University of Southern California Dissertations and Theses

Meta-analysis on the misattribution of arousal

(USC Thesis Other)

Meta-analysis on the misattribution of arousal

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content META-ANALYSIS ON THE MI5ATTRIBUTION OF AROUSAL by Kristyn Greenwood A Thesis Presented to the FACULTY OF THE GRADUATE SCHOOL UNIVERSITY OF SOUTHERN CALIFORNIA In Partial Fulfillment of the Requirements for the Degree MASTER OF ARTS Psychology May 19 95 UNIVERSITY O F SO U TH ER N CALIFORNIA THE GRADUATE SCHOOL UNIVERSITY PARK LOS ANGELES. CALIFORNIA 9 0 0 0 7 This thesis, written by under the direction of h&C.....Tkesis Committee, and approved by all its members, has been pre sented to and accepted by the Dean of The Graduate School, in partial fulfillment of the requirements fo r the degree of MASTER OF ARTS Dean Chairman 11 Contents Page List of Tables................................................. v List of Figures.............................................. vi Introduction................................................... 1 Misattribution Theories/Models............................ 2 Passionate Love Theory ................................. 3 Excitation Transfer...................................... 6 Predictions of Misattribution Theories ............. 12 Arousal and Attentional Shift Model .................... 12 Predictions of Arousal and Attentional Shift M o d e l ................................................... 14 Arousal and Hedonic Valence Model ....................... 15 Predictions of Arousal and Hedonic Valence M o d e l ................................................... 16 Reinforcement Theory......................................... 17 Predictions of the Reinforcement Theory............. 18 Incompatible Response Model ............................... 18 Predictions of the Incompatible Response Model . 19 Sex/Aggression Link Theories............................... 2 0 Predictions of Sex-Agression Link Models .......... 21 Comparisons Between Theoretical Predictions .......... 21 Overview of Met.a-Analyt.ic Procedures.................... 22 Method........................................................... 2 3 iii Page Literature Search Procedures............................... 24 Inclusionary Criteria ...................................... 25 Variables Coded From Each S t udy .......................... 2 6 Variables Judged From Each Study.......................... 27 Results........................................................ 30 Meta-Analytic Procedures ............................... 30 Results for Overall Effect Size....................... 3 9 Categorical Variables.................................... 4 6 Judged Variables ......................................... 51 Predicted Interactions ................................. 57 Discussion...................................................... 58 Unsupported Theories and Models .......................... 59 Support for the Misattribution Theory .................. 62 Conclusions................................................... 67 References...................................................... 68 Appendices...................................................... 7 6 A. Category Variables ................................. 77 B. Judged Variables .................................... 8 0 C. Predicted Affect of Categorical and Judged Variables Upon Effect. Sizes According to Theories........................................... 8 3 List of Tables Table 1. Correlations Between Category of Judges. . . . 2. Table of Means and Variances for- Each Judged Variable............................................ 3. Skew and Kurtosis Levels of Transformed and IJntransformed Ratings.......................... 4. Comparison of Effect Sizes According to Levels of Judged Variables for Each Technique. . . V List of Figures Figure Page 1. Effect Size As a Function of Time Between the 0 r i g i n a 1 S u b s e q u e n t S t i m u i u s.................... 5 0 2. Effect of Judgments of Arousal Intensity in the Original Stimulus Sit.at.ion Upon Study Effect. S i z e s ....................................................53 3. Effects of Judgments of Arousal Valence in the Original Stimulus Situation Upon Study Effect 4. Effect of Judgments of the Original Stimulus Situation Upon Study Effect Sites ............. 1 Introduction A number of researchers have examined the relationship between negative and positive emotions, such as fear and attraction, anger and love. According to Schacter & Singer (1962), emotions involve a cognitive or situacionai asses sment of the arousing stimulus. While Schacter's theory of emotion has not been universally supported, studies that examine the possibility that a misattribution of arcusal from one situation or stimulus can increase the emotional reaction to a separate situation or stimulus have generated more robust findings. "Arousal developed within one situ ation may be misattributed to an element which did not stimulate the emotional response. Or, arousal induced in one situation (and correctly or incorrectly attributed to various elements within that situation) may be carried into another situation to influence the emotional response to an element in the second situation. Finally, arousal induced by circumstances irrelevant to emotion (e.g., exercise, temperature adaptation) may contribute to emotional experience" (Dienstbier, 1978, p. 256) . Many researchers strongly supported the idea that arousal from one stimulus could be misactributed to a sec ond arousal stimulus. However, a number of other researchers failed to find support for the misattribution effect and some theorists proposed alternative explanations for those studies that did find a misattribution effect. The purpose of this paper is to examine those variables proposed in the misattribution and alternative theories. It will also attempt to determine whether these variables are significant moderators of the misattribution of arousal to a separate stimulus situation. It will attempt to de termine if these variables may be responsible for the failure of some studies to support the misattribution theory. Misattribution Theories Two groups of researchers simultaneously explored the misattribution effect. Researchers interested in romantic love (Tutton & Aron, 1974, 1989/ Waister & 3erscheid, 1971; and white i Kight, 1984) examined how misattributed arousal could increase liking and researchers examining the concept of excitation transfer (Donnerstein & Hallam, 1978; Donnerstin & 3arrett, 1978; Zillmann, 1971; Zillmann, 3ryant, Ccminsky & Meaoff, 1981) considered how it could increase dislike and aggressive behavior. The Excitation Transfer researchers and the Romantic Love researchers both examined the strength of the arousal created by the origi