B=-.245, t=-3.15, p=.002); in the odd subsample, SES was significant (B=.051, t=3.39, p=.001). R2 change: R2 change was non-significant in one subsample (odd: F<2.58, p<. 11), but significant in the other (even: F= 11.51, p=.001). R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 67 d. Y= b„ +b,Xj + b2 x2 : The regression model in which x,=PIQ significantly predicted SSPD (R^.059, F=10.74, p< 001, n=343). This result was replicated in the cross validation sample (R^.074, F=14.05, p< 001, n=354). For both subsamples, the t- values for PIQ were significant (even: B=-.315, t=-3.66, p<00l; odd: B=-.257, t=-3.43, p=.001), while the values for SES had different levels of significance in each subsample (even: t=.214, p= 831; odd: t=2.82, p=.005). R2 change: R2 change was significant only in one sample (odd: F=7.92, p= 005); this result is not replicated in the cross validation sample (F=.890, p>.34). e. Y= b0 +bjX, + b2 x2: Lower ZPIQ index scores (standardized residual described above) and lower SES predicted higher SSPD (even: R^.033, F=5.79, p=.003, n=343). This result was replicated in the cross validation sample (R2 =.064, F=12.13, p< 001, n=354). In both sub-samples, the SES variable reached significance (even: B= 037, t=2.90, p=.037; odd: B=.052, t=3.55, p<.001). The ZPIQ index reached significance in only one subsample (odd: B=-.407, t=-2.84, p=.005), and showed a trend in the other sample (even: B=-.321, t=-l .925, p=.055). R2 change: R2 change was significant in both subsamples (even: F=4.40, p=.037; odd: F=I2.62, p<.001). Effect sizes were in the small to medium range (even: d=.230; odd: d=.370). Analysis of Covariance: In order to determine whether this additive relationship was also evident in a categorical model, an analysis of covariance (ANCOVA) was carried R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 6 8 out. Results showed that when SES was entered as covariate, group effect was non significant (F=2.12, p=. 150), and effect size was in the small to medium range (f=. 173). Interaction effects: The regression equation Y = b„ + + b2 x2 + b3 x,x2 was entered and again, total R2 was significant for all equations in which Y=schizophrenism. However, the change in R2 was significant (p<.05) only for the ZPIQ standardized residual score. This one interaction effect was analyzed further, as will be explained below, after a brief presentation of results for the other dependent variables (SSPD at age 20-22, anhedonia at age 16). Significant results are listed in Table 2. 1 . SSPD at age 20-22 (Y). 10 fx,T FSI fx-A and IO/FSI interaction fxIx -.1 Y = b„ + bjX , + b2 x2 + b3 x,x2: Four of the five regression equations exploring the interactive effect of IQ variables (IQ at age 3, total IQ at age 11, VTQ, and the ZPIQ index) were non-significant in both samples (all F<2.6, all p>.053). However, for the equation where bpc, = PIQ, results were significant in one subsample (R ^.031, F=2.99, p=.031, n=283), and showed a trend towards significance in the other sample (odd: R^.026, F=2.57, p=.055, n=292). R2 change: R2 change for the interaction term (b ^ X j) was not significant for IQ at ages 3 and 11 (all F<3.8, all p>.054). R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 6 9 2. SSPD at age 20-22 fYT 10 SES (xA and the IO/SES interaction fxtx2 )_ Y = b0 + b,x, + b2 x2 + b3 x,x2: No significant results for total R2 were replicated for these regression analyses (all R2 < .014, all F<1.27, all p>.28). R2 change: No significant R2 change for the interaction term was found in both sub-samples (all F<1.23, all p>.27). 3. Anhedonia at age 16 (YI. 10 (xt). FSI fx2 ~ ) and IO/FSI interaction (x1 xO Y = b0 + b(X ( + b2 x2 + b3 xlx2 : The five regression equations exploring the individual and interactive effect of IQ variables (IQ at age 3, total IQ at age 11, VIQ, PIQ and the ZPIQ index) and FSI were non-significant in both samples (all R2 < .019, all F<2.08, all p>.10). R2 change: R2 change for the interaction term was not significant (all F<3.3, all p>.07) for any of the equations. 4. Anhedonia (Y \ 10 (b1 x1 T SES (b2 x:V and the IO/SES interaction (b2 xtx.,l Y = b0 + btx, + b2 x2 + b3 x1 x2 : No significant results were replicated for these regression analyses (all R2 < .014, all F<1.27, all p>.28). R2 change: No replicated significant R2 changes for the interaction term were found (all F<2.5, all p>. 11). 5. Schizophrenism at age 16 fY\ 10 fb,x,l. FSI fb-.x-.V and the IO/FSI interaction ('b3 x,x2) With the exception of the equation in which b,x,=lQ at age 3, the equations (for the four R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. IQ variables at age 11) showed significant results for the full model in prediction of later SSPD; however, R2 change for the interaction effect was significant only for the ZPIQ standardized residual score. Significant findings are outlined in more detail below: a. Y = b „ + b,x, + b2 x2 + b3 x,x2 : The equation in which xt=total IQ at age 11 was significant for both sub-samples (odd: RH054, F=6.34, p< 001, n=334; even: R^.061, F=7.05, p< 001, n=327). Again, lower total IQ at age 11 predicted higher levels of SSPD (odd: B=-.341, t=-4.02, p< 001; even: B=-.403, t=-4.39, p< 001). R2 change: R2 change for the interaction term was not significant in either subsample (F< 196, p>.65). b. Y = b0 + b,x, + b2 x2 + b3 x,x2 : The regression equation in which yiQ=x,, was significant in both sub-samples (odd: R^.026, F=2.99, p=.031; even: R^.044, F=5.04, p=.002). Again, lower VIQ predicted higher levels of schizophrenism (odd: B=-.l43, t=- 2.02, p=.044; even: B=-.252, t=-3.09, p=.002). R2 change: R2 change for the interaction term was not significant for either sub sample (F<1.63, p>.203). c. Y = b„ + b,xt + b2 x2 + b3 xtx2 : The regression equation in which x,=PIQ was significant in both sub-samples (odd: R^.061, F=7.16, p<.001; even: R^.058, F=6.69, P<001). Again, lower PIQ predicted higher rates of schizophrenism (odd: B=-.348, t=- 4.40, p<001; even: B=-.378, t=-4.44, p<.001). R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 71 R2 change: R2 change for the interaction term was not significant in either sub sample (F<1.35, p> 24). d. Y = b0 + b,xt + b2 x2 + b3 x,x2: When x ^ h e ZPIQ standardized residual score, the regression equation was significant for both sub-samples (odd: R^.048, F=5.51, p= 001; even: R^.034, F=3.79, p=.011). In both sub-samples, a lower ZPIQ discrepancy score predicted higher rates of schizophrenism (odd: B=-.6, t=-3.81, p< 001; even: B=- .564, t~3.23, p=.00l), and the interaction term is significant (odd: B=.055, t=1.98, p=.049; even: B=.067, t=2.07, p=.039); however, the FSI term was not significant (odd: t=1.34, p=.l82; even: t=.238, p= 812). R2 change: R2 change is significant for this equation in both sub-samples (odd: F=3.92, p=.049; even: F=4.29, p=.039). Effect size was small (odd: r=.l 10; even: r=.l 14). In order to further explore the nature of the interaction effect, the stress variable was “centered;” its mean subtracted from each score to reduce problems of multicollinearity for the interaction term (Jaccard et al. 1990). At this point, a new interaction term was computed with the new stress variable and the standardized residual (ZPIQ). The new stress variable and interaction term were then substituted in the equation, and the following equation was re-run: Y= b0 + b,xt + b2 x2 + b3 x,x2, such that, Schizophrenism = b0 + b, ZPIQ+ b2 FSI + b3 (ZPIQ)(FSI) Following the guidelines of Jaccard et al. (1990), the resultant regression equation is solved at three levels of the FSI variable (x^): low (x2 =-2.15), medium (0=mean) and high (5.36). This procedure yields three equations of the form Y= b0 + b.x,. The following R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 72 three regression lines were then plotted (see Figure 7), and the significance of the different slopes was tested, thereby showing the relationship between xt and Y at different levels of the moderator. 1. Y=22.15 + (-.564)(ZPIQ): this yielded t=-3.23, df=323, p< 01 2. Y=22.17 + (-.42l)(ZPIQ): this yielded t=-2.58, dfi=323, p<.0l 3. Y=22.21 + (-.065)(ZPIQ): this yielded t=-.269, df=323, p is non-significant. These results describe the relationship of ZPIQ to schizophrenism at different levels of family stress: At the lowest FSI score, ZPIQ has the strongest predictive value on later schizophrenism; at levels of high family stress, ZPIQ has virtually no predictive value, thus supporting Model IV (Figure 7). 5. Schizophrenism at age 16 fYI. 10 fx,\ SES (xA and the IO/SES interaction fx,x: ) All five regression equations for IQ variables at ages 3 and 11 showed significant results; however, change in R2 for the interaction term was not significant for any of the equations. An outline of significant results follows: a. Y = b„ + b,x, + b2 x2 + b3 x,x2 : The equation in which b^^total IQ at age 3 was significant for both sub-samples (odd: R^.051, F=5.46, p=.001, n=306; even: R^.033, F=3.23, p<023, n=288). R2 change: R2 change for the interaction term was not significant in either subsample (F<2.38, p>.!2). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 73 Figure 7 Model IV interaction effect 8.5 E 8 1 7.5 1 7 Q . • 6.5 6 • 2 1 0 1 2 ZPIQ (stand, res. of PIQ regressed on VIQ) low stress (b= -.56, t= -3.23, p<.01) medium stress (b= -.42, t= -2.58, p<.01) high stress (b= -.06, t= -0.27, NS) Figure 7. Three regression lines showing the relationship between age 11 ZPIQ and age 16 schizophrenism at three different levels of the Family Stress Index (FSI). At the lowest FSI score (low stress, solid line), ZPIQ has the strongest predictive value on later schizophrenism; at levels of high family stress (dotted line), ZPIQ has virtually no predictive value, thus supporting Model IV. These regression lines were derived following the guidelines of Jaccard et al. (1990): the following regression equation Schizophrenism = b0 + bt ZPIQ+ b2 FSI + b3 (ZPIQ)(FSI) is solved at three levels of the centered FSI variable (xj: low (x2 =-2.15), medium (0=mean) and high (5.36). This procedure yields three equations of the form Y= b0 + bx x,, where Y=schizophrenism and x,=ZPIQ. The following three regression lines are depicted in the figure: 1 . For low stress:Y=22.15 + (-.564)(ZPIQ): this yielded t=-3.23, df=323, p< 01 2. For medium stress: Y=22.17 + (-.421)(ZPIQ): this yielded t=-2.58, df=323, p< 01 3. For high stress:Y=22.21 + (-.065)(ZPIQ): this yielded t=-.269, df=323, p is non significant. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. b. Y = b0 + btx, + b2x2 + b3x,x2: The equation in which bjX^otal IQ at age 11 was significant for both sub-samples (odd: 1*3=.072, F=9.02, p< 001, n=354; even: RH067, F=8.16, p<001, n=343). R2 change: R2 change for the interaction term was not significant in either subsample (F<.81, p>.36). c. Y = b0 + b,x, + b2x2 + b3xtx2: The regression equation in which VTQ=b1 xl was significant in both sub-samples (odd: RH050, F=6.I5, p< 001; even: R^.051, F=6.08, p< 001). R2 change: R2 change for the interaction term was not significant for either sub sample (F<43, p>.51). d. Y = b0 + bjXt + b2x2 + b3x,x2: The regression equation in which b,x,=PIQ was significant in both sub-samples (odd: R^.077, F=9.71, p<.001; even: R^.061, F=7.35, P<001). R2 change: R2 change for the interaction term was not significant in either sub sample (F< 1.03, p>.31). e. Y = b„ + bjXj + b2x2 + b3x,x2 : When bjX^he ZPIQ standardized residual score, the regression equation was significant for both sub-samples (odd: R^.068, F=8.54, p<001; even: R^.034, F=3.93, p=.009). In one of the sub-samples, a lower SES rating R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 75 predicted higher rates of schizophrenism (odd: B=. 061, t=3.68, p< 001). However, this was not replicated in the other subsample (even: B= 033, t=1.60, p=. 110). R2 change: R2 change for the interaction term is not significant in either sub sample (F<1.36, p> 24). Hypothesis 2e Regression equations: mediator test. This prediction asserts that environmental stress leads to lower cognitive test scores and to SSPD, and that the relationship between IQ and SSPD can be accounted for by their shared relationship with environmental stress. In order to establish mediation, the following equations were tested (Baron & Kenny, 1986): (1) Y= bo+bjXj, where IQ (mediator) is Y , and the stress variable (independent variable) is x,: These regression equations were non-significant when FSI=xl (all R2 <007, all F<3.3, all p>.07), and when IQ at age 3=Y (R^.OOO, F<29, p>.59), but were significant for all IQ at age 11 measures (all p<00l, total IQ: R2 ^ 119, F=85.03; VTQ. R2 = .095, F=66.45; PIQ: R2 = .099, F=69.31; ZPIQ: R2 = .025, F=16.18) (2) This led to the second equation, Y= b0 +b,xb where the SSPD variable is regressed onto the IQ variable, which examines the predictive value of IQ on personality. These results were reported under Hypothesis la and lc above, and significant results can be seen in Table 1 . Only results when Y=schizophrenism were significant. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 76 (3) Now that the possible mediator equations were limited to those in which Y=schizophrenism, x,=SES and x2 =IQ variable (Total IQ age 11, VIQ, PIQ, ZPIQ), the following regression equation was run: Y= bo+b^-t- b2x2 . The test of significance for R2 change when the IQ variable is added to the equation determines whether that variable contributes significant unique variance in the prediction of schizophrenism. Significant p-values for R2 change were cross-validated when x2 =totai IQ at age 11 (odd: F=10.18, p< 001; even: F=15.53, p< 001), and when x2 = PIQ (odd: F=11.74, p=.001; even: F=13.38, p<00l), leading to the conclusion that total IQ and PIQ do not act as single, dominant mediators of the relationship between SES and schizophrenism. The other two IQ 11 variables, VIQ and ZPIQ, revealed mixed findings for R2 change. In one subsample (odd), when VIQ=x2 ., F=2.4, and p=. 122, leading to the possible conclusion that VIQ mediates the relationship between SES and schizophrenism. However, in the cross-validation sample (even), F=9.99, p=.002. Similarly, in one subsample (even), when x2 =ZPIQ, F=3.71, p=.055; however, in the odd sample, the p- value was significant (F=8.05, p=.005). Thus, no IQ variable revealed a purely mediating role of the relationship between psychosocial stress and SSPD. Summary of Results The following summary is a synopsis of the findings. The more comprehensive report of results above lists significance levels, and statistical coefficients. Hypothesis la: High rates of SSPD at age 16 or at age 20-22 will relate to lower intellectual test scores at ages 3 and II: This hypothesis was partially supported: R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 77 a. ANOVAs revealed significant group effects when comparing subjects with high and low scores on the schizophrenism scale from the SAE (age 16). Subjects with high schizophrenism scores had lower scores on total IQ at age 11 (see Figure 2) and on performance IQ (see Figure 2). Effect sizes were in the medium to large range3 . b. Multiple regression analyses (cross-validation technique) revealed a strong relationship between total lower IQ at age II and higher schizophrenism at age 16 (Figure 3 a, 3 b). Additionally, as depicted in Figure 4a and 4b, lower verbal IQ significantly predicted higher schizophrenism. Finally (Figure 5a, 5b), lower performance IQ emerged as a significant predictor of higher schizophrenism. These significant results are summarized in Table I. Effect sizes were in the small to medium range. c. However, no significant relationships were revealed in which cognitive performance at age 3 played a role. d. Additionally, this finding was specific to the age 16 schizophrenism scale; SSPD measured at age 20-22 and anhedonia at age 16 did not show a significant relationship to IQ. Hypothesis lb: Males with higher scores on the SSPD scales will show greater deficits on tests of cognitive functioning than will females: This hypothesis was not supported 3 Categories of “small”, “medium” and “large” effect sizes from Cohen (1988) were employed to compare f (for ANOVAs) and r (for regressions). R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 78 by the findings; sex was not a significant moderator of the relationship between childhood cognitive functioning and later SSPD. All personality group by sex interactions in the ANOVAs were non-significant. Similarly, all regression equations with a sex*IQ interaction term showed non-significant R2 change when the interaction term was added to the model. Hypothesis lc: Subjects with higher levels of SSPD will show more performance than verbal intellectual functioning deficits at age 11. This prediction was partially supported. a. ANOVAs showed a trend such that the index group with high levels of schizophrenism had significantly lower scores on the ZPIQ residual score as compared to the control group. This result is illustrated in Figure 6a. Effect size was in the medium range. b. In the regression equation (Figure 6b, 6c), lower scores on the ZPIQ significantly predicted higher levels of schizophrenism. Effect size was in the small range. c. Again, SSPD at age 20-22 as measured by the PDE and a scale of anhedonia did not have a significant relationship to levels of intellectual ability. d. Sex effects were also explored on this model, but did not approach significance. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 7 9 Supplementary analyses: Taxon vs. Dimension a. Regression equations with significant results were further analyzed; specifically, those in which Y=schizophrenism, and xt=total IQ, VIQ, PIQ, and ZPIQ. b. When the schizophrenism index group was removed from the sample, and the regression equations re-run, the change in R2 was not statistically significant, (and the difference in R2 between regression equations without the index group and without the comparison group were also non-significant), lending support to the dimensional hypothesis. Hypotheses 2a fModel P vs. Hypothesis 2b (Model ID: 2a: Subjects with higher levels of SSPD will show lower scores on intellectual tests but will not have significantly different ratings on environmental stress compared to controls. Versus 2b: An additive interaction is predicted in which both lower scores on intellectual tests and higher ratings on environmental stress relate to higher levels of SSPD. [As mentioned previously, subjects with higher levels of schizophrenism at age 16 showed lower scores on all IQ tests at age 11: total IQ, VIQ, PIQ, and ZPIQ (residual score measuring VTQ-PIQ differential). The additive effect of psychosocial stress will be reported here on these variables.] a. Additive results for the equation, Y= b0 +b,xt+ b2 x2 , in which x, is the IQ variable (IQ at age 3, total IQ at age 11, VIQ and PIQ at age 11 , ZPIQ index); x2 is the R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 80 environmental stressor variable (FSI or SES), and Y the SSPD variable (SSPD age 20-22, anhedonia and schizophrenism at age 16) again revealed significant relationships only with the schizophrenism measure. b. The Family Stress Index (FSI) did not provide any significant additive effect for all IQ variables, thereby lending support to Model I. c. The SES index did not provide an additive effect in predicting schizophrenism when X [ = total IQ, VIQ, and PIQ, further supporting Model I. d. The SES index provided a significant additive effect for ZPIQ at age 11, such that lower SES and lower ZPIQ predicted higher rates of schizophrenism. This finding supports Model II (Hypothesis 2b). Additionally, this analysis had an effect size in the small to medium range. e. The Analysis of Covariance (ANCOVA) that was intended as a follow-up analysis of significant additive effects was conducted in order to determine whether this additive relationship was also evident in a categorical model. Results showed that when SES was entered as covariate, the group effect was non-significant, and effect size was in the small to medium range. Hypothesis 2c fModel HD vs. Hypothesis 2d fModel IV) 2c: Lower intelligence test scores will interact with higher environmental stress index scores to predict higher rates of SSPD. Versus R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 81 2d: Lower intelligence test scores will interact with lower environmental stress index scores to predict higher rates of SSPD. a. The regression equation Y = b0 + bjX, + b2x2 + b3x,x2 was entered and again, total R2 was significant for all equations in which Y=schizophrenism. b. When the FSI and each IQ variable were tested for interaction effects in predicting SSPD, the change in R2 was non-significant for age 3 IQ, total IQ at age 11, VIQ and PIQ. These results continue to support Model I. c. When the FSI and ZPIQ (standardized residual score quantifying differential between VIQ and PIQ) were tested for interaction effects in prediction SSPD, the change in R2 was significant, supporting either Model m or Model IV. d. Further analysis of this interaction revealed the relationship of ZPIQ to schizophrenism at different levels of family stress: At the lowest FSI score, ZPIQ has the strongest predictive value on later schizophrenism; at levels of high family stress, ZPIQ has virtually no predictive value, thus supporting Model IV (Figure 7). e. When SES and each IQ variable were tested for interaction effects in predicting later SSPD, all changes in R2 were non-significant, supporting Model I. Hypothesis 2e: IQ mediates the relationship between childhood psychosocial stress and SSPD; psychosocial stress accounts for any significant statistical relationship between IQ and SSPD. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 82 a. As described in the main body of the results section, three equations needed to be run in order to determine mediator status. The first was Y= b0 +b,x,, where IQ (mediator) is Y, and psychosocial stress is x, b. When FSI=xt, all R2 were non-significant. c. When SES3 1 *,, R2 was not significant for IQ at age 3, but was significant for all IQ variables at age 11 (total IQ, VIQ, PIQ, ZPIQ), such that lower SES predicted lower IQ. Only these IQ variables were used in the second equation testing mediation. d. In the second equation, Y= b0 +b,x,, the SSPD variable was regressed onto the IQ variable. These results were reported under Hypothesis la and Ic above, and significant results can be seen in Table I. Only results when Y=schizophrenism were significant. e. In the third equation, variables tested were limited to those in which Y=schizophrenism, x,=SES and x2 =IQ variable (Total IQ age 11, VIQ, PIQ, ZPIQ), and the following regression equation was run: Y= b0 +b,x,+ b2x2. Note that this is the opposite order of entry from Hypothesis 2b (Model II), testing additive effects. The test of significance for R2 change when the IQ variable is added to the equation determines whether that variable contributes significant unique variance in the prediction of schizophrenism. In short, if the R2 change is significant, IQ does not fully mediate the relationship, and if it is nonsignificant, IQ is a mediator (psychosocial stress accounts for the relationship between IQ and SSPD). f. R2 change was significant when x2= total IQ at age 11, and when x2 = PIQ, leading to the conclusion that total IQ and PIQ do not fully mediate the relationship of SES to schizophrenism. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 83 g. The other two IQ 11 variables, VIQ and ZPIQ, revealed mixed findings for R2 change, insofar as R2 change was non-significant only in one of the two cross-validation subsamples. Thus, VIQ and ZPIQ do not act as single, dominant mediators of the relationship between SES and SSPD, and Model V (Hypothesis 2e) is not supported. Discussion This section will (1) review the findings of the present study in terms of their fit with the hypotheses and models proposed in this paper; (2) synthesize the significance and implications of these findings in relation to (a) neurocognitive abnormalities in schizophrenia and schizotypy, (b) previous literature examining the relationship between SSPD/schizophrenia and IQ, ( c) the single-gene and polygenic etiological models of schizophrenia, and (d) the taxonic and dimensional theories of the schizophrenia-spectrum; (3) outline possible interpretations of the lack of findings for anhedonia in the present study; (4) account for the unique additive and interactional effects for the verbal- performance discrepancy score (ZPIQ) and psychosocial stressors; (5) discuss the lack of findings at age 3 and the lack of hypothesized sex effects; (6) examine the impact of psychosocial stressors on the development of SSPD, (7) outline methodological issues in the present study such as generalizability and measurement issues, (9) discuss limitations of the study, and ( 10) address directions for future research. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 84 Review o f Findings Hypothesis I— Main Effects: The prediction that high rates of SSPD would relate to lower intellectual test scores in childhood was well-supported by multiple regression analyses; a significant relationship emerged between lower IQ scores at age 11 (total IQ, Verbal IQ, Performance IQ) and higher levels of schizophrenism at age 16. The ANOVAs also showed a significant group effect for total IQ and PIQ at age 11, but not for VIQ. Notably, the effect sizes4 for the ANOVAs were larger (medium to large range) for all three variables (even VIQ) than for the regressions (small to medium range). No significant relationships were revealed in which cognitive performance at age 3 played a role in SSPD development. Additionally, the significant findings were specific to the age 16 schizophrenism scale; SSPD measured at age 20-22 and anhedonia at age 16 did not show a significant relationship to IQ. Hypothesis lb: The prediction that males with higher scores on the SSPD scales would show greater deficits on tests of cognitive functioning than females was not supported by the findings; sex was not a significant moderator of the relationship between childhood cognitive functioning and later SSPD. Hypothesis lc: The prediction that subjects with higher levels of SSPD would show more performance than verbal intellectual functioning deficits at age 11 (as measured by the 4 Categories of “small”, “medium” and “large” effect sizes from Cohen (1988) were employed to compare f (for ANOVAs) and r (for regressions). R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 85 standardized residual score, ZPIQ) was well supported by the data. Both the ANOVAs and multiple regression analyses showed that subjects with higher levels of schizophrenism at age 16 had more performance versus verbal IQ deficits at age 11. Again, the effect size for the ANOVA (medium) was greater than for the regression equation (small). SSPD as measured by the PDE at age 20-22 and the anhedonia scale at age 16 did not show a significant relationship to levels of this discrepancy score. Additionally, no significant sex effects emerged in the relationship between schizophrenism and ZPIQ. Supplementary analyses: Dimension vs. Taxon Supplementary analyses were performed to further examining the taxon/dimension issue of schizotypy. Specifically, the significant regression equations in which schizophrenism was regressed on the IQ variables at age 11 were re-run without the index group from the ANOVA; no significant change in R2 was evident between the full and supplementary equations. A taxonic hypothesis for schizophrenism would have predicted that when the taxon (the index group from the ANOVA) is removed, R2 would be significantly reduced. However, as this did not occur, the dimensional view of SSPD is supported. Hypothesis H, Comparison of biosocial models The five models examined in Hypothesis 2 explored the effect of environment (as measured by the FSI and the SES scale) on the relationship between cognitive deficits and SSPD. Tests of the second hypothesis generally favored Model I, suggesting that in most R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 8 6 cases, the psychosocial stressors measured in the present study do not provide additive, moderating, or mediating effects for the relationship between cognitive deficits and later SSPD. However, results revealed two notable exceptions: (1) Model II was supported when SES was shown to have a significant additive effect with the ZPIQ index in the prediction of schizophrenism, and (2) Model IV was supported by findings that revealed an interaction effect such that at the lowest Family Stress Index scores, ZPIQ had the strongest predictive value on later schizophrenism, and at the highest levels of family stress, ZPIQ had virtually no predictive value for subsequent schizophrenism. Again, effect sizes for these regression analyses were small, and when the additive effect in Model II was tested in an ANCOVA, the effect size was also small. Unfortunately, the interaction effect could not be tested in a group comparison model because of small cell sizes. It is important to note that two of the models were not at all supported by the data: (1) Tests of Model HI revealed that psychosocial stress as measured by the present study was not a moderator of the relationship between intellectual and personality functioning such that higher stress and lower IQ interacted to predict higher levels of SSPD, and (2) examination of Model V demonstrated that the relationship between IQ and SSPD is not spurious, and cannot be entirely accounted for by the relationship of IQ and SSPD to psychosocial stress. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 87 Significance o f Findings IQ, the schizophrenia spectrum, and neurodevelopment. To the author’s knowledge, this is the first study to report a significant relationship between IQ deficits and later SSPD in a longitudinal study of a general population. This finding speaks to a significant question considered in the literature: are the IQ deficits found in schizophrenics neurodevelopmental or neurodegenerative in origin? The concept of neurodegeneration in schizophrenia dates back to the original formulation by Kraeplin of “dementia praecox” (Kraeplin, 1971/1919). More recent research has demonstrated that a significant deterioration of functioning occurs in schizophrenia, primarily in the first five years of the illness (e.g. Abrahamson, 1983; Kolakowska et al., 1985), and that this functional decline is, in part, a neurodegenerative process that can manifest itself in neuropsychological deficits (Payne, 1961; Hamlin, 1969), including those measured by IQ tests (Bilder et al., 1992; Bilder et al., 1991). It is now well-accepted that a neurodegenerative process occurs after the first psychotic break. However, it remains unclear whether a neurodevelopmental process is at play before onset of the illness. If IQ deficits signal a neurodevelopmental process which occurs premorbidly, and they may be indicative of the integrative neural defect hypothesized by Meehl (1962, 1990). To achieve status as a biological marker, a number of criteria must be filled, among which are the following: first, the putative indicator must predate the onset of the illness; second, it must be demonstrated to occur in genetically related forms of the disease (e.g. SSPD); and third, it must be stable over time (Garver, 1987). The present study attempted to examine intellectual deficits as a potential biological marker, as part of a R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 88 neurodevelopmental process underlying the schizophrenic spectrum. Findings of significant relationships between IQ measures at age 11 and personality at age 16 do lend support to the theory of a neurodevelopmental origin for IQ deficits in the development of schizophrenia. It is important to note, however, that other studies examining this possible neurodevelopmental process have shown mixed results. Lending support to the hypothesis that IQ is a biological marker, a number of studies have demonstrated that IQ deficits predate the onset of psychosis in schizophrenics and are stable across time (Bower et al., 1960; Offord, 1974; Watt & Lubensky, 1976; Albee et al., 1964; Lane & Albee, 1968; Offord & Cross, 1971; Erlenmeyer-Kimling, et al., 1991; Bilder et al., 1992). Interestingly, these studies used control groups consisting of non-siblings. When preschizophrenics are compared to their siblings, findings are less conclusive. Whereas some studies have found significant differences between preschizophrenic and sibling IQ scores (Lane & Albee, 1964; Lane & Albee, 1965; Offord & Cross, 1971), two studies (Watt & Lubensky, 1976; Offord, 1974) found no significant differences at all. This led Offord (1974) to hypothesize that IQ and the diathesis to schizophrenia are independently inherited traits, and that IQ acts as a moderator of the development of schizophrenia. Based on his study, Offord (1974) proposed that a high IQ can act as a protective factor against becoming schizophrenic, whereas a low IQ does not offer this protection. In a model analogous to Model IV in the present study, he surmised that if a high-IQ family contained a preschizophrenic child, then the genetic vulnerability to schizophrenia must be severe, and if the predisposition is weak, overt expression of schizophrenia will occur only in Iow-IQ families. Similarly to Offord (1974), but in R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 8 9 contrast with the present study, other researchers also have also viewed IQ as a possible moderator of the relationship between a putative marker and schizotypal traits in childhood (e.g. Bergman & Walker, 1995). However, in the present study, IQ has been proposed as the possible biological marker, and not the moderator. It is possible that the lack of findings at age 3 as well as the lack of a relationship between IQ and, respectively, anhedonia and SSPD at age 20-22, may be explained if IQ is itself a moderator or potentiator of the development of schizophrenia, and not a putative indicator. There is, however, another possible interpretation of the mixed findings that lends support to the hypothesis of IQ deficit as a biological marker. If lower IQ is in fact a manifestation of the integrative neural defect, it would be expected to be found in family members of schizophrenics, just as higher rates of schizotypal personality disorder have been found in families of schizophrenics (Kendler, et al., 1995). Thus, a preschizophrenic’s sibling would be expected to have lower than normal IQ. However, because siblings share, on average, 50% of their genes, studies with small sample sizes may sometimes find significant differences in IQ levels between preschizophrenics and their siblings. It is important to note that the neurodevelopmental and neurodegenerative models are not mutually exclusive. For instance, Bilder et al. (1992) tested both models by comparing controls, first episode schizophrenics and chronic schizophrenics on IQ measures. Findings demonstrated poorer PIQs for chronic than for first-episode schizophrenics, supporting the neurodegenerative model; both groups of schizophrenics showed lower IQS (on all IQ measures) than controls, supporting the neurodevelopmental model. Yet findings from this study (Bilder, et al., 1992) do not eliminate the possibility R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 9 0 that IQ is part of a neurodevelopmental process that is independent from but potentiates the process leading to schizophrenia. Similarly, while findings from the present study show a clear relationship between IQ deficits at age 11 and later schizophrenism, the possibility remains that IQ is not a result of the integrative neural defect, but is rather a potentiator of the neurodevelopmental process such that lower IQ is a stressor that interacts or adds to the genetic diathesis. Future studies examining the interaction of IQ and other putative markers, such as psychophysiological variables (e.g. electrodermal orienting response) are needed to test this hypothesis. IQ and schizotypy. Examining IQ deficits as a possible biological marker for schizophrenia also necessitates an exploration of its presence in genetically related forms of the disorder. Several studies have shown schizotypy and other SSPDs (e.g. schizoid, avoidant) to have increased prevalence in the relatives of schizophrenics (e.g. Kendler et al., 1995) as well as increased rates of schizophrenia in relatives of schizoptypals (Siever et al., 1991). Consequently, the present study extends research by examining whether IQ deficits predict SSPD. As discussed previously, the present study’s findings lend support to the relationship between lower IQ and schizotypy, at least when SSPD is measured as schizophrenism, and IQ is measured at age 11. A number of other researchers have also studied intellectual functioning in subjects with SPD or SSPD, with mixed results (Asamow and Ben -Meir, 1988; Fenton and McGlashan, 1989; Poreh et al., 1995; Trestman et al., 1995; Lencz, et al., 1993; Kelley & Coursey, 1992; Raine, 1987; Bergman & Walker, 1995). R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. Findings from two studies using SSPD patient samples (Asamow & Ben-Meir, 1988; Fenton & McGlashan, 1989) found lower IQ to be significantly related to schizophrenia-spectrum disorders, thus lending support to the hypothesis of IQ deficit as a putative indicator. However, in another patient sample, Trestman et al. (1995) did not find any significant IQ differences when comparing schizotypals, other personality disordered patients and controls. However, it is important to note that conclusions about the results of these studies must be made carefully. First, in the Asamow & Ben-Meir (1988) study, hospitalized children with schizophrenia and SPD were compared to children with depression and dysthymia. The early onset of the illness in the subjects may indicate: (1) a more severe genetic vulnerability to schizophrenia, in which case the significant finding may signal the IQ deficit as a biological marker, or, alternately, (2) extreme environmental stress (e.g., birth complications, psychosocial stress, maternal exposure to influenza) that either moderated or provided an additive effect to the genetic diathesis, or (3) both a greater vulnerability and higher levels of environmental stress. Additionally, some methodological issues arise: (1) given the comparison of the SSD group to a depressed sample, it is important to note that these disorders may be related (Kendler et al., 1995), (2) differences between groups may not persist after discharge from the hospital, and (3) IQ tests were given during hospitalization, where because of psychosis, patients did not give a good indication of their premorbid abilities. However, the examination of children in this study is notable in light of the present study’s focus on childhood predictors of late adolescent SSPD. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 9 2 In the Fenton & McGlashan (1989) 15-year longitudinal study of adults hospitalized for SPD, schizotypals who became schizophrenic showed lower IQ than schizotypals who did not decompensate. These findings suggest several possible interpretations that also fit the present study: (1) lower IQ is an indicator of the genetic diathesis, and the patients who became schizophrenic had a higher genetic loading for the illness, or (2) IQ moderates or provides an additive effect with the diathesis, such that schizotypals with lower IQS are more likely to become schizophrenic. Thus, as in the present study, it is clear that in attempting to test IQ deficit as a biological marker it is difficult to distinguish between whether the IQ variable plays the role of indicator or potentiator. In contrast to these studies (Asamow & Ben-Meir, 1988; Fenton & McGlashan, 1989; Trestman et al., 1995), the present study does not use a small patient sample to explore the relationship between IQ and schizophrenia; instead, analysis of a large general population sample allows for an examination of SSPD as a dimensional variable as well as a diagnostic category. However, a number of other researchers also used non-patient samples to examine IQ deficits in schizotypy (Poreh et al., 1995; Lencz et al., 1993; Kelley & Coursey, 1992) and did not find differences in IQ sub-test levels between undergraduates with schizotypal symptoms and controls. It is important to note that these three studies (Lencz, et al., 1993; Kelley & Coursey, 1992; Poreh et al., 1995) utilized undergraduate samples. Schizotypals who are attending a university may score higher on IQ tests as a reflection of their higher cognitive functioning, thus restricting the range of scores. In order to overcome this possible R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 93 restriction of range, the present study examines an unselected sample. Additionally, none of these studies of SSPD and IQ have used a biosocial longitudinal approach starting from childhood so as to more accurately examine the development of the disorder; the present study uses such a longitudinal design. This may account for the significant relationship between low IQ and schizophrenism in the present study. The impact of a biosocial study is evidenced by findings from two studies that did use a biosocial model to explore the relationship between IQ and SSPD (Raine, 1987; Bergman & Walker, 1995). These will be examined at length in a later section (The Biosocial Approach). In short, Raine (1987) found that schizotypal criminals from “intact” homes showed lower VIQs than schizotypal criminals from “broken” homes. This finding was the basis for Model IV, predicting that lower psychosocial stress will moderate the relationship between lower IQ and SSPD. In the present study, this prediction was realized when an interaction effect emerged between lower family stress and lower VIQ- PIQ discrepancy score (ZPIQ) in predicting later schizophrenism. Findings in the Bergman & Walker (1995) study, however, do not parallel findings in the present study. They seem to have largely conceptualized IQ as Offord (1974) did; IQ was examined as a possible mediator or moderator of the relationship between parental maltreatment or parental psychiatric status and childhood personality. Findings revealed that IQ moderated the relationship between parental maltreatment and schizoid behavior, but did not play a role in the relationship between parental psychiatric status and schizoid behavior. Thus, in contrast to the findings of the present study and of Raine (1987), Bergman & Walker (1995) found an interaction effect between higher family stress and lower IQ in predicting R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 94 SSPD in children. However, the difference in findings may be explained by (1) differences in the measure of family stress, and (2) the measurement of personality in childhood as opposed to adulthood. In sum, studies on IQ and SSPD have led to inconclusive results regarding the status of IQ deficits as a biological marker. The present study provides some evidence for IQ deficits as a putative indicator of “positive” symptomatology measured by the schizophrenism scale, but conclusions must be made with caution, as the possibility remains that IQ is a genetic potentiator in the development of SSPD. Etiological and Nosological models of SSDs. In order to understand the findings of the present study, they must be considered in the context of theories about vulnerability to schizophrenia, specifically, the single-gene and multifactorial polygenic models of the disease. These theories lead to different predictions about the relationship of cognitive ability and psychosocial stress to the development of schizophrenia and SSPD. Genetic theories Meehl’s genetic hypothesis of schizotaxia, schizotypy, and schizophrenia (Meehl 1962, 1989, 1990) is a diathesis-stress model asserting that a single major locus (SML), or the “schizogene,” constitutes the genotype for schizophrenia, termed “schizotaxia.” Under “all real reinforcement schedules” (Meehl, 1962), schizotaxia is manifested phenotypically as schizotypy; that is, no psychosocial stressors are needed to potentiate the expression of the phenotype (Figure la). Thus, schizotypy is the expression of the diathesis for R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 95 schizophrenia and is hypothesized to result from the “integrative neural defect,” a flaw in the workings of the central nervous system (CNS), that can show itself in abnormalities of neurocognitive, neuromotor, and personality functioning. Schizophrenia, the model posits, develops only when polygenic and environmental potentiators (e.g. birth complications, psychosocial stressors) interact with this diathesis. Thus, Meehl’s diathesis-stress model would predict that individuals with SSPD would show signs of the integrative neural defect (e.g. cognitive deficits), but would not necessarily experience the environmental stressors (e.g. high family stress, low SES), or have the polygenic potentiators necessary for the development of schizophrenia. Thus, Model I proposes that subjects with SSPD would not show significantly higher or lower psychosocial stress. Note that this prediction would be different for older subjects with SSPD; schizotypals beyond the age of risk for schizophrenia might show lower levels of environmental stress as compared to schizophrenics. In fact, a study by Schulsinger et al. (1984) and a follow-up study by Cannon et al. (1994) demonstrated that schizotypals do seem to lack certain stressors as compared to schizophrenics. Specifically, in the Copenhagen High Risk Study, schizotypals had the same cortical abnormalities as schizophrenics, but did not show the ventricular enlargement present in schizophrenics. It was hypothesized that while the schizotypal group shared a genetic liability with the schizophrenics, they did not suffer the same environmental stressor as the schizophrenics, and that their lack of enlarged ventricles served as a protective factor against development of the full disease course. Unfortunately, the present study is not designed to address this hypothesis due to the current young age of the subjects. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 9 6 The polygenic model of schizophrenia suggests that schizotypy and schizophrenia are not just two manifestations of a single schizotaxic gene, but rather are a result of many genes causing neurodevelopmental abnormalities in different parts of the brain (Gottesman & Shields, 1982; Risch, 1990; Reiss et al., 1991; Fowles, 1992). These abnormalities combine with environmental stress, and at a certain threshold, schizophrenia occurs. SSPD falls below this threshold, but may manifest lower levels of the genetic diathesis and/or the environmental stress. Findings in present study: Genetic models In the present study, tests of the first hypothesis show a relationship between cognitive deficits and later schizophrenism, suggesting that intellectual deficits in childhood may in fact be a putative indicator, and an expression of the integrative neural defect proposed by Meehl (1962, 1989). However, these findings are also supportive of the polygenic model which also expects neurocognitive expressions of the multifactorial, polygenically created diathesis to schizophrenia. It is important to note that because the present study did not use several genetic variables with proven distinct etiologies, it is difficult to tease out if the manifestation of the genetic liability is determined by one schizogene or by several genes. The majority of the results from the examination of the biosocial effects seem to support Meehl’s diathesis-stress model of schizophrenia, which maintains that the schizotaxic gene leads to schizophrenia only under the influence of a certain amount of polygenic potentiators and environmental stressors; without these potentiators, the gene R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 9 7 manifests as personality dysfunctions less severe than those found in schizophrenia — schizophrenia-spectrum personality disorders (Meehl, 1962; 1990). Thus, in Meehl's model (see Figure la), individuals with SSPD would show the neuropathology (e.g. cognitive deficits) that reflect the single defective gene, but not experience the environmental stressors (e.g. high family stress, low SES); a majority of findings in the present study reflected a similar pattern. However, Meehl’s model cannot account for the additive and interaction effects of psychosocial stressors with the verbal/performance IQ discrepancy score (ZPIQ) in predicting later schizophrenism. In fact, these findings are at odds with Meehl’s theory which predicts that schizotypals would not experience increased levels of these psychosocial stressors. On the other hand, the multifactorial model takes into account the additive effects found in the present study, in which SES independently predicted schizophrenism when examined with ZPIQ partialled. This model predicts that certain environmental stressors will combine with the genetic diathesis, and that schizophrenia will develop only when a certain critical level is reached. Thus, an individual does not have to be schizophrenic to have manifestations of the genetic diathesis such as cognitive deficits, and also be subject to psychosocial stress; instead, that individual can fall somewhere along the illness continuum. The interaction effect found in the present study, in which lower family stress interacted with lower ZPIQ to predict schizophrenism can also be explained by the multifactorial polygenic model. Because different levels of environmental stress and R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 98 genetic vulnerability may combine and appear along the continuum, individuals with high levels of environmental stress will have a stronger push toward psychopathology, independent of genetic risk. On the other hand, individuals from low stress environments would need a greater genetic “hit” to manifest psychopathology such as SSPD. Thus, in the present study, the schizotypal subjects with a stronger manifestation of the genetic diathesis (greater cognitive deficits), came from low stress homes; it is possible that they had more of a genetic vulnerability to SSPD than the subjects from the high family stress environment, who reached the critical level necessary for schizotypy by means of higher psychosocial stress. These findings are similar to those of Raine (1987), in which schizotypals from “intact homes” showed lower verbal IQS than schizotypals from “broken” homes. The findings will be discussed in relation to the Raine (1987) study in greater detail in a later section. In sum, the present study provides mixed support for Model I and Meehl’s single gene hypothesis (1962, 1990). Whereas the general lack of interaction and additive effects by the psychosocial stressors do seem to support the single-gene model, the findings of an additive effect of SES and ZPIQ as well as the interaction effect between family stress and ZPIQ imply that psychosocial stress does at times act as a potentiator, thus supporting the multifactorial, polygenic approach to the development of SSPD. Findings in present study: Taxon vs. dimension As stated previously, the taxonic view of schizotypy would predict that if SSPD is a true taxon, a group comparison (such as ANOVA) would best capture the relationship R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 99 between the putative marker and SSPD, whereas a regression or correlational model would introduce error variance or noise in the dependent variable (SSPD). Alternately, if schizotypy is a true dimension, an ANOVA would restrict the range and would omit the full variance of the variable (Gangestad & Snyder, 1985). In the present study, both ANOVAs and multiple regression analyses were carried out on the data. Interestingly, significance of the findings was largely the same for both sets of analyses, with the exception of a non-significant finding in the ANOVA on VIQ; otherwise, as mentioned previously, cognitive deficits at age 11 related to later schizophrenism. However, when effect sizes were examined, it became clear that the effect sizes of the ANOVAs were in the medium to large range, whereas the effect sizes of the multiple regression equations were in the small to medium range. It is important to note that, in the case of VIQ, the ANOVA was not significant, but also had a larger effect size than the significant regression equation; the regression equations had higher significance levels due to the large sample sizes in the analyses. Thus, the ANOVAs seemed to more clearly show the relationship between the putative indicator and SSPD, supporting the hypothesis that schizophrenism is a taxon that can be best measured as a discrete category. Using the single gene model, we may surmise that the IQ variables serve as the putative indicators of a taxon. On the other hand, with the polygenic model, we might speculate that at least two genes are involved in the development of SSPD, one expressed taxonically through total IQ, PIQ, and ZPIQ. Other variables, such as SES, may serve as potentiators of the genetic vulnerability (see Biosocial section below). R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 100 However, supplementary analyses revealed a different picture, this one supporting a dimensional view of SSPD. When the significant regression equations in which schizophrenism was regressed on the IQ variables at age 11 were re-run without the index group from the ANOVA, no significant change in R2 was evident between the full and supplementary equations. A taxonic hypothesis for schizophrenism would have predicted that when the taxon (the index group from the ANOVA) was removed, R2 would be significantly reduced. However, as this did not occur, the dimensional view of SSPD is supported by the findings. This fully dimensional model is akin to that of Eysenck (1960), who proposes that mental illness is entirely dimensional, and who does not take into account any discontinuities that my occur in symptoms or traits (Claridge & Beech, 1995). Interestingly, findings from the tests of the additive effect of SES lend support to this model. Specifically, regression analyses found that lower SES provided a unique, additive effect to ZPIQ in predicting later schizophrenism (small to medium effect size). When the index group was removed, the change in R2 was non-significant, further supporting a dimensional hypothesis. Additionally, when an ANCOVA was performed, SES provided an additive effect with a similar effect size (small to medium). Thus, it seems that SES is able to predict schizotypy both when it is viewed as taxon and when it is viewed as a dimension; in accordance with the model proposed above, SES may account for schizotypy as it occurs as a dimension in the “normal” population. It is also possible that the findings in the present study are indicative of a different, more complicated model of schizotypy (see Figure 8) theorized by Claridge and Beech (1995): perhaps schizotypy is dimensional when in the normal range, but at a certain R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 101 critical level, becomes taxonic, and can be diagnosed as schizotypal personality disorder. At this point, schizotypy emerges as a continuum between SPD and schizophrenia. Perhaps most of the subjects in the present study, even those in the index group, do not show taxonic levels of schizotypy, and results reflect the dimensionality of the “normal range” of schizotypy. FIGURE 8 SCHIZOPHRENIA SCHIZO TYPAL P.D. GENE(S) : ► SCHIZOTYPY (PERSONALITY TRAITS) Figure 8. “Fully dimensional” model of SSPD as described by Claridge & Beech (1995). This model suggests that schizotypy is dimensional in the general population (lower horizontal line), and that there might be many genetic and environmental predictors of this personality dimension in the normal range. Additionally, the model posits that the presence of a specific genetic diathesis (diagonal line) results in a qualitative change to a higher level of SSPD, which may be expressed variably from schizotypal personality to schizophrenia (curved line). The taxon/dimension issue also speaks to another discrepancy in the findings. Specifically, the present study did not find any significant relationships between cognitive Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 102 deficits and SSPD on the PDE at age 20-22 (Loranger, 1988). Although the semistructured interview score was used as a dimensional index in the regression equation, it is not primarily used in that fashion. The scale was created with the Diagnostic and Statistical Manual III and Ill-Revised (American Psychological Association, 1980, 1987) in mind, and is concerned only with these diagnostic criteria. On the other hand, the other personality measure used in the study, the SAE (Venables et al., 1990), was constructed as a dimensional index, and, if schizotypy is itself dimensional, the SAE may measure SSPD more accurately. However, this conclusion conflicts with the findings in the current study in which schizotypy seems best viewed as a taxon. Another interpretation for the lack of findings on the PDE is that the PDE, based on the DSM (American Psychiatric Association, 1987; 1980), taps into a wide variety of SSPD symptoms, whereas the schizophrenism and anhedonia scales each focus on specific types of symptomatology. Thus, it is possible that the relationship between lower IQ and later SSPD is manifested only in “positive” symptomatology assessed by the schizophrenism scale at age 16, and that the PDE is tapping into a set of traits too heterogeneous to show the significant relationship between the “positive” traits it measures and lower IQ. In summary, findings provided support for a fully dimensional hypothesis for SSPD, similar to that proposed by Eysenck (1960). However, findings may also implicate a more complicated model of schizotypy and schizophrenia, based on Claridge & Beech’s (1995) integrated model. The results from the present study suggest a more comprehensive analysis of the SSPD measures; a direction for further research may Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 103 include MAX-COV-HIT-MAX analyses of the schizoprhenism questionnaire, as described by Golden & Meehl (1979) and Lenzenweger & Korfine (1992). Positive/negative symptomatology: Anhedonia and Schizophrenism at age 16 The non-significant relationship between cognitive deficits and anhedonia also merits discussion. Anhedonia was originally thought to be a central, fundamental, symptom of schizotypy (Meehl, 1962), a manifestation of the integrative neural defect. This hypothesis led the Chapmans and their colleagues to create a questionnaire that could be used in studies of schizophrenia and SSPD (Chapman et al., 1976). Many studies have shown significant relationships between anhedonia and cognitive deficits, especially in psychophysiological and attentional processing (e.g. Simons, 1981; Bernstein & Reidel, 1987; Raine et al., in press), backward-masking tasks (e.g., Balough & Merritt, 1985), and eye tracking (e.g. Kendler et al., 1991). However, because the mechanisms underlying backward masking are theoretically unclear (Silverstein, et al., 1992), and the deficit has been demonstrated to be reversible in schizophrenics (Sacuzzo & BrafF, 1981), it may not be a good indicator of the genetic vulnerability to schizophrenia. In recent years, a number of investigators, including Meehl (1987, 1989), have changed their conception of anhedonia as a manifestation of the genetic vulnerability to schizophrenia. This change of theory is due to findings of anhedonia in other psychopathologies such as depression. Meehl’s current theory of “hedonic capacity” (Meehl, 1987; Meehl, 1989) proposes that anhedonia does not qualify as a pathological taxon, but rather is a very low level of a “normal range individual differences variable” Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 104 termed “hedonic capacity.” (Meehl, 1987, p. 48). Meehl posits that hedonic capacity is polygenically determined, and distinct from the schizotaxic gene. In this theory, hedonic impairment is a polygenic potentiator, increasing the probability that a schizotypal person will develop schizophrenia. Consequently, anhedonia would not necessarily show a relationship to markers of the integrative neural defect. Research has substantiated this claim by demonstrating that unlike schizophrenics, anhedonics do not show abnormal performance on a variety of neurocognitive measures, including perceptual organization tasks (Silverstein et al., 1992), and visual switching attention reaction time tasks (Wilkins & Venables, 1992), and the Continuous Performance Task (Venables, 1990). Additionally, in a 10-year longitudinal study (Chapman et al., 1994) on the predictive utility of psychosis of the Chapman scales, the Physical Anhedonia scale was found not to be an effective predictor of psychosis proneness. It is important to note that none of the Chapman scales specifically predict schizophrenic psychosis (Chapman et al., 1994). However, the Chapman research group do suggest the use of their Perceptual Aberration and Magical Ideation scales as a good predictor of general psychosis proneness, including both schizophrenic and mood related psychoses (Allen et al., 1987; Chapman et al., 1994); these scales measure traits similar to those measured by the schizophrenism scale of the SAE utilized in the present study. Based on these findings, cognitive deficits as measured by the present study would not be expected to relate to later anhedonia, as anhedonia is not necessarily a manifestation of the genetic vulnerability to schizophrenia. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 105 ZPIQ: The verbal/performance 10 discrepancy score The ZPIQ at age 11 is a standardized residual created when Performance IQ (PIQ) was regressed on Verbal IQ (VIQ). Hence, ZPIQ is the amount of PIQ “left over,” when VIQ is partialled. As was mentioned previously, based on the literature (e.g., Parker & Davidson, 1963), it was predicted that schizotypals would show more PIQ versus VIQ deficits than normals; this was supported by ANOVAs and regression analyses. Additionally, ZPIQ was the only cognitive variable implicated in an additive model with SES, as well as in an interaction with the FSI in predicting later schizophrenism. Prior to a discussion regarding these findings, a number of issues regarding the structure of this index need to be explicated. Several methodological issues emerge under close scrutiny of the VIQ and PIQ measures. First, neither subscale includes all the WISC-R items necessary to provide a total IQ score, and hence neither may tap into the full constellation of verbal and performance functions. Second, only two sub-tests form the VIQ index, whereas four form the PIQ index. This led to increased variance in the VIQ index, and makes it a less stable measure. Third, the subtests in the VIQ index are “Similarities” and “Digit Span”; these are not the usual subtests used as substitutes for the total battery (i.e. vocabulary and comprehension). However, as the VIQ and PIQ subscales were given to the entire sample, and their relative performance analyzed, relationships between performance on these tasks and later SSPD can still be assessed, albeit with caution. Fourth, the verbal and performance tests probably involve both verbal and non-verbal processes (Lawson & Inglis, 1983; Benton, 1962). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 106 It is also important to view the findings of ZPIQ in terms of findings of abnormal laterality in schizophrenia and schizotypy. Whereas VIQ taps language-mediated cognitive processes, and can be loosely considered a measure of left hemisphere (LH) functioning, PIQ measures visuo-spatial and visuomotor abilities, and can be broadly categorized as an index of right hemisphere (RH) functioning; ZPIQ thereby provides a crude measure of the relative functioning of the two cerebral hemispheres. Thus, in the present study, findings of significantly reduced PIQs in future schizotypals may indicate an abnormality in hemispheric functioning. In fact, abnormal development and functioning of lateralized brain structures has been characterized as a possible manifestation of the genetic vulnerability to schizophrenia (Bilder & Degreef, 1991), and it is believed they may reflect abnormalities in the genes controlling normal neuronal migration (Cannon et al., 1989; Crow et al., 1989; Weinberger, 1987; Bilder et al., 1994). Although studies that have examined left/right cerebral hemisphere differences in schizophrenia have shown mixed results (Bilder et al., 1994; Hoff et al., 1992), studies at the neuroanatomical level (Bilder & Degreef, 1991; Crow et al., 1989), and at the functional level (Flor Henry, 1987) tend to indicate that schizophrenics show a reduction in asymmetries found in normal subjects. The dominant theory for brain asymmetry in schizophrenia is that of LH dysfunction. This functional deficit in the LH has been attributed by various theorists either to LH under-activity or to LH over-activation (Flor-Henry, 1976; Venables, 1977). Venables (1983) proposed that the LH has a prime level of activation, above or below which deficits occur. Venables adds that LH over-activation may be associated with an early impairment in the right Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 107 hemisphere (RH). This early RH deficit may be offset by the LH having to process information that would usually be treated by both hemispheres. In the present study, lower ZPIQ for subjects showing high levels of schizophrenism may indicate an early RH deficit which leads to a diminished PIQ, and also leads to LH over-activation manifesting perhaps in schizotypal symptoms. However, this is speculative, especially in light of other theories of cerebral hemispheric abnormality in schizophrenia. Based on a study of skin conductance in which Gruzelier (1981) found some patients showing RH advantages (i.e., stronger responding), and others showing LH advantages, led Gruzelier & Manchanda (1982) to hypothesize that these conflicting results reflected heterogeneity in the independent variable; that is, that different types of schizophrenia might be accounting for the different types of asymmetries. Consequently, a relationship between symptomatology and hemispheric asymmetry was proposed: positive symptom schizophrenia is related to over-activation of the LH and under-activation of the RH, whereas negative symptom patients show under-activation of the LH, and possible RH superiority (Gruzelier, 1984; Levin et al., 1989). Interestingly, findings from the present study fit this model as well; schizophrenism, or positive symptomatology is related to underactivation of the RH as evidenced by lower PIQs. Findings in the literature on laterality in schizotypy have also produced mixed results. A number of these studies showed reduced or reversed asymmetries for subjects with higher levels of cognitive-perceptual (“positive”) symptoms of schizotypal personality (Claridge & Broks, 1984; Rawlings & Claridge, 1984; Broks et al., 1984; Broks, 1984; Rawlings & Borge, 1987; Benishay, 1993; Raine & Manders, 1988). In fact, several of Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 108 these studies found improved performance in non-dominant hemispheric functioning (cf a review, Lencz et al., 1995), showing less asymmetry, especially for male schizotypals. One study (Gruzelier, 1995) stands out in sharp contrast to the studies mentioned above. Gruzelier (1995) found no deviations from the normal asymmetry on the Warrington Recognition Memory Test for subjects who scored high on the cognitive- perceptual factor. Instead, subjects in the "withdrawn" group ("negative" symptoms) showed an RH advantage on the test indicating relatively greater RH activation, whereas subjects in the "active" group ("disorganized" symptoms) showed an LH advantage. In general, the present study’s findings lend some support to the LH overactivation/ RH underactivation hypothesis of schizophrenia (Gruzelier, 1984), as positive symptomatology is associated with a lower PIQ. However, the present study does not show similar findings to the studies of schizotypals regarding reductions in the usual cerebral asymmetries (e.g., Claridge & Broks, 1984); instead, higher schizophrenism levels are associated with more asymmetrical IQ performance. The discrepancy cannot be accounted for in difference in symptom types, as the present study and several of the other studies of schizotypy used measures of “positive” symptoms. However, the difference may be explained by the measures of asymmetry used for each study; for instance, Claridge & Broks (1984) use a very specific laterality task (dichotic listening), whereas in the present study, the IQ measures are more inclusive of both RH and LH processing. It is possible that the finding in the present study may best be accounted for by an abnormality in interhemispheric transfer (IHT); schizotypals may have difficulty in tasks that utilize flexible interactions between hemispheres, but may perform well on tasks that are limited Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 109 to the processing of mainly one hemisphere. Thus, it is possible that findings of lower ZPIQ in the present study may actually indicate a defect in the corpus callosum. With this in mind, the following question arises: Why was ZPIQ the only cognitive measure to be involved in an additive and interaction effect with psychosocial stressors? In the additive model (see Figure 9a), ZPIQ and SES independently predicted schizophrenism. Contrastingly, when SES was added into the regression equations with the other age 11 IQ measures, SES lost its unique predictive power; its influence on schizophrenism was accounted for by its higher correlation with total IQ, VIQ and PIQ than with ZPIQ (Figure 9b). Two possibilities stem from this finding: First, it is possible that ZPIQ is less related than the other IQ measures to influences of SES such as birth complications and poorer education, and is more “pure” as a genetic vulnerability marker; in other words, ZPIQ may be a better indicator of the genetic vulnerability to schizophrenia than the other IQ variables. This is supported by the tests of SES as a mediator. Second, it could be hypothesized that ZPIQ is reflective of a separate potentiator of SSPD risk that is more nearly orthogonal to the potentiating role of SES. In sum, the strong findings relating to ZPIQ suggest that the measure may be a clearer manifestation of the genetic diathesis than the other IQ variables utilized in the study. Additionally, ZPIQ may be a manifestation of an IHT abnormality, implying perhaps that a defect in IHT is the biological marker that should be investigated. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 110 FIGURE 9a Figure 9. The Venn diagrams represent variance of schizophrenism accounted for by IQ measures and SES in multiple regression. Figure 9a. SES had a significant additive effect when ZPIQ was the other predictor of schizophrenism. As seen in the figure, to the extent that SES (horizontal lines) was preditive of schizophrenism, the shared variance (crossed lines) of the two variables is shared only slightly by ZPIQ (vertical lines). Thus, ZPIQ shares only a small amount of their mutual variance. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. I l l FIGURE 9b VIC SES Schiz’ism Figure 9b. SES did not have a significant additive effect when, for example, VIQ was the other predictor of schizophrenism. To the extent that SES (horizontal lines) was predictive of schizophrenism, the shared variance of the two variables (crossed lines) was already largely accounted for by VIQ (vertical lines). Age 3 versus age 1 1 findings in the present study The absence of significant findings at age 3 may be a reflection of a developmental process in which the putative marker (cognitive deficits) manifests itself differently over the life-span. In fact, a number of researchers (Walker & Lewine, 1990; Walker and Gale, 1995; Davis et al., 1993; Fish et al., 1992) have found that neuromotor abnormalities, also believed to be manifestations of the genetic vulnerability to schizophrenia are most pronounced in the first two years and at the end of life, creating a U-shaped relation between age and neuromotor abnormalities in schizophrenics (Walker & Gale, 1995). Interestingly, in a study on the same Mauritian sample used in the present study, Lencz Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (1995) also found a U-shaped relation between age and a putative marker of schizotypy; adolescents who had high scores on the schizophrenism scale showed psychophysiological abnormalities in an orienting-tone paradigm at age 3, but not at age 11 . Other studies (reviewed by Raine et al., 1995) have revealed similar electrodermal abnormalities in adult schizotypals. However, whereas these findings point to changes over time in the manifestations of putative indicators of SSPD, the pattern demonstrated in the present study is not U shaped, but seems to be linear, with no manifestation of the indicator at age 3, and a strong relationship between indicator and later SSPD at age 11 . Meehl asserts (1962) that though the integrative neural defect is genetic and present at birth, a process occurs over time by which the defect may implicate an increasing number of neuronal pathways, and may manifest itself in different systems at different ages. In the present study, the presence of cognitive deficits at age 11 but not at age 3 might be explained by this notion; the neurocognitive functions measured by IQ tests at age 3 might not yet have been affected by the integrative neural defect. The findings of other putative indicators at earlier ages such as skin conductance abnormalities or basic neuromotor abnormalities may be accounted for by the fact that they are “closer to the gene” (Meehl, 1990) than neurocognitive indicators. However, some studies which have directly looked for a prodromal decline in IQ by comparing early childhood scores with later childhood scores have not found statistical differences between preschizophrenic children and controls (e.g. Watt & Lubensky, 1976). However, findings are mixed; Lane & Albee (1968) did find stable differences on IQ between preschizophrenics and controls in both 2nd and 6th grade. The conflicting Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 113 findings in the literature may be explained by small sample sizes; the large sample in the present study lends power to the analyses, and does not find that IQ deficits are consistent over time. There are several other possible explanations of the disparate findings at age 3 and age 11. First, it is possible that the age 11 finding is the result of chance variance; however, that seems unlikely as the sample was cross-validated, and findings were significant on both subsamples. Second, a methodological question arises in terms of the use of the age 3 measure of cognitive ability. The age 3 test is less widely used than the WISC-R. Additionally, as it is based on Piagetian principles, it is unclear whether or not such a test would be expected to show a strong relationship to later measures which are reflective of a range of cognitive processes, including language, education, attention, psychomotor speed, etc. Thus, the cognitive measure at age 3 may be testing a different process than the WISC-R at age 11, and this process may not be a putative indicator of schizotypy. In sum, it is difficult to interpret the lack of findings at age 3, but several explanations can be attempted: (1) It is possible that IQ deficits grow over the course of childhood, and cannot be assessed at early ages, and (2) the age 3 and age 11 measures may not be assessing the same cognitive skills. Sex Differences Based on literature revealing significantly lower IQ scores in preschizophrenic (Offord, 1974; Watt & Lubensky, 1976) and high-risk (Rieder et al., 1977; Lane et al.. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 114 1970) males than in females, it was predicted that males would show greater deficits in IQ at ages 3 and 11 as compared to females. However, findings did not support this prediction; instead, no main effects of sex and no group by sex interactions emerged in the ANOVAs or in the regression analyses. This finding is surprising in light of research on sex differences in schizophrenia in age of onset (e.g. Rosenthal, 1970; Lewine, 1980; Goldstein et al., 1989), clinical expression (e.g. Goldstein et al., 1989), concordance rates among primary relatives (e.g. Goldstein et al., 1989), and biological stressors (Mednick, 1970; Mirdal et al., 1974). In these studies, male schizophrenics are characterized as having higher levels of biological abnormalities and poorer prognosis than females. Additionally, studies have found sex differences in schizotypy in expression of “positive” and “negative” SPD traits (Raine, 1992), and left hemisphere overactivation (Raine & Manders, 1988), such that males show more “negative” traits and females show more “positive” traits. However, similar to the finding in the present study, several studies did not find differences in IQ levels between male and female schizophrenics (e.g. Bilder, et al., 1991) or schizotypals (e.g. Trestman et al., 1995) A possible confound in testing sex effects in the present study is that IQ assessments were done in childhood. Because boys and girls mature physically and cognitively at different rates, the differences usually found in adults may not have been evident, leading to similar results for males and females. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 115 The Biosocial Approach The results from the present study lend limited support to findings from a number of studies examining family stress and SES as moderators in the development of schizophrenia (e.g. Sameroff et al., 1987; Mednick et al., 1987; Walker et al., 1981), and SSPD (e.g. Raine, 1987). Specifically, the present study indicated two biosocial interactions: ( 1) an additive effect of lower SES to lower ZPIQ in predicting later schizophrenism, and (2) an interaction effect in which family stress moderated the relationship between ZPIQ and schizophrenism such that lower levels of family stress and lower ZPIQ predicted higher levels of schizophrenism. As discussed previously, these results indicate that ZPIQ may be a “purer” measure of the genetic diathesis. The findings also seem to strongly support a multifactorial view of the etiology of SSPD and schizophrenia. After close scrutiny of the data, a number of other issues emerge regarding each of these findings. SES and IQ: Additive effect. The additive model implies that SES-of-origin and IQ both independently predict later schizophrenism, so that SES is not a moderator or a mediator of the relationship between intellectual deficits and schizotypy. As in Model II (Figure lb) proposed in the present study, lower SES and lower IQ might additively predict a greater likelihood of higher levels of schizophrenism than either alone (see Figure 9a). The finding of an additive effect for SES with ZPIQ in the present study lends support to a number of studies which have found significant relationships between lower SES and schizophrenia (e.g. Hollingshead & Redlich, 1968; Miller, 1958; Haas, 1963; Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 116 Dohrenwend, 1974; Gift et al., 1988), and is at odds with studies that have not found SES to be a predictor of later schizophrenia (LaFosse, et al., 1994; Erlenmeyer-Kimling, et al ., 1991; Walker, et al., 1981). It is important to note that whereas the present study and those seemingly at odds with it used an SES-of-origin measure, the studies which found significant relationships between SES and schizophrenia used current SES as their measure. This discrepancy makes it more difficult to interpret the present study’s findings. To clarify this issue, it can be helpful to look at studies with similar designs to the present one. Unfortunately, few studies have reported examining both IQ and SES as predictors in the same model; however, in those studies that have (Erlenmeyer-Kimling et al., 1991; Walker et al., 1981), no significant additive effects of low SES and low IQ have been found to predict schizophrenia. Thus, the relationship between SES and schizophrenia in other studies might be accounted for by the “social drift” theory which posits that schizophrenics “drift” into lower levels of SES as their functioning becomes progressively impaired (Silverton & Mednick, 1984; Malama, et al., 1988). The social drift theory stands opposed to, for example, Model II, which assigns a causative role to low SES in the development of psychopathology. To the author’s knowledge, there are no reports of a biosocial study investigating IQ and SES as predictors of SSPD, and, as such, conclusions based on findings in the schizophrenia literature should be made with caution. Not only is support for the previous literature equivocal, but the findings of this study are also mixed; the additive effect of SES was present with only one IQ variable (ZPIQ), and not with any others. Because the high-risk studies described (Erlenmeyer- Kimling et al., 1991; Walker et al., 1981) did not use ZPIQ as a measure, findings from Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 117 the present study even more strongly indicate that the ZPIQ measure is possibly unique in acting as an indicator/biological marker of schizotypy. Additionally, it should be noted that, the relationship between lower SES and schizophrenia is often understood to result in poorer prenatal care and an increased risk of birth complications. Thus, a biological environmental variable (birth complications) may mediate the relationship of SES to schizophrenia, as well as IQ to schizophrenia. This would imply that multiple potentiators can mediate each other and affect the development of the illness (Erlenmeyer-Kimling et al., 1991). In the present study, tests of SES demonstrated that this variable was not a significant mediator or moderator of the interaction between IQ and later schizophrenism, although SES did contribute unique variance to the regression model predicting schizophrenism when ZPIQ was partialled. Family Stressors: Interactive effect. The interactive effect of the family stress index and ZPIQ in predicting schizophrenism showed a similar pattern to that hypothesized by Model IV (Figure Id) in the present study: schizophrenism was predicted by lower levels of ZPIQ and at lower levels of family stress, but not at higher levels of family stress (Figure 2). This finding seems at first glance to be at odds with literature which has demonstrated a significant relationship between increased childhood family stressors and later schizophrenia (e.g. Walker et al., 1981; Schulsinger et al., 1987; Mednick et al., 1987; Mednick & Schulsinger, 1968); the literature has shown to be significantly predictive stressors ( 1) absence of father from the home and (2) institutionalization of child. Watt & Nicholi (1979) also found a relationship between Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 118 childhood family stress and schizophrenia: a history of death of a parent was more common among schizophrenics than controls. In the present study, no significant main effects were found for family stress in directly predicting SSPD. However, family stress did show a moderating effect on the relationship between a lower IQ and schizophrenism. This effect, although appearing to be in the opposite direction of main effects found in the literature, can be seen to support the hypothesis that the ZPIQ deficit is a putative indicator of schizophrenism. A child who develops SSPD in a lower stress environment must have a stronger genetic vulnerability, and hence lower IQ scores, than schizotypals developing in a high stress environment where the stress itself more likely to influence the development of SSPD. Two studies of SSPD also examined family stress and SSPD in a biosocial model. A study that structurally resembles the present one (Bergman & Walker, 1995) examined a sample of children over a the course of a year who were drawn from a large high-risk, longitudinal project, to determine whether cognitive factors mediate or moderate the relation between parental psychiatric status (schizophrenia, psychiatric comparison, normal comparison) or parental maltreatment(maltreated as defined by Child Protective services, not maltreated) and child behavior. Findings reflected an interaction effect between greater parental maltreatment and lower IQ in predicting schizoid behavior in the children. Additionally, the interaction between Parental Psychiatric group and Maltreatment group did not have any significant effects. Group differences on IQ scores examined with an ANOVA revealed that the Maltreatment group, and not the Parental Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 119 Psychiatric Group, had significantly lower IQ scores and poorer attention on the digit-span task. Multiple regression analyses on data from both assessments demonstrated that, similar to the group comparisons, parental psychiatric status and parental maltreatment were predictive of schizoid behavior, but the only interaction predicting SSPD was between parental psychiatric status and lower WISC-R scores. Cognitive test scores on their own were not predictive of schizoid behavior, thus eliminating the possibility of a mediating effect. Pearson correlations indicated that lower intelligence was associated with schizoid behavior for the schizophrenia risk-group, but not for the other groups. In comparing the Bergman & Walker (1995) study to the present one, several important issues emerge. First, the decision to equate parental psychiatric status with parental maltreatment as vulnerability indicators seems an unusual choice, insofar as parental maltreatment is more parsimoniously conceptualized as a psychosocial stressor. By contrast, their findings of an interaction of increased maltreatment with IQ deficits can be re-interpreted with the roles of indicator and moderator reversed; viewed in this way, Bergman & Walker’s (1995) findings support Model HI. A second concern and possible reason for the lack of an interaction between the psychosocial stressor and the putative indicator in the Bergman & Walker (1995) study is the use of parental psychiatric status as an indicator variable. This choice is based on the higher probability of later schizophrenia in the high-risk sample. However, the generalizability of high-risk studies is limited by the fact that only approximately 10% of Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 120 schizophrenics have a schizophrenic parent (Gottesman & Shields, 1982). Thus, this variable may be limited as a putative indicator. A third finding of the Bergman & Walker (1995) study that is at odds with the present one is the lack of a finding of IQ differences in the parental psychiatric status groups. This result lends support to a hypothesis of IQ as potentiator of the diathesis (as discussed previously), and is consistent with other studies that showed no difference between high-risk and control children on standardized IQ tests (e.g. Cohler et al., 1977; Worland & Hesselbrock, 1980). However, the lack of parent psychiatric status group differences on an attentional task is at odds with most literature on executive function (e.g. Poreh, et al., 1995; Trestman et al., 1995; Harvey et al., 1981; Neale et al., 1984) which implies that there may be some other variable at work influencing cognitive test scores. One possible explanation lies with the choice of SSPD measure. While schizoid behavior is certainly in the schizophrenia spectrum, it taps into mostly “negative,” or deficit symptomatology. In the present study, deficit symptoms (anhedonia) were not related to childhood IQ deficits, but “positive” symptoms as measured by the schizophrenism scale were. Thus, it is possible that a different measure of schizotypal personality might have better elucidated the relationship between IQ deficits and SSPD in the Bergman & Walker (1995) study. Alternately, Bergman & Walker propose that this lack of group differences on cognitive performance is due to the mediating effect of low SES for both groups. They suggest that a psychosocial stressor (SES) is impacting on IQ to such an extent that any group differences in cognitive ability due to genetic factors are “washed out.” Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 121 In sum, the Bergman & Walker (1995) study utilized a similar methodology to the one used by the present study, but showed a number of differences in findings possibly due to (I) differences in measures, (2) designation of variables as moderator, mediator, independent variable and dependent variable, and (3) difference in sample (high-risk versus general population. Raine (1987) also examined the relationship of biosocial interactions in schizotypy. He studied assessed prisoners for psychopathy and schizotypy, intact versus broken home background, current IQ scores, and skin conductance response. Schizotypy was measured by a dimensional index of schizotypal personality derived from 9 scales that were indexed into two measures: (I) Schizophrenism, and (2) Anhedonia-Psychoticism. Home background was designated as “broken” if, before the age of 10, the subject had (1) been brought up primarily by institutions or foster parents, or (2) parents who were divorced. Twelve of the 36 subjects had a broken home background based on these criteria. Separate analyses on the “broken” home and “intact” home background groups revealed that in the “intact home” group, higher levels of schizophrenism were significantly related to lower IQ indices. There were no significant findings in the “broken home” sample, and no significant findings relating to psychopathy. These findings lend support to the hypothesis that lower IQ may be a specific indicator of the schizophrenia spectrum; results demonstrated a strong relationship between schizotypal traits and lower IQ, but not between psychopathic traits and IQ. Interestingly, in the Raine (1987) study, the relationship between lower IQ and SSPD was significant only in the subjects from intact homes, providing strong evidence that Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 122 psychosocial stress is a moderator of the relationship between IQ and SSPD. In the present study, the specificity of which schizotypal traits were related to IQ in low stress conditions was further examined, and the IQ deficits seem to be limited to ZPIQ. Additionally, findings (Raine, 1987) provide support for the diathesis-stress model proposed by Model IV in the present study: the genetic diathesis (as measured by deficits in skin conductance response and by lower IQ) is more visible in schizotypals from stable environments because they needed a stronger genetic “hit” to become schizotypal, whereas in stressful environments, schizotypal personality can develop partly from the influence of the psychosocial stress, consequently necessitating a weaker genetic diathesis. The present study sought to extend Raine’s study by (1) increasing generalizability of findings by studying similar interactions in the general population as opposed to in a criminal sample, (2) using a prospective (rather than retrospective) longitudinal design so as to more clearly elucidate the role of childhood IQ as a predictor of later SSPD, (3) test group differences in addition to dimensional relationships to explore the possibility that schizotypy might be better conceptualized as a taxon, and (4) examine psychosocial stress not only as a moderator but also as a mediator of the relationship between IQ and SSPD. Additional Methodological Issues The findings of the present study must be viewed in the context of a number of methodological issues. The following section will outline these issues as related to measurement, power, cultural issues and specificity. Methodological strengths of the study will also be discussed. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Measurement Cognitive measures: Several methodological issues emerge under close scrutiny of the cognitive measures. First, even while the age 3 cognitive measure was created for a Mauritian population and based on a reliable and valid test (Boehm, 1989), the measure itself has not been widely used. Thus, any interpretation for the lack of findings at age 3 must be made with caution. Second, as discussed previously, the performance and verbal factors of the WISC-R IQ probably involve both verbal and non-verbal processes, so that assumptions made about specific types of cognitive functioning should be made carefully (Lawson & Inglis, 1983; Benton, 1962). Additionally, neither subscale includes all the WISC-R items necessary to provide a total IQ score. Because only two sub-tests form the VIQ index and four form the PIQ index, the VIQ showed increased variance making it a less stable measure. However, as the VIQ, PIQ and ZPIQ were given to the entire sample, and their relative performance analyzed, relationships between performance on these tasks and later SSPD can still be assessed, albeit with caution. An important issue to consider is that IQ measures of cognitive functioning are necessarily indirect as compared to methods such as brain imaging. As Meehl asserted, when we search for indicators of the integrative neural defect, as it is better to get “closer to the gene” (Meehl, 1990). When measures may not be precise in testing which brain processes are involved in functioning, interpretation of results is also in question. Measures of SSPD. One noticeable outcome of the present study which may suggest a source of error is the lack of significant findings for IQ differences between the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 124 SSPD index and control groups as defined by the PDE interview. This is especially noteworthy because the PDE groups did differ significantly on the schizophrenism scale, and many significant findings were demonstrated using the schizophrenism measure. As discussed previously, it is possible that IQ deficits tap specifically into positive symptoms measured by the schizophrenism scale, whereas the PDE measures a wide range of SSPD symptoms, thus “diluting” any effects that might have manifested themselves. Another important issue regarding the PDE is the relatively low base rate of schizotypy as compared to other studies (e.g. Baron & Risch, 1987; Raine, 1991). Based on the literature, the decision was made to base SSPD diagnosis on one less criteria than usual (e.g. Fenton & McGlashan, 1989; Thaker et al., 1993). However, the reduced diagnostic specificity may have contributed to the lack of findings for the PDE. The lower rate of SSPD may be due to another issue: due to financial and practical constraints, only one rater was available to diagnose subjects, making any analysis of inter-rater reliability impossible. Given that there can be a large degree of inter-rater variability in the diagnosis of Axis II pathology in particular, and that raters can often be subject to a minimizing bias, (Zimmerman, 1994), it seems possible that the interviewer for the present study may have downplayed or normalized reported behavior that other raters might have viewed as pathological. Indeed, cultural issues may have played a role in producing such a minimizing effect. First, the diagnostic interviewer may not have rated certain criteria of SSPD, such as magical ideation, as pathological because they are culturally normative, as is indicated by DSM-IH-R (American Psychiatric Association, 1987). Second, Mauritian subjects may be more guarded than the Westerners on whom the PDE was normed, and Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. less comfortable talking about personal and psychological issues, and may self-report less psychopathology. In contrast, the SAE scales (Venables et al., 1990) were normed on the Mauritian population, and may provide a better basis for assessment of certain experiences. Additionally, the anonymous nature of the self-report format may have reduced defensive or guarded responding. Further, although the SAE scales do not make judgments or determinations as to whether certain criteria are or are not culturally acceptable, base rates of symptoms as assessed by the SAE were equivalent in the Mauritius and a British sample (Venables and Bailes, 1994). Finally, it is possible that the translation of the PDE into the native language of Mauritius did not accurately reflect the culture-specific ways in which SSPD pathology may be expressed. Again, the SAE avoids this source of measurement error because it was normed on a Mauritian sample. Psychosocial stress measures. The most striking possible source of error in relation to the psychosocial measures is the high degree of skew and kurtosis on both measures, even after square root transformations. The skew was due to the large number of subjects (1) with lower SES, and (2) with no family stressors. Consequently, the SES variable did not show the normal distribution found in the samples on which it was first standardized (Hollingshead & Redlich, 1958). Thus, findings on SES should be considered with caution. Several additional points can be made regarding the Family Stress Index (FSI). Only a limited number of stressor variables were available to form the index, thus possibly neglecting other important moderating stressors that would have influenced findings. Also, Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 126 although the FSI has some construct validity based on other studies using similar variables (e.g. Sameroff et al., 1987), this specific index has not been tested for reliability. Power: Power was clearly not an issue in terms of the regression analyses, in which sample sizes were between 300 and 400 subjects. However, the ANOVA equations had fewer subjects in each group (Anhedonia, index=54, control=29; Schizophrenism: 51,43; PDE: 38,38) The relatively low sample sizes increase the risk for Type II error. Still, sample size in the present study was as large or larger than all but one of the previous schizotypal studies of IQ. Cultural issues/specificity. Both a strength and weakness of this study pertains to the nature of the Mauritian sample, and the generalizability of findings to Western societies. Cross-cultural studies of schizophrenia (e.g. Orley et al., 1979; Gillis et al. 1982; Radhakrishnan et al. 1983) carried out through the International Pilot Study of Schizophrenia (IPSS) (World Health Organization, 1973) have demonstrated an essential similarity in the profile of core symptoms of schizophrenia, and have shown roughly equivalent population base-rates in developing countries and Western countries, indicating that SSPD might be similarly distributed. In fact, Murphy & Raman (1971) found very similar base rates of schizophrenia on Mauritius as compared to England and Wales. It is important to note, however, that the strategy used to determine the cross-cultural profile of schizophrenia relied on a priori criteria based on the Western diagnosis, and then looked for similar phenomena in other cultures, while excluding phenomena that did not Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 127 match the initial criteria (Sartorius et al., 1986; Fabrega, 1989). Thus, certain manifestations of schizophrenia and of SSPD may be present in other cultures, but not accounted for by clinical ratings of the disorders. Interestingly, when predictors of the course and outcome of schizophrenia, such as sociodemographic factors and past history factors, were examined cross-culturally, researchers concluded that predictive factors identified in Eastern cultures might not be relevant or sufficient for the study of prognosis or chronicity in other cultures (Murphy & Raman, 1971; Sartorius et al., 1978). Additionally, Murphy & Raman (1971) found that schizophrenics in Mauritius had a better prognosis than in Britain, and ascribed this difference to increased support of the ill subjects by their families and communities. However, in a study of an ethnically similar population, Kulhara & Wig (1978) found no significant differences in prognosis between North West Indians and white, European populations. The authors posited that the findings of different prognosis in the Mauritius study (Murphy & Raman, 1971) might be explained by the use of nurse’s reports on relapse as opposed to clinical diagnoses made personally by the researchers. While any assumptions about generalizability of the findings should be viewed with these cultural issues in mind, it is important to note that in initial studies of the self-report SSPD measures (SAE) indicate that results obtained in Mauritius are comparable to results obtained in the West (Venables & Bailes, 1994). Methodological Strengths. It is felt that a major strength of this study involves its longitudinal design. As a prospective investigation, it studies a relatively stable population Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 128 on key variables before they have developed a major psychological disorder. This minimizes the risk of confounding effects of labeling, institutionalization, medication, which are common in studies of schizophrenia, as well effects of drug and alcohol abuse on cognitive/intellectual functioning. The temporal sequencing of a longitudinal study allows for the examination of possible cause and effect variables in the etiology of SSPD, although causality cannot be absolutely determined. Another strength of this study is its sample size. Indeed, even after taking into account different sample sizes for each measure, the study provides a large enough sample to provide the statistical power needed to observe effects based upon previous studies which used smaller samples. Lastly, the study's biosocial approach provides a broad and integrative view of schizotypy, and fills a void in the research on the interaction between vulnerability and environmental stressors in the development of schizotypal personality. Summary and Directions for Further Research In summary, the present study demonstrated mixed support for both main hypotheses of the study. Hypothesis I, based on Meehl’s theory (1962; 1990) that IQ deficits associated with schizophrenia are genetic vulnerability markers, was supported by findings which revealed significant relationships between lower IQ scores at age 1 1 and later schizophrenism in both group comparisons and regression equations. However, predicted sex effects did not emerge in the analyses. Additionally, in the test of SSPT) as a Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 129 taxon or dimension, supplementary analyses suggested that schizophrenism is best understood as a dimension. Tests of the five biosocial models largely supported Model I (Meehl, 1962; 1990), which predicted that higher levels of schizotypy would not be related to psychosocial stress. However, there were two notable exceptions to this trend; an additive effect supporting Model II of lower SES and lower ZPIQ in predicting schizophrenism, and an interaction effect supporting Model IV between lower family stress and lower ZPIQ in predicting schizophrenism. The presence of ZPIQ in both of these findings suggested that the variable may be a “purer” indicator of the genetic vulnerability to schizotypy and schizophrenia, and that the ZPIQ deficit may be caused by abnormalities in interhemispheric transfer. Additionally, the fact that psychosocial stress did play an additive and interactive role in predicting schizophrenism lent support to the multifactorial, polygenic diathesis stress model of schizotypy and schizophrenia (Fowles, 1992). The lack of significant results when anhedonia and SSPD at age 20-22 were the dependent variables has many possible explanations, including (1) specificity of the IQ deficit as a putative indicator of only “positive” symptomatology as assessed by the schizophrenism scale, and (2) measurement issues including problems in interrater reliability, cultural factors and lack of specificity of symptoms (PDE). The lack of findings at age 3 may also be attributable to measurement error, or may be explained by a developmental course in which the manifestation of the integrative neural defect does not occur until the child has developed cognitively to some extent. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 130 Directions for future research. One important direction for future research would be to choose more valid and a greater variety of measures. For instance, if the full WISC was administered, interpretations of findings could be more conclusive. Additionally, if more specific measures of laterality were analyzed, the possible involvement the corpus callosum in manifesting the integrative neural defect could be clarified. Similarly, if biological environmental stressors such as birth complications were added into the analysis, a clearer picture of the mediating and moderating relationships between the variables might result. When examining the findings of the present study, an important question arises . Are IQ deficits biological markers of schizophrenia, or are they potentiators of the diathesis? The findings lend some support to the relationship between childhood IQ deficits and later SSPD, but only with age 11 IQ scores and schizophrenism. Thus, there does seem to be a relationship between IQ deficits and a genetically related disorder to schizophrenia (SSPD), partially supporting one of the criteria for biological marker status (Garver, 1987). Additionally, the IQ deficits are neurodevelopmental in origin, and thus support another criteria for biological marker. Although findings were not significant at ages 3 and 11, there are some limited findings in the literature showing IQ deficits in adult schizotypals, thus implying that there may be stability over time, again partially fulfilling one of the criteria for biological marker status. While there does seem to be limited support for IQ deficits as a biological marker, a strong argument can be made that IQ is a potentiator of the diathesis and not a putative indicator itself. In the present study findings were not conclusive enough to reject this possibility. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 131 The difficulty in the present study in determining the role of IQ deficits as an indicator or moderator (potentiator) leads to an important question: How can a variable be clearly defined as an indicator? Garver (1987) attempts to answer this question by identifying six criteria for a biological marker (indicator): ( 1) that the trait be distributed differently in psychotic than in control populations, (2) that the marker be stable over time, (3) that the trait be manifest at higher frequencies in family members than in the normal population and be associated with SSPD in family members, (4) that the trait predate the development of psychosis, (5) that there is a relationship between the trait and later SSPD and psychosis, and (6) that the marker can be measured reliably by a variety of investigators. Only a few potential markers have been tested on all these criteria at this time. Garver (1987) suggests that the Continuous Performance Test sensitivity score might be a biological marker, a view supported by Comblatt & Erlenmeyer-Kimling (1985). Meehl (1990) proposed that the P50 anomaly and smooth pursuit eye movement dysfunction may be strong indicators of the schizogene. Interestingly, Meehl, who originally suggested that researchers should be exploring potential risk markers (Meehl, 1962), also speaks to the difficulty in determining the indicator status of a trait (Meehl, 1990, p. 44): We must be able to answer affirmatively the question, “Are we now studying something basic, not schizophrenia, but a predisposer condition?” and yet be able to answer negatively the dangerous question “Are we now studying something not causally related to schizophrenia at all?” The latter, for example, can easily happen if an investigator latches onto a powerful polygenic potentiator, especially when the clinical penetrance of the disease is low, as the dominant model requires. While the presence of the specific schizogene is the sine qua non, the decompensation variance receives a heavy contribution from such a potentiator, so it could statistically swamp Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 132 the influence of the schizogene, due to the looseness of the syndrome and our lack of clear guidelines as to how diagnostic indicators change their relative importance when we move from the decompensated through the semicompensated to the compensated range. Meehl suggests that researchers should use a taxometric approach, and should examine possible indicators at different levels and from “qualitatively diverse domains” (Meehl, 1990, p. 44). This would enable a corroboration/discorroboration of a relationship between, for example, psychopathology and closer links to the CNS defect (e.g. P50 anomaly). Additionally, Meehl proposes that indicators should distinguish between subjects with psychosis, affective disorders, and normal controls. Thus, a biological marker should be specific to the psychotic process. This last point suggests that IQ deficits may not be an indicator of SSPD, but rather a powerful potentiator. A number of studies have shown that IQ deficits have been found in populations of sociopaths (e.g. Purcell, 1956), and depression is believed to produce cognitive deficits. Thus, it is likely that generalized IQ deficits may moderate a number of illnesses. However, research on specific patterns of IQ performance has been sparse; it is possible that a pattern of deficits in certain areas and advantages in others may better predict SSPD (e.g. ZPIQ). Exploring the possibility of such a pattern is an important direction for future research. Alternative Models Future research could clarify the answer to the questions posed above by testing parts of heuristic models of vulnerability, stress, and protective factors related to the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. development of schizophrenia such as the one proposed by Nuechterlein et al. (1992, 1994). In this model, personal vulnerability factors such as autonomic hyper reactivity, environmental potentiators and stressors such as psychosocial stressors, and protective factors all interact in the development of neurocognitive and information processing deficits. These, in turn, lead to prodromal symptoms, and, then finally to social, occupational, and psychopathological outcomes. This model incorporates the findings in the literature and includes many interactions between variables as well as feedback loops. This type of model can help in understanding not only the role of the variables that were examined in the present study, but also those which could still be tested and included in a more comprehensive analysis. Thus, a direction for future research involves investigating the predictive value as well as possible moderating and mediating effects of a number of proposed environmental stressors and putative indicators, such that comparisons can be made on which variables account for the most unique variance. This type of model, which used multiple predictors, mediators/moderators, and dependent variables, could be modified for the Mauritius sample, and employed to predict SSPD. This is especially important in light of the paucity of studies that examine both the psychosocial and cognitive aspects of schizotypy. Additionally, this study should include not only psychosocial stressors, but also biological stressors such as birth complications and maternal prenatal influenza. One model which were not explored in the present study, but which merits attention is Holzman’s pleiotropy model (Holzman et al., 1995). The pleiotropy model (Holzman et al., 1995) is a single-gene model in which a single schizotaxic gene can Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 134 manifest in many different ways, much like neurofibromatosis ("elephantitis"). Afflicted individuals with the same gene can have very different phenotypic presentations, from the distinctive tumorous lesions to more minor dermatological anomalies. Similarly, an individual with the schizotaxic gene can have psychotic-like symptoms, or verbal deficits, or performance deficits, or abnormal eye tracking, or any combination of these and other phenotypic manifestations of the single gene. One way to explore this model in the context of SSPD is to increase the number of possible indicators being investigated, and compare variance between index subjects and controls. It is important to note, however, that short of a direct gene study, it would be difficult to disprove the pleiotropy model. The nature of most statistical analyses examines linear relationships; however, a plausible model for the development of schizotypy may lie in a transactional model of the dynamic interaction between the biological and the environmental. In such a model, environmental and biological factors would not interact with each other in a direct and linear fashion; instead, they would affect and change each other so that the factors themselves would be different the next time they interacted. This type of model might clarify whether variables with a genetic component, such as IQ, are biological markers of the genetic vulnerability or independent potentiators of the diathesis. Additionally, such a model could explain differences over time in putative indicators, as in the present study. This model takes into account dynamic real-life interactions in the development of schizotypy. As such, it is has potentially more verisimilitude. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 135 References Abrahamson, D. (1983). Schizophrenic deterioration. British Journal of Psychiatry. 143: 82-83. Albee, G., Lane, E., & Reuter, J. (1964). Childhood intelligence of future schizophrenics and neighborhood peers. Journal of Psychology. 38:141-144. Allen, J.J., Chapman, L.J., Chapman, J.P. Vuchetich, J.P., & Frost, L.A. (1987). Prediction of psychotic-like symptoms in hypothetically psychosis-prone college students. Journal of Abnormal Psychology. 96:83-88. American Psychiatric Association, (1987). Diagnostic and Statistical Manual of Mental Disorders. 3rd Edition. Revised. Washington, D.C.:APA Press. American Psychiatric Association, (1980). Diagnostic and Statistical Manual of Mental Disorders. 3rd Edition. Washington, D.C.:APA Press. Andreasen, N.C., Shore, D., Burke, J.D., Grove, W.M., Lieberman, J.A., Oltmanns, T.F., Pettegrew, J.W., Pulver, A.E., Siever, L.J., Tsuang, M.T. & Wyatt, R.J. (1988). Clinical phenomenology. Schizophrenia Bulletin. 14:2-21. Arndt, S., Alliger, R.J., & Andreasen, N.C. (1991). The distinction of positive and negative symptoms: The failure of a two-dimensional model. British Journal of Psychiatry. 158:317-322. Asamow, J.R. & Ben-Meir, S. (1988). Children with schizophrenia spectrum and depressive disorders: a comparative study of premorbid adjustment, onset pattern and severity of impairment. Journal of Child Psychology and Psychiatry and Allied Disciplines. 29:477-88. Aylward, E., Walker, E., & Bettes, B. (1984). Intelligence in schizophrenia: Meta-analysis of the research. Schizophrenia Bulletin. 10:430-459. Balough, D.W., & Merritt, R.D. (1985). Susceptibility to Type A backward masking among hypothetically psychosis-prone college students. Journal of Abnormal Psychology. 94: 377-383. Baron, M. & Risch, N. (1987). The spectrum concept of schizophrenia: evidence for a genetic-environmental continuum. Journal of Psychiatric Research. 21:257-67. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 136 Baron, R.M., & Kenny, D.A. (1986). The moderator-mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology. 51: 1173-1182. Benishay, D.S. (1993). Cerebral Asymmetry and Schizotypal Personality. Unpublished Master’s thesis. Benton, A.L. (1962). Clinical symptomatology in right and left hemisphere lesions. In V.B. Mountcastle (ed.) Interhemispheric Relations and Cerebral Dominance. Baltimore: Johns Hopkins University Press. Bernstein, A.S. & Riedel, J.A. (1987). Psychophysiological response patterns in college students with high physical anhedonia: scores appear to reflect schizotypy rather than depression. Biological Psychiatry. 22, 829-847. Bergman, A.J. & Walker, E. (1995). The relationship between cognitive functions and behavioral deviance in children at risk for psychopathology. Journal of Child Psychology and Psychiatry and Allied Disciplines. 36:265-278. Bilder, R.M. & DeGreef, G. (1991). Morphologic markers of neurodevelopmental paths to schizophrenia. In S.A. Mednick (Ed.), Developmental neuropathology of schizophrenia. New York: Plenum. Bilder, R.M., Lipschutz, B.L., Reiter, G., Geisler, S.H., MayerhofF, D.I., & Lieberman, J.A. (1992). Intellectual deficits in first-episode schizophrenia: evidence for progressive deterioration. Schizophrenia Bulletin. 18:437-448. Bilder, R.M., Lipschutz, B.L., Reiter, G., Geisler, S.H., MayerhofF, D.I., & Lieberman, J.A. (1991). Neuropsychological deficits in the early course of first episode schizophrenia. Schizophrenia Research. 5: 198-199. Bilder, R.M., Wu, H., Bogerts, B., Degreef, G., Ashtari, M„ Alvir, J.M.J., Snyder, P.J., & Lieberman, J.A. (1994). Absence of regional hemispheric volume asymmetries in first-episode schizophrenia. American Journal of Psychiatry, 151: 1437-1447. Boehm, A. (1989). Boehm Test of Basic Concepts - Preschool Version. San Antonio: The Psychological Corporation. Bower, E., Shellhamer, T., and Daily, J. (1960). School characteristics of male adolescents who later became schizophrenic. American Journal of Orthopsychiatry. 30: 712- 729. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 137 Broks, P. (1984). Schizotypy and hemisphere function-11: Performance asymmetry on a verbal divided visual field task. Personality and Individual Differences, 649- 656. Broks, P., Claridge, G, Matheson, J, and Hargreaves, J. (1984). Schizotypy and hemisphere function-IV: Story comprehension under binaural and monaural listening conditions. Personality and Individual Differences. 5: 665-670. Cannon, T.D., Zorilla, L.E., Shtasel, D., & Gur, R. (1994). Neuropsychological functioning in siblings discordant for schizophrenia and healthy volunteers. Archives of General Psychiatry. 51: 651-661. Cannon, T.D., Mednick, S.A., and Parnas, J. (1989). Genetic and perinatal determinants of structural brain deficits in schizophrenia. Archives of General Psychiatry. 46: 883-889. Chapman, L.J., Chapman, J.P., Kwapil, T.R., Eckblad, M., & Zinser, M.C. (1994). Putatively psychosis-prone subjects 10 years later. Journal of Abnormal Psychology. 103:171-183. Chapman, L.J., Chapman, J.P., & Raulin, M.L. (1976). Scales for physical and social anhedonia. Journal of Abnormal Psychology. 85:374-382. Claridge, G. (1987). ‘The schizophrenias as nervous types’ revisited. British Journal of Psychiatry. 151:735-743. Claridge, G. & Beech, T. (1995). Fully and quasi-dimensional construction of schizotypy. In A. Raine, T. Lencz, & S. A. Mednick, (eds.) Schizotypal Personality. New York: Cambridge University Press. Claridge, G. and Broks, P. (1984). Schizotypy and hemisphere function-I. Theoretical considerations and the measurement of schizotypy. Personality and Individual Differences. 5: 633-648. Cohen, J. (1988). Statistical Power Analysis for the Behavior Sciences. Hillsdale, N.J.: Lawrence Erlbaum Associates. Cohen, J., & Cohen, P. (1983). Applied Multiple Regression for the Behavioral Sciences. 2nd Edition. Hillsdale, N.J.: Lawrence Erlbaum Associates. Cohler, B., Brunebaum, H., Weiss, J., Gamer, E., & Gallant, D. (1977). Disturbance of attention among schizophrenic, depressed, and well mothers and their young children. Journal of Child Psychology and Psychiatry. 18:115-135. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 138 Comblatt, B. & Erlenmeyer-Kimling, L. (1985). Global attentional deviance as a marker of risk for schizophrenia: Specificity and predictive validity. Journal of Abnormal Psychology. 94:470-486. Comblatt, B. & Erlenmeyer-Kimling, L. (1984). Early attentional predictors of adolescent behavioral disturbances in children at risk for schizophrenia. In: N.F. Watt, E.J. Anthony, L.C. Wynne, & J.E. Rolf (eds.) Children at Risk for Schizophrenia: A Longitudinal Perspective. N.Y.: Cambridge University Press, 198-212. Comblatt, B.A., & Keilp, J.G. (1994). Impaired attention, genetics, and the pathophysiology of schizophrenia. Schizophrenia Bulletin. 20:31-46. Comblatt, B. A., Lenzenweger, M.F., Dworkin, R.H., & Erlenmeyer-Kimling, L. (1992). Childhood attentional dysfunctions predict social deficits in unaffected adults at risk for schizophrenia. British Journal of Psychiatry. 161:59-64. Crawford, J.R., Besson, J.A, Bremner, M., Ebmeier, K.P., Cochrane, R.H., & Kirkwood, K. (1992). Estimation of premorbid intelligence in schizophrenia. British Journal of Psychiatry. 161:69-74. Crow, T.J. (1985). The two-syndrome concept: Origins and current status. Schizophrenia Bulletin. 11: 471-486. Crow, T.J. (1980). Molecular pathology of schizophrenia: More than one disease process? British Medical Journal. 280:66-68. Crow, T.J., Ball, J., Bloom, S.R., Brown, R., Bruton, C.J., Colter, N., Frith, C.D., Johnstone, E.C., Owens, D.G.C., and Roberts, G.W. (1989). Schizophrenia as an anomaly of development of cerebral asymmetry: a post-mortem study and a proposal concerning the genetic basis of the disease. Archives of General Bsychiatry, 4ft 1145-1150. Davie, R., Butler, N., and Goldstein, H. (1972). From Birth to Seven. London: Longman Group. Davis, K.L., Kahn, R.S., Ko, G., & Davidson, M. (1991). Dopamine in schizophrenia: A review and reconceptualization. American Journal of Psychiatry. 148: 1474-1486. Dohrenwend, B.P. (1974). Social and cultural differences in psychopathology. Annual Review of Psychology. 25:417-452. Dohrenwend, B.P., & Dohrenwend, B.S. (1969). Social Status and Psychological Disorder: A Causal Inquiry. New York: Wiley & Sons. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 139 Dohrenwend, B.P., Levav, I., Shrout, P.E., Schwartz, S., Naveh, G., Link, B.G., Skodol, A.E. & Stueve, A. (1992). Socioeconomic status and psychiatric disorders: The causation selection issue. Science. 255: 946-952. Drillien, C. The Growth and Development of the Prematurely Bom Infant/1964T Baltimore: Williams & Wilkins Company. Eaves, L. Schulz, S.C., Sharp, M., Asher, D.M., Bromet, E., Cloninger, C.R., Gershon, E.S., Hall, J.C., Kaufmann, C, Kendler, K., McFarlin, D.E., Pert, C.B., & Stevens, J. (1988). Genetics, immunology, virology. Schizophrenia Bulletin. 14:21-39. Erlenmeyer-Kimling, L. & Comblatt, B. (1987). High-risk research in schizophrenia. A summary of what has been learned. Journal of Psychiatric Research. 21:401-411. Erlenmeyer-Kimling, L. & Comblatt, B. (1984). Biobehavioral risk factors in children of schizophrenic parents. Journal of Autism and Developmental Disorders. 14:357- 374. Erlenmeyer-Kimling, L., Rock, D., Squires-Wheeler, E., Roberts, S., and Yang, J. (1991). Early life precursors of psychiatric outcomes in adulthood in subjects at risk for schizophrenia and affective disorders. Psychiatry Research, 39: 239-256. Eysenck, H.J. (1960). Classification and the problem of diagnosis. In H.J. Eysenck (Ed), Handbook of Abnormal Psychology. London: Pitman. Fabrega Jr., H. (1989). The self and schizophrenia: A cultural perspective. Schizophrenia Bulletin. 15: 277-290. Fenton, W.S., & McGlashan, T.H. (1989). Risk of schizophrenia in character disordered patients. American Journal of Psychiatry. 146:1280-1284 Ferraro, F. & Okerlund, M. (1995). Implicit memory of nonclinical schizotypal individuals. Perceptual and Motor Skills. 80: 371-376. Fish, B., Marcus, J., Hans, S.L., Auerbach, J.G., & Perdue, S. (1992). Infants at risk for schizophrenia: Sequelae of a genetic neurointegrative defect. Archives of General Psychiatry. 49:221-235. Flor-Henry, P. (1976). Lateralized temporal-limbic dysfunction and psychopathology. Annals of the New York Academy of Sciences, 280: 777-795. Flor-Henry, P. (1987). Cerebral dynamics, laterality, and psychopathology: A commentary. In R. Takahashi, P. Flor-Henry, J. Gruzelier, and S.I. Niwa, Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 140 (eds.) Cerebral Dynamics, laterality, and psychopathology. Amsterdam: Elsevier. Fowles, D.C. (1992). Schizophrenia: Diathesis-stress revisited. Annual Review of Psychology. 43:303-336. Gangestad, S., & Snyder, M. (1985). “To carve nature at its joints”: On the existence of discrete classes in personality. Psychological Review, 92: 317-349. Garver, D.L. (1987). Methodological issues facing the interpretation of high-risk studies: biological heterogeneity. Schizophrenia Bulletin. 13:525-529. Goldberg, T.E., Karson, C.N., Leleszi, J.P., & Weinberger, D.R. (1988). Intellectual impairment in adolescent psychosis. A controlled psychometric study. Schizophrenia Research. 1:261-266. Gift, T.E., Strauss, J.S., Ritzier, B.A., Kokes, R.F., & Harder D.W. (1988). Social class and psychiatric disorder: The examination of an extreme. Journal of Nervous and Mental Disease. 176:593-597. Gillis, L.S., Elk, R , Ben-Arie, 0. & Teggin A. (1982). The Present State Examination: Experiences with Xhosa-speaking psychiatric patients. British Journal of Psychiatry. 141: 143-147. Golden, R.R., & Meehl, P.E. (1979). Detection of the schizoid taxon with MMPI indicators. Journal of Abnormal Psychology. 88:212-233. Goldstein, J.M., Tsuang, M.T., & Faraone, S.V. (1989). Gender and schizophrenia: Implications for understanding the heterogeneity of the illness. Psychiatry Research 22:243-253. Gottesman, I.I., & Shields, J. (1982) Schizophrenia: The epigenetic puzzle New York: Cambridge University Press. Green, M. (1990). Family relations as risk factors for the onset and course of schizophrenia. In J. Rolf, A.S Masten, D. Cicchetti, K.H. Nuechterlein, & S. Weintraub, S. (eds.) Risk and Protective Factors in the Development of Psychopathology.. N.Y.: Cambridge University Press. Green, M.F., Satz, P., Ganzell, S., & Vaclav, J.F. (1992). Wisconsin Card Sorting Test performance in schizophrenia: remediation of a stubborn deficit. American Journal of Psychiatry. 149:62-67. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 141 Griffith, J., Mednick, S., Schulsinger, F. & Diderichsen, B. (1980). Verbal associative disturbances in children at high risk for schizophrenia. Journal of Abnormal Gruzelier, J. (1995). Syndromes of schizotypy: Patterns of cerebral laterality, arousal, and gender. In A. Raine, T. Lencz, and S.A. Mednick (eds.) Schizotypal Personality. New York: Cambridge University Press. Gruzelier, J. (1984). Hemispheric imbalances in schizophrenia. International Journal of Psychophysiology. 1: 227-240. Gruzelier, J. (1981). Cerebral laterality and psychopathology: Fact and fiction. Psychological Medicine. 11: 93-108. Gruzelier, J. & Manchanda, R. (1982). The syndrome of schizophrenia: Relations between electrodermal response, lateral asymmetries and clinical ratings. British Journal of Psychiatry. 141: 488-495. Gunderson, J.G., Siever, L.J. & Spaulding, E. (1983). The search for a schizotype: Crossing the border again. Archives of General Psychiatry. 40: 15-22. Hamlin, R. (1969). The stability of intellectual function in chronic schizophrenia. Journal of Nervous and Mental Disease. 149:495-503. Haas, K. (1963). The middle class professional and the lower class patient. Mental Hygiene. 47: 408-410. Harvey, P., Winters, K.C., Weintraub, S. & Neale, J.M. (1981). Distractability in children vulnerable to psychopathology. Journal of Abnormal Psychology. 90: 298-304. H off*, A.L., Riordan, H., O’Donnell, D.W., Strizke, P., Neale, C., Boccio, A., Anand, A.K., & DeLisi, L.E. (1992). Anomalous lateral sulcus asymmetry and cognitive function in first-episode schizophrenia. Schizophrenia Bulletin. 18: 257-270. Hollingshead, A.B. (1965). Two factor index of social position ('copyrighted 1957), privately printed, Yale Station, New Haven, Connecticut. Hollingshead, A.B., & Redlich, F.C. (1958). Social Class and Mental Illness. John Wiley Holzer, C.E., Shea, B.M., Swanson, J.W., & Leaf, P.J. (1986). The increased risk for specific psychiatric disorders among persons of low socioeconomic status. American Journal of Social Psychiatry. 6:259-271. 89:125-131. and Sons. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 142 Holzman, P.S., Bivens, L.W., Bower, G.H., Campos, J.J., Carpenter, P., Ekman, P., Klatzky, R.L., Suomi, S.J., Tellegen, A. & Thompson, R.F. (1988). Basic behavioral sciences. Schizophrenia Bulletin. 14:68-82. Holzman, P.S., Coleman, M., Lenzenweger, M.F., Levy, D.L., Matthysse, S., O’Driscoll, G., & Park, S. (1995). Working memory deficits, antisaccades, and thought disorder in relation to perceptual aberration. In A. Raine, T. Lencz, & S. A. Mednick, (eds.). Schizotypal Personality. N.Y.: Cambridge University Press. Jaccard, J., Turrisi, R., & Wan, C.K. (1990). Interaction Effects in Multiple Regression. Newbury Park: Sage Publications. Kelley, M.P. & Coursey, R.D. (1992). Lateral preference and neuropsychological correlates of schizotypy. Psychiatry Research. 41:115-135. Kendler, K.S. & Gruenberg, A.M. (1982). Genetic relationship between paranoid personality disorder and the ‘schizophrenic spectrum disorders. American Journal of Psychiatry. 139: 1185-1186. Kendler, K.S., Ochs, A.L., Gorman, A.M., Hewitt, J.K., Ross, D.E., & Mirsky, A.F. (1991). The structure of schizotypy: apilot multitrait twin study. Psychiatry Research. 36: 19-36. Kendler, K.S., Neale, M.C., & Walsh, D. (1995). Evaluating the spectrum concept of schizophrenia in the Roscommon Family Study. American Journal of Psychiatry. 152:749-754. Kendler, K.S., Walsh, D., Su, Y., McGuire, M., Spellman, M., Lytle, C., McCormick, O., O’Neill, A., Shinkwin, R., Nuallain, N., O’Hare, A., Kidd, K., MacLean, C., & Diehl, S. (1989). The Roscommon family and linkage study of schizophrenia: Preliminary report. Abstract presented at the 28th annual meeting of the American College of Neuropsychopharmacology. Kolakowska, T., Williams, A.O., Arden, M., Reveley, M.A., Jambor, K., Gelden, M.G., & Mandelbrote, B.M. (1985). Schizophrenia with good and poor outcome. I. Early clinical features, response to neuroleptics, and signs of organic dysfunction. British Journal of Psychiatry. 146:229-246. Kraeplin, E. (1971/1919). Dementia Praecox and Paraphrenia. R.M. Barclay, Trans., G.M. Robertson, Ed. Huntington, N.Y.: Robert Krieger. Kulhara, P. & Wig, N.N. (1978). The chronicity of schizophrenia in North West India. British Journal of Psychiatry. 132: 186-190. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 143 LaFosse, J.M., Mednick, S.A., Praestholm, J., Vestergaard, A., Paraas, J., & Schulsinger, F. (1994). The influence of parental socioeconomic status on CT studies of schizophrenia. Schizophrenia Research. 11.285-290. Lane, E., & Albee, G. (1968). On childhood intellectual decline of adult schizophrenics: A reassessment of an earlier study. Journal of Abnormal and Social Psychology. 73:174-177. Lane, E. & Albee, G. (1964). Early childhood intellectual differences between schizophrenic adults and their siblings. Journal of Abnormal and Social Psychology. 6 8: 193-195. Lane, E. & Albee, G. (1965). Childhood intellectual differences between schizophrenic adults and their siblings. American Journal of Orthopsychiatry. 35: 747-753. Lane, E., Albee, G., & Doll, L. (1970). The intelligence of children of schizophrenics. Developmental Psychology. 2:315-317. Lawson, J.S., & Inglis, J. (1983). A laterality index of cognitive impairment after hemispheric damage: A measure derived from a principle-components analysis of the Wechsler Adult Intelligence Scale. Journal of Consulting and Clinical Psychology. 51:832-840. Lencz, T. (1995). Psychophvsiological and behavioral antecedents of schizophrenia- spectrum personality disorder. Unpublished doctoral dissertation. Lencz, T., Raine, A., Benishay, D.S., Mills, S., and Bird, L. (1995). Neuropsychological abnormalities associated with schizotypal personality. In A. Raine, T. Lencz, & S. A. Mednick, (eds.). Schizotypal Personality. N.Y.: Cambridge University Press. Lencz, T., Raine, A., Scerbo, A.S., Redmon, M., Brodish, S., Holt, L., & Bird, L. (1993). Impaired eye tracking in undergraduates with schizotypal personality disorder. American Journal of Psychiatry. 150:152-154. Lenzenweger, M.F., Comblatt, B.A., & Putnick, M.E. (1991). Schizotypy and sustained attention. Journal of Abnormal Psychology. 100:84-89. Lenzenweger, M.F. & Korfine, L. (1992). Confirming the latent structure and base rate of schizotypy: a taxometric analysis. Journal of Abnormal Psychology 101:567-71. Levin, S., Yurgelun-Todd, D., & Craft, S. (1989). Contributions of Clinical Neuropsychology to the Study of Schizophrenia. Journal of Abnormal Psychology. 