nal situation, the salience of the original stimulus, and the amount of time between the original stimulus and their emotional reaction to the target individual. Passionate Love Theory The source of love has been of considerable interest to scholars as well as poets for hundreds of years. In their search for the determinants of love and attraction, a consistent theme has been the link between passionate love and arousing, or even aversive situations. Following the tenets of Schacter's general theory of emotion, Walster and Berschied (1971) proposed that passionate love would be experienced whenever individuals were physiologically aroused, and the appropriate cues existed to indicate that the arousal was due to feelings of love (Berscheid & Walster, 1974). A number of experimental studies have supported this theory (Dutton & Aron, 1974; White & Kight, 1984). In these studies, subjects were found to be more attracted to an attractive female after being exposed to sexually arousing material (Stephen et al., 1971), or to fearful circumstances (Brehm, Gatz, Goethals, & McCrimmon, 1967; Dienstbier, 1979; Dutton & Aron, 1974, 1984). The idea that the misattribution of arousal can actu ally create passionate love has been criticized (Zillmann, 1984). Even 3erscheid and Walster (1974) are more likely to use terms such as "enhance" rather than "create"; and 4 they limit their discussions and predictions about the ef fect of naisattributed arousal to the more ephemeral pas sionate love, rather than the longer lasting companionate love. Critics observed that those studies which found in creased attraction, only did so in situations where the individuals would nave generated attraction anyway. In some studies, the experimental manipulations of arousal did not increase, or sometimes even decreased attraction to unattractive or same sex individuals (White, 1979) . This critical observation, however, does not weaken the support for misattribution of arousal theories. Arousal when mis- attributed to another stimulus or situation, will enhance or create the emotion appropriate to that situation. When the situation involves an unattractive person, then the subject is more likely to mistake their aroused state as aversion. Likewise "the maleness or femaleness of a stimu lus individual will influence greatly what emotional states other individuals are likely to attribute to the presence of that stimulus individual" (Dutton & Aron, 1974, p. 258). For most individuals, the most common interpretation of increased physiological arousal when in the presence of a member of the same sex is not one of increased attraction, but of dislike. On Che other hand, there is a strong association between arousal and an attractive member of the opposite sex (White & Kight, 1984, p.52). The relative salience of the arousing stimuli is an important factor within misattribution theories. The awareness of a prior causal connection or relationship be tween a stimulus and arousal, will increase the salience of that stimulus and increase the likelihood that it will be perceived as the source of the arousal. Individuals have well-defined ideas about why they might experience in creases in arousal during social interactions. However, "when the true source of one's arousal is salient, attri bution to plausible alternatives is unlikely, even if they are salient" (White & Kight, 1984, p. 56). It is this variable, the salience of the original arousal source, therefore which has the greatest influence over the misat tribution process. Failures to replicate the misattribution effect have been interpreted by supporters of the misattribution para digm as due to high levels of salience for the original arousal source. White, Fishbem, S Rustein (1981) found that misattribution of arousal from one stimulus situation to another occurred only when cues associated with the initial arousal were minimized. Other studies have found that attraction to the target individuals increased when 6 che cues associated with the initial arousal were made more salient (White, et al., 1981; White & Kight, 1984) . All of the studies reviewed in this analysis contain individuals to whom the experimentally induced arousal may be misattributed. A number of studies, even prime this relationship, by informing the subjects that they will have an opportunity to go on a date with this individual or by having the target individual provoke the subject. Excitation Transfer Theory The Excitation Transfer paradigm applies a three fac tor theory of emotion to the misattribution of arousal from one stimulus to a subsequent or secondary stimulus. The three components of this emotion theory are the disposi tional factor (skeletal motor responses), the excitatory factor (heightened activity in the sympathetic nervous system), and the experiential factor (cognitive evaluation of the stimulus and modifier of the reactions of the stimulus). This three factor theory is based upon the general emotion theory proposed by Schacter and Singer (1962) . Changes in a stimulus situation or a new stimulus will evoke quick changes in the dispositional and experiential factors, but excitatory adjustment or termination is slow and inefficient. Therefore, the residues of excitation 7 caused by one stimulus may enter into subsequent emotional reactions (Zillmann 1984/ Zillmann, Hoyt & Day, 1974). This transfer of excitation can occur because people are largely inaccurate in their assessments of their physiological state. This inaccuracy is due to a variety of reasons. One is that individuals do not continually account for their excitatory activity. However, even when subjects are asked to monitor their arousal, they are inaccurate. In Cantor, Ziliman, and Bryant (1975), che subject's judgment; of che cessation of their arousal oc curred on average a minute before their actual return to their physiological base rate. A third reason for inaccu racy, is chat people are incapable of discriminating be tween two separate excitatory reactions. Somatic feedback systems are not set up so as to allow individuals to appor tion the correct percentage of their excitatory r.eactions to the correct stimulus. Consequently, "individuals tend to ascribe their excitatory reaction in toto one specific, though potentially complex, inducing condition" (Zillmann, 1984, p. 151). Excitation Transfer Theory was initially developed to explain misattribution in successive stimulus situations. However, it is also applicable co transfer between simul taneous sources of excitation. Eor excitation transfer to 8 occur between subsequent sources of excitation, the two situations must be close enough in time that excitation still exists, but far enough apart so that the individual has lost awareness of the excitation. For simultaneous or concurrent excitation transfer, subjects should be aware of only one of the stimulus situations as a conceivable source for their excitatory reactions. In 1987, Baron found that subjects bombarded with positive ions experienced increased levels of arousal. Because they were unaware of the source of their arousal, these subjects attributed their increased arousal to the individuals with whom they were interacting, and so reported increased feelings of attraction and dis like. A reaction to another individual was a mere conceiv able cause of their arousal than the idea that they were under the influence of some invisible force. The facilitating affects of arousal were increased when Donnerstein, Donnerstein and Svans (1975) changed the basic experimental paradigm for excitation transfer re search. In the original paradigm, subjects were initially provoked, then exposed to the arousing and potentially distracting material. They were then reminded of their anger and were given an opportunity to retaliate toward the initial provoker. In the altered paradigm, subjects were first exposed to the arousing material and then provoked. 9 This second paradigm produced stronger facilitation effects because it did not rely on the "cognitive reinstatement of annoyance" (Zillmann, 1984, p. 153). A number of theorists have proposed that there is a specific link between sexual arousal and aggression. A number of studies have shown that, when provoked subjects were exposed to sexually arousing material, the level of their subsequent aggression against their tormentor was increased (Fisher & Harris, 1976; Jaffa, Malamuth, Feingold & Feshbach, 1974; Meyer 1972; Rosene, 1971; Sapolsky, 1977; Zillman, et al. , 1974 ; Zillmann, 1984) . According to the Excitation Transfer theory, however, this effect is not a function of the substantive effects of nudity or sexuality upon aggression, but is due to the arousing nature of such material. Meyer (1972) and Rosene (1972) found similar increases in aggressive behavior for an arousing violent film as they did for a sexually arousing film. Arousal from other sources such as physical exercise, noise, humor or an eye operation were also found to produce a similar facilitation effect (Donnerstein & Wilson, 1975; Mueller & Donnerstein, 1977; Zillmann, Bryant, Comiskv, £ Medoff, 1981; Zillmann, Katcher, & Milavsky, 1972) . According to Zillmann, these "findings challenge the view that the ef fect of erotica is mediated by a specific sex-aggression 10 incompatibility or by the generalization of disinhibition from sex to aggression" (Zillmann et al., 1981, p. 249). The theory that there is a specific link between sex ual arousal and aggression, is further invalidated by the fact that misattribution does not occur unless there is a relevant stimulus to which it may be transferred. Subjects who are not provoked, show no increases in aggressive be havior in response to sexually arousing material (3aron, 1979; Cantor, Zillmann, & Einsiedel, 1978; Donnerstein & Barrett, 1973). Furthermore, the enhancement effects of sexually arousing material are not limited to aggression. If subjects are exposed to sexually arousing material when predisposed to act in a positive manners then there is an increase in their positive behavior (Mueller & Donnerstein, 1981; Mueller & Donnerstein, 1977; Tannenbaum & Zillmann, 1971) . The salience of the arousal inducing stimuli is an important factor in the misattribution of arousal, accord ing to the Excitation Transfer Theory. Zillmann (1984) states that if "individuals are not provided with obtrusive intero and/or exteroceptive cues that link their excitatory state to a prior stimulation, residues of excitation from prior stimulation will inseparably combine with the excita tion reaction to the present stimulus and thereby intensify 11 both emotional behavior and emotional experience" (p. 153). Zillmann also pointed out chat abrupt changes from one stimulus source to a second may prevent misattribucion as the abrupt cessation of a stimuli, might make the individ ual highly aware of its residual effects upon their excita tory state. Salience's role in concurrent excitatory reactions is not discussed by Zillmann, however, it is an especially important factor in this type of misattribution, as it is only the relative salience of the two stimuli which determine to which the arousal is attributed. As mentioned earlier, time is also an important factor in the excitation transfer process. Aroused individuals perceive their arousal as dissipating sooner than it does in reality. In Cantor et al. (1975), experimenters deter mined that "arousal was perceived to be elevated, on txhe average for somewhat less than 5 minutes, but actually re mained elevated for over 8 minutes" (p. 301). According to the Excitation Transfer Theory, "transfer effects occur only if residual arousal from a prior activity is present and the individual has lost track of these residues" (Cantor, et al., 1975, p. 312). Therefore, excitation transfer would only occur when the secondary arousal source occurs (on average) between 5 and 3 minutes after the original arousal source. The sole exception would be cases of simultaneous presentation of the two arousal sources, in which subjects would lack temporal and ocher salience cues to identify a sole source for their arousal. In summary, this theory predicts: Predictions of Misattribution Theories 1. Increases in attraction or aversion will occur only when subjecus are aroused by the original stimulus. 2. Increases in attraction or aversion will occur in response to arousal from any source (not just erotic or aggressive material). 3. Misattribution of arousal will not occur between two concurrent situations, if the time between the two situations is too short for subjects to lose their aware ness of the original arousal source, or if it is too long and the original arousal dissipates. 4. Cues that overtly remind subjects of the original source of arousal will decrease the misattribution effect. 