98: 341-356. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 144 Lewine, R.R. (1980). Sex differences in age of symptom onset and first hospitalization in schizophrenia. American Journal of Orthopsychiatry 50:316-22. Liddle, P.F. (1987). The symptoms of schizophrenia: A re-examination of the positive- negative dichotomy. British Journal of Psychiatry. 151:145-151. Loranger, A.W. (1988). Personality Disorder Examination (PDE) Manual. Yonkers, N.Y.: DV Communications. Malama, I.M., Papaioanno, D.J., Kaklamani, E.P., Katsouyanni, K.M., Koumantaki, I.G., and Trichopolous, D.V. (1988). Birth order sibship size and socio-economic factors in risk for schizophrenia. British Journal of Psychiatry. 152: 482-486. Mednick, S.A. (1970). Breakdown in individuals at high risk for schizophrenia: Possible predispositional perinatal factors. Mental Hvgeine. 54:50-63. Mednick, S.A., Pamas, J. & Schulsinger, F. (1987). The Copenhagen high-risk project, 1962-1986. Schizophrenia Bulletin. 13:485-495. Mednick, S.A. & Schulsinger, F. (1968). Some pre-morbid characteristics related to breakdown in children with schizophrenic mothers. In D. Rosenthal & S.S. Kety (eds.) The Transmission of Schizophrenia. New York: Pergamon Press. Meehl, P.E. (1990). Toward an integrated theory of schizotaxia, schizotypy, and schizophrenia. Journal of Personality Disorders. 4:1-99. Meehl, P.E. (1989). Schizotaxia revisited. Archives of General Psychiatry. 46:935-944. Meehl, P.E. (1987). “Hedonic capacity” ten years later: Some clarifications. In D.C. Clark & J. Fawcett (eds.) Anhedonia and Affect Deficit States. New York: PMA Publishing. Meehl, P.E. (1975). Hedonic capacity: some conjectures. Bulletin of the Menninger Clinic. 39: 295-307. Meehl, P.E. (1962). Schizotaxia, schizotypy, schizophrenia. American Psychologist 17:827-38. Miller, W.B. (1958). Lower class culture as a generating milieu of gang delinquency. Journal of Social Issues. 14:5-19 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 145 Mirdal, G.K.M., Mednick, S.A., Schulsinger, F., & Fuchs, F. (1974). Perinatal complications in children of schizophrenic mothers. Acta Psvchiatrica Scandinavica- 50:553-568. Murphy, H.B.M. & Raman, A.C. (1971). The chronicity of schizophrenia in indigenous tropical peoples: Results of a twelve-year follow-up survey in Mauritius. British Journal of Psychiatry. 118: 489-497. Myers, F.R. (1979). Emotions and the self: A theory of personhood and political order among Pintupi Aborigines. Ethos. 7: 343-370. Myers, J.K., & Bean, L.L. (1968). Two-factor index of social position, in A decade later: a follow-up of social class and mental illness. John Wiley and Sons, 235-237. Myers, J.K. & Roberts, B.H. (1959). Family and Class Dynamics in Mental Illness. New York: Wiley & Sons. Nuechterlein, K.H., & Dawson, M.E. (1984). A heuristic vulnerability/stress model of schizophrenic episodes. Schizophrenia Bulletin. 10:300-312. Nuechterlein, K.H., Dawson, M.E., Gitlin, M., Ventura, J., Goldstein, M.J., Snyder, K.S., Yee, C.M., & Mintz, J. (1992). Developmental processes in schizophrenic disorders: Longitudinal studies of vulnerability and stress. Schizophrenia Bulletin. 18:387-424. Nuechterlein, K.H., Dawson, M.E., Ventura, J., Gitlin, M., Subotnik, K., Snyder, K.S., Mintz, J., & Bartzokis, G. (1994). The vulnerability/stress model of schizophrenic relapse: a longitudinal study. Acta Psvchiatrica Scandinavica. 89:58-64. Offord, D. (1974). School performance of adult schizophrenics, their siblings, and age mates. British Journal of Psychiatry. 125:12-19. Offord, D. & Cross, L. (1971). Adult schizophrenia with scholastic failure or low IQ in childhood. Archives of General Psychiatry. 24:431-435. Orley, J., Blitt, D.M., & Wing, J.K. (1979). Psychiatric disorders in two African villages. Archives of General Psychiatry. 36: 513-520. Parker, R. & Davidson, N. (1963). A comparison of students of nursing and hospitalized patients on scores derived from an intelligence test (WAIS). Psychiatric Quarterly Supplement. 37:298-306. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 146 Payne, R. (1961). Cognitive abnormalities. In H.J. Eysenck (ed.) Handbook of Abnormal Psychology. New York: Basic Books. Poreh, A.M., Ross, T.P., & Whitman, D.R. (1995). Re-examination of executive functions in psychosis-prone college students. Personality and Individual Differences. 18: 535-539. Purcell, K. (1956). A note on the Porteus Maze and the Wechsler-Bellevue scores as related to antisocial behaviour. Journal of Consulting Psychology. 20: 361-364. Radhakrishnan, J., Matthew, K., Richard, J. & Verghese, A. (1983). Schneider’s first-rank symptoms— prevalance, diagnostic use and prognostic implications. British Journal of Psychiatry, 142: 557-559. Raine, A. (1992). Sex differences in schizotypal personality in a nonclinical population. Journal of Abnormal Psychology. 101:361-364. Raine, A. (1991). The SPQ: a scale for the assessment of schizotypal personality based on DSM-DI-R criteria. Schizophrenia Bulletin. 17: 555-564. Raine, A. (1987). Effect of early environment on electrodermal and cognitive correlates of schizotypy and psychopathy in criminals. International Journal of Psychophysiology. 4:277-287. Raine, A., Benishay, D.S., Lencz, T., & Scarpa, A. (In press). Abnormal electrodermal orienting in schizotypal personality disorder. Schizophrenia Bulletin. Raine, A., Brennan, P., Mednick, B., & Mednick, S.A. (1996). High rates of violence, crime, academic problems, and behavioral problems in males with both early neuromotor deficits and unstable family environments. Archives of General Psychiatry. 53:544-549. Raine, A., Lencz, T., & Benishay, D.S. (1995). Schizotypal personality and skin conductance orienting. In A. Raine, T. Lencz, & S.A. Mednick, (eds.). Schizotypal Personality. N.Y.: Cambridge University Press. Raine, A., and Manders, D. (1988). Schizoid personality, inter-hemispheric transfer, and left hemisphere over-activation. British Journal of Clinical Psychology. 27: 333-347. Raine, A., Reynolds, C. Lencz, T., Scerbo, A., Triphon, N., & Kim, D. (1994). Cognitive- perceptual, interpersonal, and disorganized features of schizotypal personality. Schizophrenia Bulletin. 20: 191-201. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 147 Rawlings, D., and Borge, A. (1987). Personality and hemisphere function: two experiments using the dichotic shadowing technique. Personality and Individual Differences. fi:483-488. Rawlings, D., and Claridge, G. (1984). Schizotypy and hemisphere function-III: Performance asymmetries on tasks of letter recognition and local-global processing. Personality and Individual Differences. 5: 657-663. Reiss, D., Plomin, R., & Hetherington, E.M. (1991). Genetics and psychiatry: An unheralded window on the environment. American Journal of Psychiatry. 148: 283-291. Rieder, R., Broman, S., & Rosenthal, D. (1977). The offspring of schizophrenics. Archives of General Psychiatry. 34:789-799. Risch, N. (1990). Genetic linkage and complex diseases, with special reference to psychiatric disorders. Genetic Epidemiology. 7:3-16. RofF, J., Knight, R & Wertheim, E. (1976). Disturbed preschizophrenics. Journal of Nervous and Mental Disease. 162: 274-281. Rosenthal, D. (1970). Genetic Theory and Abnormal Behavior. New York: McGraw-Hill. Rubin, R., & Balow, B. (1979). Measures of infant development and socioeconomic status as predictors of later intelligence and school achievement. Developmental Psychology. 15:225-227. Rutschmann, J., Comblatt, B., & Erlenmeyer-Kimling, B. (1986). Sustained attention in children at risk for schizophrenia: Findings with two visual continuous performance tests in a new sample. Journal of Abnormal Child Psychology. J L 4 : 365-385. Rutter, M. (1990). Psychosocial resilience and protective mechanisms. In Risk and Protective Factors in the Development of Psychopathology. Rolf, J., Masten, A.S., Cicchetti, D., Nuechterlein, K.H., & Weintraub, S. (eds.) N.Y.: Cambridge University Press. Saccuzzo, D.P., & Braff, D.L. (1981). Early information processing deficit in schizophrenia: New findings using schizophrenic subgroups and manic control subjects. Archives of General Psychiatry. 38:175-179. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 148 Sameroff, A., Seifer, R., Zax, M., & Barocas, R. (1987). Early indicators of developmental risk: Rochester longitudinal study. Schizophrenia Bulletin. 13:383-394. Sartorius, N., Jablensky, A., Korten, A, Emberg, G., Anker, M., Cooper, J.E., & Day, R. (1986). Early manifestations and first-contact incidence of schizoprhenia in different cultures. Psychological Medicine. 16: 909-928. Sartorius, N., Jablensky, A., & Shapiro, R. (1978). Cross-cultural differences in the short term prognosis of schizophrenic psychoses. Schizophrenia Bulletin. 4:102-112. Sattler, J.M. (1982). Assessment of Childrens' Intelligence and Special Abilities. (2nd ed.). Boston: Allyn and Bacon. Schreiber, H., Stolz-Bom, G., Heinrich, H., Komhuber, H.H., & Bom, J. (1992). Attention, cognition and motor perseveration in adolescents at genetic risk for schizophrenia and control subjects. Psychiatry Research. 44:125-140. Schulsinger, F., Pamas, J., Mednick, S.A., Teasdale, T.W., & Schulsinger, H. (1987). Heredity-environment interaction and schizophrenia. Journal of Psychiatric Research. 21:431-436. Schulsinger, F., Pamas, J., Petersen, E.T., Schulsinger, H., Teasdale, T.W., Mednick, S.A., Moller, L., & Silverton, L. (1984). Cerebral ventricular size in the offspring of schizophrenic mothers: A preliminary study. Archives of General Psychiatry. 41:602-606. Siever, L.J. (1995). Brain structure-function and the dopamine system in schizotypal personality disorder. In A. Raine, T. Lencz, & S.A. Mednick, (eds.). Schizotypal Personality. N.Y.: Cambridge University Press. Siever, L.J., Amin, F., Coccaro, E.F., Bernstein, D., Kavoussi, R.J., Kalus, O., Horvath, T.B., Warner, P., Davidson, M., & Davis, K. (1991). Plasma homovanillic acid in schizotypal personality disorder patients and controls. American Journal of Psychiatry. I4&: 1246-1248. Siever, L.J., Coursey, R.D., Alterman, I.S., Buchsbaum, M.S., and Murphy, D.L. (1984). Impaired smooth pursuit eye movement: vulnerability marker for schizotypal personality disorder in a normal volunteer population. American Journal of Psychiatry, i l l , 1560-1566. Siever, L.J., Friedman, L., Moskowitz, J., Mitropolou, V., Keefe, R., Roitman, S.L.. Merhige, D., Trestman, R., Silverman, J., and Mohs, R. (1994). Eye Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 149 movement impairment and schizotypal psychopathology. American Journal of Psychiatry. 15. , 1209-1215. Siever, L.J., Kalus, O.F., & Keefe, R.S. (1993). The boundaries of schizophrenia. Psychiatric Clinics of North America 16:217-44. Silverstein, S.M., Raulin, M.L., Pristach, E.A., & Pomerantz, J.R. (1992). Perceptual organization and schizotypy. Journal of Abnormal Psychology. 101:265-270. Silverton, L., & Mednick, S.A. (1984). Class drift and schizophrenia. Acta Psvchiatrica Scandinavica. 70:304-309. Simons, R.F. (1981). Electrodermal and cardiac orienting in psychometrically defined high-risk subjects. Psychiatry Research. 4, 347-356. Strauss, J.S., Carpenter, W.T., & Bartko, J.J. (1974). The diagnosis and understanding of schizophrenia. Part m . Speculations on the processes that underlie schizophrenic symptoms and signs. Schizophrenia Bulletin. 11:61-69. Thaker, G.K., Adamhi, H., Moran, M., and Lahti, A. (1993). Psychiatric illnesses in families of subjects with schizophrenia-spectrum personality disorders: High morbidity risks for unspecified functional psychoses and schizophrenia. American Journal of Psychiatry. 15Q, 66-71. Trestman, R.L., Keefe, R.S.E., Mitropolou, V., Harvey, P.D., deVegvar, M.L., Lees- Roitman, S., Davidson, M., Aronson, A., Silverman, J., & Siever, L.J. (1995). Cognitive function and biological correlates of cognitive performance in schizotypal personality disorder. Psychiatry Research. 59: 127-136. Tyrka, A.R., Haslam, N., & Cannon, T.D. (1995). Detection of a latent taxon of individuals at risk for schizophrenia-spectrum disorders. In A. Raine, T. Lencz, & S.A. Mednick, (eds.). Schizotypal Personality. N.Y.: Cambridge University Press. Venables, P.H. (1995). Schizotypal status as a developmental stage in studies of risk for schizophrenia. In A. Raine, T. Lencz, & S.A. Mednick, (eds.). Schizotypal Personality. N.Y.: Cambridge University Press. Venables, P.H. (1990). Fro p. 105 re cpt & anhedonia Venables, P.H. (1983). Cerebral mechanisms, autonomic responsiveness, and attention in schizophrenia. Nebraska Symposium on Motivation. 47-91. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 150 Venables, P.H. (1977). The electrodermal psychophysiology of schizophrenics and children at risk for schizophrenia: controversies and developments. Schizophrenia Bulletin. 3: 28-48. Venables, P.H., & Bailes, K. (1994). The structure of schizotypy, its relation to sub diagnoses of schizophrenia and to sex and age. British Journal of Clinical Psychology. 33: 277-294. Venables, P.H., Wilkins, S., Mitchell, D.A., Raine, A., & Bailes, K. (1990). A scale for the measurement of schizotypy. Personality and Individual Differences. 1 1 :481-95. Walker, E.F., & Gale, S. (1995). Neurodevelopmental processes in schizophrenia and schizotypal personality disorder. In A. Raine, T. Lencz, & S.A. Mednick, (eds ). Schizotypal Personality. N.Y.: Cambridge University Press. Walker, E., Hoppes, E., Emory, E., Mednick, S., & Schulsinger, F. (1981). Environmental factors related to schizophrenia in psychophysiologically labile high-risk males. Journal of Abnormal Psychology. 90:313-320. Walker, E.F. & Lewine, R.J. (1990). The prediction of adult-onset schizophrenia from childhood home-movies of patients. American Journal of Psychiatry. 147: 1052- 1056. Watt, N., Grubb, T, & Erlenmeyer-Kimling, L. (1982). Social, emotional, and intellectual behavior at school among children at high risk for schizophrenia. Journal of Consulting and Clinical Psychology. 50:171-181. Watt, N. & Lubensky, A. (1976). Childhood roots of schizophrenia. Journal of Consulting and Clinical Psychology. 44:363-375. Watt, N., & Nicholi, A. (1979). Early death of a parent as an etiological factor in schizophrenia. American Journal of Orthopsychiatry. 49: 465-475. Webb, C.T., & Levinson, D.F. (1991). Schizotypal and paranoid personality disorder in the relatives of patients with schizophrenia and affective disorders: a review. Schizophrenia Research. J_ L : 81-92. Wechsler, D. (1974). The Wechsler Intelligence Scale for Children-Revised. New York: The Psychological Corporation. Weinberger, D.R. (1987) Implications of normal brain development for the pathogenesis of schizophrenia. Archives of General Psychiatry. 44: 660-669. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 151 Wender, P.H., Rosenthal, D., Kety, S.S., Schulsinger, F., & Weiner, J. (1973). Social class and psychopathology in adotpees. Archives of general Psychiatry. 28: 318- 325. Wilkins, S. & Venables, P.H. (1992). Disorders of attention in individuals with schizotypal personality. Schizophrenia Bulletin. 18:717-723. Wing, J.K., Cooper, J.E., & Sartorius, N. (1974). The Measurement and Classification of Symptoms. Cambridge: Cambridge University Press. Worland, J. & Hesselbrock, V. (1980). The intelligence of children and their parents with schizophrenia and affective illness. Journal of Child Psychology and Psychiatry and Allied Disciplines. 21:191 -201. Worland, J., Weeks, D., Weiner, S., & Schechtman, J. (1982). Longitudinal prospective evaluations of children at risk. Schizophrenia Bulletin. 8:135-141. World Health Organization. ('1973V Report of the international Pilot Study of Schizophrenia. Vol. 1. Geneva: World Health Organization. Zimmerman, M. (1994). Diagnosing personality disorders. Archives of General Psychiatry. 47: 527-531. Zubin, J. & Steinhauer, S. (1981). How to break the logjam in schizophrenia research. Journal of Nervous and Mental Disease. 169:477-492. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 152 Appendix 1 Cognitive Test at age 3 (based on Boehm Test of Concepts-Preschool Version, developed by Brian Sutton Smith) Comprehension 1 . Put the spoon in the cup. 2. Put the spoon in the cup and put both on the saucer. 3. Put the spoon in the cup. Put both in the saucer and put all of them on the tray. 4. Put the marble in the cup. Cover the cup with the saucer. Put the block on top of the saucer. 5. Put the marble in the cup. Cover the cup with the saucer. Put the block on top of the saucer. Put them all on the table. Movement 1. See the doll there. Bring me the doll and put it on the table. 2. See the ball. See the stick. Bring me the ball and the stick together and put them on the table. 3. See the cup. See the block. See the car. Bring the car, then bring me the cup, then bring me the block. 4. See the marble. See the cup. See the block. See the car. Bring me the marble, then bring me the block then bring me the cup, then bring me the car. 5. See the doll. See the marble. See the cup. See the block. See the car. Bring me the cup, then bring me the block, then bring me the marble, then bring me the car. Attention— 4 point rating scale of how child listens to and follows direction. Comprehension— 4 point rating scale of comprehension and directions. Expressive Language Labeling: 1. What do you call this? (Arm) 2 . 3. (Leg) (Neck) (Arm) (Leg) (Neck) 4. Show me your 5. 6 . 7. What is this? 8. What is this? 9. Show me the 10. 11. What color is this? (boy) (Girl) (Red) (Yellow) (Blue) (Green) (Black) (Picture of boy) (Picture of girl) 12. 13. 14. 15. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 153 16. Show me the one that is (Red) 17. (Yellow) 18. (Blue) 19. (Green) 20. (Black) 21. Tell me what you see in this picture. (Mother cooking) 22. (Girls washing clothes) 23. (Children playing ball) Verbalization: 4 point rating of child’s verbalization during assessment. Arithmetic Size: 1. How are these different? (Size) 2. (Length) 3. Which is the biggest? Which is the smallest? 4. Which is the longest? Which is the shortest? Classifications: 1. Which 2 are the same? 2. Which 2 are different? 3. Which one is different from the others? Discriminations: 1. Show me the one with a few. 2. Show me the one with a lot. Number concepts: 1. How many marbles are there? (3) 2 How many now? (5) 3. How many now? (7) 4. Give me 3 marbles. 5. Give me 5. 6. Give me 7. Adaptive behavior 1. Imitates examiner’s 3 block bridge 2. Builds a five block tower after demonstration (no model) 3. Circle construction with 10 blocks 4. Draw a circle. 5. Draw a square. 6. Draw a triangle. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 154 Appendix 2 Family Stress Index (FSI) As described in the text, two independent judges with experience in the area of social factors in schizophrenia rated the following family events/situations, representing potential stressors, each on a scale of 0- 10, with 0 meaning not stressful and 10 meaning maximally stressful. The number in the column after each item represents the mean of the two judges’ weightings for that item. For each item that is true for a subject, the corresponding weight is added to his or her total FSI score. CATEGORY VARIABLES AVERAGE WEIGHT 1. Father's employment status a. part-time 4 when subject is age 3: b. unemployed 6.5 2. Father's employment status age 11: a. part-time 4.5 b. unemployed 7.5 3. Change in father's employment status bet. ages 3 and 1 1 a. full-time to part-time 5.5 (be aware that this variable is related to #1 & #2 and could potentially b. part-time to unemployment 7.5 lead to overweighting) c. full-time to unemployment 7.5 4. Parental presence in home a. Subject lives in orphanage 9 when subject is age 3 b. orphaned, substitute parents 8.5 c. no father in home 5.5 d. one parent in home, not the mother 7 5. Parental presence in home at age 11: a. father not living with family 5.5 b. mother not living with family 7 c. neither father nor mother 9.5 living with family Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 155 Appendix 3 Survey of Attitudes and Experiences (Venables, et al., 1990) Schizophrenism Scale 1 . Sometimes people I know well begin to look like strangers. 2. Now and then when I look in a mirror, my face seems quite different from normal. 3. When introduced to strangers, I often wonder if I have seen them before. 4 .1 have sometimes felt that strangers were reading my mind. 5. Good luck charms don’t work. 6. 1 often get a restless feeling that I want something but do not know what. 7 .1 often change between positive and negative feelings towards the same person. 8. 1 suddenly feel shy when I talk to a stranger. 9 .1 prefer others to make decisions for me. 10. People can easily influence me even though I thought my mind was made up on a subject. 11.1 often have great difficulties controlling my thoughts when I am thinking. 1 2 .1 am not easily confused if a number of things happen at the same time. 13.1 find it difficult to concentrate; irrelevant things seem to distract me. 14.1 am not usually self-conscious. 15.1 am not much worried by humiliating experiences. Anhedonia Scale 1 . Beautiful scenery has been a great delight to me. 2. A brisk walk has sometimes made me feel good all over. 3. When I pass flowers I often stop to smell them. 4 .1 have been fascinated with the dancing of flames in fireplace. 5 .1 don't understand why people enjoy looking at the stars at night. 6. When I have been extremely happy I have sometimes felt like hugging someone. 7. Writing letters to friends is more trouble than it's worth. 8. When anticipating a visit from a friend, I have often felt happy and excited. 9 .1 get a lot of pleasure from listening to music. 1 0 .1 attach little importance to having close friends. 11. Getting together with close friends has been one of my greatest pleasures. 1 2 .1 have thoroughly enjoyed laughing at jokes with other people. 13. The idea of going out and mixing with people at partied has always pleased me. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Childhood precursors of adult reality distortion: A prospective study.