5. Arousal may be misattributed to feelings of liking or disliking, and may enhance aggression or attraction. Arousal and Attentional Shift Model The Arousal or Attentional Shift Model (Donnerstein, Donnerstein, \ Evans, 1975) is actually a modified version of the Excitation Transfer Theory. To explain those stud ies that found decreases in aggression after subjects were 13 erotically aroused, Donnerstein et al. (1975) theorized that erotic material had two effects upon individuals. It produced arousal and it was attentionaily diverting. "Various stimuli such as erotic, aggressive, or humorous presentations have some measure of attentional shift which diverts a subject's attention away from prior anger insti gation" (Mueller & Donnerstein, 1977, p. 74). Residue of arousal produced by erotic material enhances subsequent aggressive behavior. However, if the material does not produce enough arousal, then the attentionaily diverting aspects of the arousing material will divert the individ ual's attention from memories of their prior experience with the target individual and away from the emotions en gendered by that encounter (Donnerstein, Donnerstein, & Barrett, 1976; Zillmann & Johnson, 1973). Therefore, when subjects come into contact with these target individuals, they do not experience the same degree of feeling they would have normally. In support of their model, Donnerstein et al. (1975) found that while subjects who saw photographs of sexual material did exhibit less aggression than those subjects who viewed neutral subjects, pictures of explicit sexual behavior decreased the level of aggressive behavior less than did mild erotica. This suggests that more explicit and filmed sexual material should produce enough arousal to outweigh the material's distracting aspects, and thus, in crease aggression levels. A later examination of chose studies that found increases in aggression following expo sure to erotica, revealed that those studies used more arousing (more explicit) erotica then that used in those studies that did not find increased aggression (Donnerstein, Lint, & Penrod, 1987). Another variable affecting attentional shift and its impact on the misattribution of emotions, is whether indi viduals are primed to like or dislike the target prior to the presentation of the arousing stimulus. Attentional shift diverts the subjects attention from prior experiences and feelings. If there are no prior experiences, then at tentional shift will not impact the subject. Therefore, this model would predict that decreases in liking or dis liking will only occur within the original excitation transfer paradigm, in which subjects were primed to like or dislike the target, were exposed to the arousing stimulus, and then were placed in a situation that reminded them of their earlier emotional experience. Predictions of Arousal and Attentional Shift Model 1. All of the predictions of the Misattribution The ory will be supported. 2. There will be an interaction between the level of arousal and the attentionaily diverting potential of the original arousal stimulus, such that the reactions to the second stimulus will be reduced when the original material is attentionaily diverting and produces lew arousal. 3. Reactions to the second stimulus will be increased when the stimulus is not diverting and/or the arousal pro duced is high. 4. The predicted decreases in feeling due to atten tional shift will only occur within these studies, in which the subjects were primed to like or dislike the target be fore the arousing stimulus. Arousal and Hedonic Valence Model Some modification had to be made to the Excitation Transfer Theory, to account for the fact that material that produced negative affect would increase aggressive behavior and feelings of dislike, while material that produced positive affect would decrease these feelings and behavior (Sapolsky & Zillmann, 1931). This effect should be visi ble, however, only when the arousal level is low. At high levels of arousal, these valence effects are suppressed by the stronger effects of the arousal. There seem to be no specific predictions by the model as to the effaces of negatively and positively valenced m 1 o arousal upon positive feelings and behaviors. These ef fects should be the same as those mentioned above; posi tively valenced arousal enhancing feelings of attraction and negatively valenced arousal decreasing these feelings of attraction. However, even past studies are divided upon this issue. Rickwooa and Price's (1983) experiment found excitation transfer effects only when the prior and subse quent arousal were similar. However, other studies found no differences between the misattribution effects of simi lar and dissimilarly valenced arousal producing situations or stimuli (Cantor, et al., 1975; White et al., 1981) . Predictions of the Arousal and Hedonic Valence Model 1. All of predictions for the Misattribution Theory will be supported. 2. If the original situation produces positively va lenced arousal and then aggressive responses or other indications of dislike will be diminished. 3. If the original situation produces negatively va lenced arousal, then aggressive responses or other indica tions of dislike will be enhanced. 4. The effects of the arousal valence produced by the original arousal source, will only be apparent when this arousal level is low or moderate. 17 Reinforcement Theory The Reinforcement Theory (Kenrick, Cialdini, s Linder, 1979) has been proposed as an alternative to the misattri- bution explanation of how aversive situations can lead to increased attraction. Its proponents focused their efforts on reinterpreting the results of the passionate love para digm experiments performed by Dutton and Aron (1974) and 3erscheid and Waister (1974), but they did not attempt to account for the results of other misattribution theories and experiments. The Reinforcement Theory claims that arousal does lead to increased attraction, but that the attraction is a re sult of the dissipation of the arousal. The presence of another individual is reassuring, and therefore the dissi pation of arousal due to their presence causes us to like them (Kenrick, Cialdini, i Linder, 1979). According to Kenrick and Johnson (1979), the presence of another individual would have no reassuring effect in the absence of an aversive situation, therefore their at tractiveness should not increase. Also, because it is the presence of the other individual which causes the arousal dissipation, the gender of the other individual should not affect the results. 13 Predictions of the Reinforcement Theory 1. Increases in attraction should not occur in re sponse to positively valenced or neutral arousal. 2. The reassuring, calming aspects of the second stimulus situation should increase liking. 3. The gender of the target individual should not effect the feelings for the target individual. Incompatible Response Model The Incompatible Response Model states that it is the compatibility or incompatibility of hedonic valences, which produce the enhanced attraction or dislike for target individuals. According to 3aron, "If the valence of the second emotions state is opposite to the disposition pro duced by the first state, then the two valences should counterbalance each other, resulting in a less intense re sponse to the second stimulus " (Branscombe, 1985, p. 157). This is because "people are incapable of engaging in two incompatible responses at once" (Berkowitz, 1974, p. 261). On the other hand, if the two situations or stimuli produce similar valences then the reaction to the second stimulus should be enhanced. The enhancing or diminishing effects of the valences are limited to those stimuli which occur close together. "As the time delay between the two states increases, the valence of the first state should presumably exert a smaller effect on the second state" (Branscombe, 1985, p. 157). This is in contrast to the misattribution theo ries, which predict that increases in attraction or dislik for a target individual will only occur after a shore de lay, which successfully reduces the subject's awareness of the earlier situation (except in those cases where the earlier situation is not salient for other reasons). The Incompatible Response Model does not include arousal as a factor in the enhancement process, and there fore would not predict that the effects of valence would vary according to the level of arousal created by the original stimulus. This is in contrast to the Hedonic Va lence Model, which predicts that the effects of valence upon the enhancement of liking or disliking will only be visible when arousal is low, because the effect will be washed out when arousal is high. Predictions of the Incompatible Response Model 1. If the valences of the arousal created by the original and subsequent stimulus situations are incompati ble, then the response to the second stimulus will be diminished. 2. If the valences of the two situations are similar, then the response to the second stimulus will be enhanced. 20 4. . The effects of valence will not be affected by the level of arousal created by the initial situation; (d) the enhancing effects of compatibly vale.nced arousal upon the second stimulus will be strongest when there is little or no delay between the two stimulus situations. Sex/Aggression Link Theories A number of different theories and models are based upon a hypothesized link between sex and aggression (Cofer & Apley, 1964; Stoller, 1975). The drive diffusion chec- ries, for example, believe that the two emotions share a common origin, and therefore, increases in one emotion fuel increases in the ocher. Barclay and Haber (1965) found thac angered students produced more aggressive and sexual imagery than nonangered students. Clark (1952) found chat sexually aroused individuals produced more aggressive im agery than nonsexually aroused individuals. Other theo rists believe that the increases in aggression observed in sexually aroused individuals are due to the disinhibiting effects of the erotic material used to create the sexual arousal (Heshbach & Malamuth, 1978). These theories are in direct contrast with the excication transfer theory which states there is no spe cific link between sexual and aggressive drives, and thac 21 there are no differences between the arousal created by erotic means and that created by exercise and humor. Predictions of Sex-Aggression Link Models 1. Subjects will only experienced enhanced feelings for the target individual after being aroused by erotic or aversive stimuli. 2. Arousal created by humor or exercise will not en hance feelings of attraction or dislike. Comparisons Between Theoretical Predictions A number of key variables differentiate between the theories proposed in this meta-analysis (Appendix A). The source of the arousal experienced in the original stimulus situation is the main differentiation between the Sex/ Aggression Link theories and the other theories within this meta-analysis. In contrast to other theories, the Sex/ Aggression Link theories argue that arousal from humor or physical exercise will not enhance feelings toward the target individual. The predictions regarding attentionally diverting as pects of the original arousal source are the main differ entiators between the Arousal and Attentional Shift Model and the other theories within this meta-analysis. When the intensity of the original arousal source is low and sub jects are primed to either like or dislike the target 22 individual, then increases in the actentionaliv divert- ing aspects of the original stimulus will decrease the re sponses to the second stimulus. The other theories do not predict this interaction. The predictions regarding the calming effects of the second arousal source differentiate the Reinforcement The ory from the other theories in this meta-analysis. The Reinforcement Theory predicts that feelings for the target individuals will be enhanced to the degree that the second stimulus has a calming effect. The other theories do not predict that this variable will have any effects. The predictions regarding the valences of the original and secondary arousal sources differentiate the Incompati ble Response Theory from the Hedonic Valence Model and the other theories. If the valences of the two stimuli are incompatible, then the response to the target individual will be decreased. If the valences are similar, then the responses to the target individual will be increased. The Hedonic Valence Model predicts that increases in the posi tive valence of low intensity arousal will decrease dislike for the target individual. Overview of Meta-Analvtic Procedures Meta-analvsis statistically combines the findings of a number of studies on a common topic. The findings of these 23 separate studies are transformed into a statistic known as "effect size," which may be compared and combined across the studies. The effects of a number of variables on these "effect sizes" may be determined through normal statistical procedures. Method In this meta-analysis, I collected published studies on the effects of misattributed arousal upon the liking/ disliking for a neutral individual. Each of these studies, manipulated the degree of arousal experienced by the sub jects and compared their attraction toward an individual. This comparison between high and low arousal states was used to calculate the effect sizes. The studies were then coded on a number of objective and subjective theoretical variables, and analyses was performed to determine if there were any relationships between these variables and the strength and direction of the "effect sizes." The majority of the variables used within this meta analysis were not actively manipulated or measured within the original studies. To allow these theoretically impor tant variables to be examined within this analysis, infer ential judgments were performed by independent raters. 