PDF

Childhood videotaped neuromotor and social precursors of schizophrenia: A prospective investigation

PDF

Associations and mechanisms among attention deficit hyperactivity disorder symptoms, cognitive functioning, and drinking habits

PDF

A longitudinal cross-ethnic study of sociocentricity in schizophrenia

PDF

Anatomic and functional specificity of relationships between explicit memory dysfunction and atrophy of medial temporal structures in Alzheimer's disease

PDF

Concordant and discordant drug use in intimate relationships: A longitudinal study

PDF

A behavioral -analytic model for assessing social competence in male batterers

PDF

Childhood Sexual Abuse and Depressive Symptoms In A Lesbian Population: An Exploratory Study.

PDF

An examination of reliable component analysis

PDF

Client characteristics, service characteristics, and psychosocial outcomes in the community treatment of schizophrenia

PDF

Cognitive predictors of intention -behavior consistency in safer sex practices

PDF

A correlational study of feminist/womanist identity development and depression in women

PDF

Social and psychophysiological correlates of under socialized-aggressive conduct disorder and socialized-aggressive conduct disorder in institutionalized male youth and adolescents

PDF

Cognitive aging: Expertise and fluid intelligence

PDF

Adolescent predictors and adult consequences of alcohol dysphoria: A longitudinal study

PDF

Creative aptitude patterns of college honors students

PDF

Characteristics and cortical localization of subliminal visual priming

PDF

Experimental evaluation of a new intervention designed to promote dementia caregivers' acceptance and empathic responding towards care -recipients

PDF

Attachment as a predictor of self -monitoring and sexual behavior

PDF

Hyperactive symptoms, cognitive functioning, and drinking habits

Asset Metadata

Creator Benishay, Deana S. (author)

Core Title Biosocial antecedents of schizophrenia-spectrum personality disorders: A longitudinal study

Degree Doctor of Philosophy

Degree Program Clinical Psychology

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag OAI-PMH Harvest,psychology, clinical,psychology, developmental,psychology, personality

Language English

Contributor Digitized by ProQuest (provenance)

Advisor Raine, Adrian (committee chair), Brekke, John (committee member), Dawson, Michael (committee member), Earleywine, Mitchell (committee member), Mednick, Sarnoff (committee member)

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c17-510495

Unique identifier UC11353479

Identifier 9705074.pdf (filename),usctheses-c17-510495 (legacy record id)

Legacy Identifier 9705074.pdf

Dmrecord 510495

Document Type Dissertation

Rights Benishay, Deana S.

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the au...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus, Los Angeles, California 90089, USA