24 Literature Search Procedures The studies for this analysis were collected by a number of methods. First, a literature search of studies published from 1955 to 1990 was conducted on the computer ised version of the Psychological Abstracts. The terms used for this search were: (a) aggression, (b) arousal, (c) sexual arousal, (d) attraction, (e) misattribution, and (f) excitation transfer. Further, the names of Aron, Saron, Berschied, Cialdini, Donnerstein, Dutton, Kenrick, Walster and Zillmann, were used as additional search terms, as these individuals were strongly associated with research on the misattribution phenomena. Second, the volumes of Journal of Personality and Social Psychology, Journal of Experimental Social Psychology, European Journal of Social Psychology, and Personality and Social Psychology 3ulletin for the years 1960 to 1991 were manually reviewed for studies missed by the computerized literature search. Third, the citations index was searched for citations of each relevant study and the reference sections of these studies were searched for prior studies. Fourth, this process was repeated for review articles on the subjects of misattribution and excitation transfer. 25 Inclusionary Criteria To be included within this meta-analysis, studies had to meet the following requirements: (a) They must measure attraction (liking/disliking) toward a human target, (b) they must contain two stimulus situations; the original arousal inducing situation and a second simultaneous or subsequent stimulus situation, and (c) they must manipulate the level of arousal experienced by the subject. Forty-nine studies conformed to this paradigm, con taining 111 separate comparisons. The studies contain on the average, two separate comparisons of the arousal conditions and some contained as many as four. These con ditions were treated as separate and distinct observations if they differed along the dimensions of arousal valence (e.g., positive, negative, neutral), type of arousal induction (e.g., erotic, exercise, etc.), salience (many cues or few cues to original arousal source), and whether provoked or not. If two conditions failed to differ along these dimensions, then they were combined and treated as a single condition. Duplicate comparisons of conditions were conducted for those studies that included more than one dependent variable. When a study used several measures of attraction, separate effect sites were calculated for each dependent measure (e.g., behavior measures vs. questionnaires). There was no variation between effect sires based on type of dependent measure. Therefore, ef fect sires were averaged across the types of dependent measures to form one composite effect size for each com parison. This was accomplished by calculating the mean of the effect sires. The significance level of each effect sire were also averaged. This procedure involved, trans forming effect sire probabilities into Z scores, computing the mean Z Score, and then translating this mean Z back into a probability measure. The effect sires for dependent measures which were not administered at the end of the ex perimental procedures with other dependent variables, but were measured a significant amount of time later, were not computed and averaged into the mean effect sire. This policy was implemented because the focus of this study is on the impact of unrelated arousal upon the reaction to a second stimuli, and not on the memory of feelings or about how attitudes may change. Variables Coded From Each Study The following variables were recorded from each study (Appendix B). Values for these variables were determined according to information that was reported in each study and thus could be objectively coded by the author of this paper. These variables included: (a) type of experimental 9 7 paradigm (e.g., excitation transfer paradigm, the modified transfer, and the simultaneous paradigm); (b) gender of subjects (e.g., male, female, mixed); (c) number of sub jects within experiment; (a) gender of subjects and target (same, different); (e) type of dependent variable (e.g., questionnaire, positive behavior); (f) study orientation (e.g., increased attraction, increased dislike and aggres sive behavior); (g) directionality of dependent measure (e.g., increases in effect size indicate an increase in attraction, decreases indicate an increase in dislike); (h) type of original arousal stimulus (e.g., erotica, exercise, anxiety); and (i) the length of time between the original and second stimuli (e.g., 0 sec., 0-2 min., 2-4 min., 4-5 min., 6-8 min., > 3 min.). Variables Judged From Each Study Seven variables are included in this meta-analysis that had not been manipulated nor measured in the studies, but were potential moderating variables (Appendix C). They were obtained by having four raters judge their levels af ter reading each experimental study. These variables in cluded: (a) strength of arousal produced by the original stimulus, (b) the valence of the arousal produced by the original stimulus, (c) the number and strength of the cues in the second stimulus situation which would remind the 2 3 subject of the original stimulus, (d) the strength of she arousal produced by the secondary stimulus, (e) the valence of the arousal produced by secondary stimulus, (f) the po tential of the original stimulus to be attentionally di verting, and (g) the arousing or calming effects of the second stimulus situation. Appendix C presents the definition of each variable; and rating scales for each, with examples of the various values. The rating procedures and material were adapted from those used previously in an examination of displaced ag gression. The method sections of each study were removed, and all indications of author and title were removed. Next brackets were used to indicate the sections of the proce dure of interest to the judges (e.g., priming period, presentation or the original arousal stimulus, presentation of the secondary arousal stimulus). Then, sections detailing differing procedures for conditions were high lighted in different colors. Finally, each theoretical variable was judged independently by four (two males, two females) social psychology graduate students. Judges were trained in the coding process before be ginning their ratings. The experimenter explained how the judges should approach the studies, and identified the im portant sections of the method section. They also 29 discussed the definitions for each variable to be judged, and practiced rating hypothetical situations. Judges were given rating sheets, the method sections, variable defini tions and rating instructions for each variable. The judges rated all the studies for a single variable, before beginning the next variable ratings. The order in which the variables were rated was randomly determined for each judge, as was the order in which studies were examined for each variable. The ratings for 43 comparisons were chosen to test for the correlation between the male and female raters. The correlations between the two male and the two female judges were high and significant, and so the ratings for one male and one female judge were included within chis analysis. (Correlation between the two male judges and between one two female judges for each judged variable, were significant at o < .001 level: Arousal Intensity of SI - M (r = .31), F (R= .32); Valence of 51 - M (r = .92), F (r = .31); Salience - M (r = .32), F (r = .39); .Arousal Intensity of 51 - M (r = .33), F (r = .31); Valence of SI - M (r = .92), F (r = .30); Calming Potential - M (r = .82), F (r = .39); Diverting Potential - M (r = .90), F (r = .83)). 3° Results Meta-Analytic Procedures The effect size for each study was calculated and then tested for significance. If significant variability ex isted between the effect sizes collected in this analysis, then various analyses were performed to determine the mod erating variables. These analyses involved categorical modeling (Hedges & Clkin, 1935), one-way analyses of vari ance, univariate regressions, and ANOVAs. Effect size calculations. Effect sizes were calcu lated using a statistical software program for meta- analysis called DSTAT (Johnson, 1990). For those studies which reported the means and standard deviations of the aroused and unaroused conditions, d was estimated by sub tracting the mean of the unaroused condition from that of the aroused and dividing the result by the pooled estimate of standard deviation. If a pooled standard deviation was provided within the study's results section, this was the SD used. This was the preferred method of calculation. If the standard deviation information was not reported for a study, then effect sizes were calculated form the individ ual cell means and the E, T, or X2 value using the formulas of Glass, McGaw, and Smith (1981). 31 Most studies within this analysis contributed more than one effect size due to their factorial design. For example, if a study crossed arousal level with prior pro vocation, then separate effect sizes were computed for the conditions where subjects were provoked and for the condi tions where the subjects were not provoked. Both effect sizes were included within these analyses. Control conditions were generally those in which no experimental manipulations were performed on the subjects before the measures. These control conditions were net included in the escimates of che effecz sizes if zhey ex isted in a multivariate design. In these designs, zhe control condition differed from each experimental condition along other dimensions in addition to the manipulation of arousal. Therefore, any effect size estimates could be due to the effect of those ocher variables. In most cases, this issue is moot because che control conditions are not included in the reported analyses. Effect sizes are com puted between the low and high arousal conditions. Control condition are used however in the cases of one way analy ses. In these designs, the control conditions cannot vary along a second dimensions that is distinct from the experi mental conditions. 32 Analysis of effect: sizes. The estimated effect-sizes were combined to a mean effect size. This effect size was weighted by sample size. The number of subjects was also included within the analysis as a moderator variable. This variable did not correlate with effect size (corr. = .0705, d > .05). Therefore, unweighted analyses of effect size were not conducted. The mean effect size was then tested for significance. A test of homogeneity (Qw) was performed to test for vari ability between the studies. A lack of homogeneity would indicate the possible existence of moderator variables that cause the difference between the studies (Hedges i Olkin, 1985). Categorical variables analysis. Categorical modeling was used to test for variability in effecc sizes between studies which differ on the various categorical variables, using the Qb statistic (Hedges & Olkin, 1985). This is similar to the F statistic in an Analysis of Variance. In addition to the Qb statistic, the Qw statistic was also computed for each level of the categorical variable. A significant Qw statistic would indicate the existence of variance among the studies within the category. Analysis of judged variables. Analysis of the judged variables showed on.lv moderate correlations between the 33 male and female judges. The correlation between the male and female judges ratings for the intensity of the arousal produced by the second stimulus was very low (r = .03) . Although, correlations between the male and female judges for other judged variables were significant at the .01 level, none of the correlations exceeded r = .71 (Table 1) . This was not a completely unexpected finding as many of the studies deaic with pornographic stimuli and aggressive situations. Some researchers (Kinsey, Pomeroy, & Martin, 1943) have found differences between male and female reac tions to erotica. Because of their low levels of correlation, averaging the male and female judgements might have distorted the analyses of the effect sizes. Therefore, the judgements used for each variable were the male judges ratings when the study's subjects were male, and the female judge's ratings when the subjects were female. The average value of male and female judgements were used for those few studies where subjects were of mixed gender (Table 2) . These judgements were tested for normality. Some variables were found to have high levels of skewness. Therefore, the judgements were transformed to normalize the distribution using the transformation formulas suggested by Tabachnik and Fidel (1939) . Bach variable was transformed using the 34 Table 1 Correlations Secwesn Categories of Judges Groups Judged Variables Ms1e/Ma1e Female/Female Mai 0 / F* 0male r (n) r (n) r {n) Intensity of SI Arousal . 32 (45) . 32(45)** .45(111 ) * Valence of Si Arousal .92(45)** . 81 (45)+* .71 tin ) + Valence of S2 Arousal .j2 (45)** .80(45)** . 03 (in ) * Salience of 51 Arousal .32(45) .89(45) <_n CO tin ) * Calming effects of S2 . 32(45) II .89(45)** .25 (in ) * Attentional Divert ing Effects of 52 .90(45)*' .83(45) * * . 62 (in ) * ^Indicates correlations of .01 level ++Indicates correlations at the .001 level Table 2 Table of Means and Variances for Each Judged Variable According to Various Techniques Male Female Pooled Assigned Transformed Judge Variables Mean Mean Mean Mean* Mean** Intensity of SI Arousal 2.41 ( .92) 2 . 32 ( .68) 2 .36 ( .57) 2 . 31 ( -69) 2 . 31 ( -69) Valence of SI Arousal .16 (3.62) . 15 (3,63) .15 (3.11) .24 (3.55) .51 ( .06) Intensity of S2 Arousal 1 . 65 ( .91) 1 . 93 (1.01) 1 .79 ( .47) 1 . 91 (1.05) 1 . 91 (1 .05) Valence of S2 Arousal - .80 (1.81) - . 90 (1.73) - . 08 (1.25) - . 90 (1.59) 1 .73 ( .12) Salience of SI Arousal - .80 (1 -35) 1 .23 (1.30)' 1 . 02 (1.04) 1 .24 (1 .29) .29 ( .05) Calming Effects of S2 .36 (1.35) .44 (2.36) .40 (1.15) . 52 (2.15) . 52 (2.15) Attentionally Diverting Effects of S2 2 . 87 (1.81) 2 .81 (1.44) 2.84 (1.32) 2.74 (1.48) 2.74 (1.48) ‘Technique in which gender of subject and judge must correspond. ‘‘Technique in which assigned variables are transformed in order to reduce. u> Ur formula for mildly and moderately skewed values. The for mulas for the transformation of positively skewed distribu tions are New Value = SQRT (Old Value) for mildly skewed variables, and New Value = SQRT (K - Old Value) for moder ately skewed variables. The formulas for the transforma tion of negatively skewed distributions are new Value = Log 10 (Old Value + C) for mildly skewed variables and New Value = Log 10 (K - Old Value)- for moderately skewed vari ables. In these formulas, C is a constant adaedto each value so that the smallest value is one; and K is a con stant from which each value is subtracted so that the smallest value is one. Table 3 shows the amount of skew for each variable after the mild and moderate transforma tions and after no transformations. This table was used to choose which of the transformed or untransformed values for each variable would be used for subsequent analyses, based on which value had the least amount of skew. Univariate regression analyses were con ducted on these transformed values for the seven judged variables using the statistical software package SPSS-PC. These analyses were conducted to test the predictions re garding the intensity, valence, salience, and attentionaliy diverting effects of the arousal produced by the original stimulus, ana the calming and/or exciting effects of the T a t> 1 e 3 Skew and Kurtosis Levels of Transformed and Untransformed Ratings Arousal Level SI Skew Kurtosis Valence S2 Skew K u r L o s i s Sa1i ence Skew Kurtosis Untrans formed Rating .13570 .81350 • .35310 1.2475 . 3977 1 .2467 Mildly Transformed Rathings . 3056 .4383 . 3960 .8046 .1355 1.6648 Highly Transformed Ratings .7641 . 3062 . 0960 . 0087 ■1 . 501 0 OJ Table 3. (continued) Untrans formed Rating Mildly Transformed Ratinys Highly Transformed Ra t i ngs Arousal Level S2 Skew - .02100 -].3973 -1.7641 Kurtos i s -1.41250 -2 .3107 3.157 6 Valence S2 S kew .06279 .18 58 . 5357 Kurtosis - .63330 -1.7 692 -1 . 0570 Relax Skew - .0714 - .8919 -1.3968 Kurtosis -1 .0737 .2786 2.1992 Attention Skew . 3052 -1.4 690 -1.2820 Kutosis - .5317 5.84 30 4.52 4 3 u> oo 39 second stimulus source duplicate regression analyses were conducted separately upon the male and female judgements, as well as the pooled value judgements. Because the SPSS PC program does not use the correct degrees of freedom for an analysis of effect sizes, the results of these regres sions were entered into a D-Stat program which adjusted significance levels by correcting the standard error values. One-way and multivariate ANOVAs were also performed to further test for predicted relationships within, and among the categorical and judged variables. Duplicates of these analyses were also performed for the sole male and female judgements, and the pooled male/female judgements (Table 4). Results for Overall Effect Site Effect size distribution. Preliminary analyses were conducted to examine the distribution of effect-size esti mates for extreme outliers. These extreme values may be excluded or modified, as they can have severe effects upon analyses results (Tabachnik & Fidel, 1989) . One extreme outlier was identified (d = -2.79). Upon examination, it was determined that this extreme value came from a study which was the sole user of negative ions as a source of arousal. The effect size produced by this study, Baron Table 4 Comparison of Effect Sizes According to Levels of Judged Variables for Each Technique Intensity of SI Arousal Neutral Mild Moderate Strong Very Strong Ass igned -.2088“ - .1370“ .2186“ - .01860“ Male -.0634“ -.1878“ -.1077“ .2055“ .04440“ Female - . 7 7 8 9C - .2368“ - . 097 9" . 1741° . 04 7 50- Pooled - . 3 019A - . 1841A - . 0623A . 18 59A . 04 S95A Intensity of S2 Arousal Ass igned -.2244“ -.0333“ - .2249“ - . 0099“ - . 39040“ Male -.1242“ . 0573“ -.0622“ . 1427“ - . 36370“ Female -.397 7 A - . 04 80A . 2 7 5 5A - . 017 8a - . 39040A Pooled - .1622“ -.0962 . 1 31 3V -.0587“ -.37705“ ■£> o Table 4. (continued) Valence of S1 Arousal Very Negative Negative Mildly Negative Mildly Neutral Positive Positive Very Positive Assigned . 3818" - . 233 6bt - . 3 32 6 ' " ' - . 4 64 9" - . 007AbJ - . 002Ab" - ,272 6Afc Male . 09b6b - . 3758f a - . Q433b - . 1746b . 0203b - . 1901b - . 1531b Fema1e .3 67 4' - . 2 2 0 7: - . 3545' -.57 31 - . 024 4bJ . 0061b; - . 3045Ab Poo1ed . 4 428Ab . 2 589Ab -. 1535Afc . 04 07 Ab . 07 58Ab . 2809b - . 2 17 5b Valence of S2 JUousal Assigned . 0316b . oo93b - . 10 3 21 - . 0506b -.1712e Mai e -.3 637f a . 0609b - .2153" -.1524b . 2708b Female .04 37a .0136" - .0592A . 18 4 4 Ab . 1182Ab Pooled - .1600- . 0301': - . 1061- - .0017- .0608: Table 4. (continued) Calming Effects of S2 Very Mildly Mildly Very Negative Negative Negative Neutral Positive Positive Positive Ass igned - .2 87 2* . 0*111'' . 1 2 60i ; - . 07 31t . 0 9 31t ; . 005 6b Male - . 014 6Ab - . 242 9b . 08 3 lAb . 0 7 6 7 Ab -. 1336Ab . 4399b Female - . 17 63” . 0719” - . 0415* -.0178A . 0135A .04 57a Pooled - .3 637- -.074 8 ' -.02 52- . 03 60: . 2 01 0: .2428' Table 4. (continued) Ass igned II a 1 e Fein a 1 e Pooled AlLentional Diverting Effects of D2 Not at All Mildly Moderately Extremely Completely Constantly Distracting Distracting Distracting Distracting Distracting Distracting - . 06991 1 - .0157b -.6191a - . 5003 ' ' . 0093“ -.3 027b - . 33 33" - . 0319 l ! - .0521L . 0 8 0 2b . 008 r - . 0980 b . 0105A . 02 8 9b - . 0135" - . 0691b .0 617a . 0769b .0877" . 1411b . 2712a -.0131“ . 2195" . llllb Note. Matching superscripts (a, b, c, d, and e) .indicate values that do not differ. 44 (1987) was significantly lower then the effect sizes pro duced by ether studies of arousal. Although Baron (1987) did find that the presence of negative ions produced arousal, it nay be that this arousal was accompanied by or was the result of some discomfort; much like burning a finger will produce arousal. It is not known exactly how negative ions produce arousal, so sub jects in this 'study may have been undergoing some active discomfort while experiencing the ion bombardment. Due to this ignorance regarding the actual effects of negative ions, it was decided to drop this study, Baron (1987), from the analysis. Dependency. More than one effect size was obtained for many of the comparisons between the control and experi mental conditions, due to the presence of more than one dependent variable. This may have affected the validity of the analyses. Therefore, the effect sizes were averaged across studies and comparisons. Mean overall effect size estimate. The overall anal ysis did not support the hypothesis that there is a rela tion between increased feelings of attraction or dislike toward another individual and unrelated arousal. The mean weighted effect size (m = + .0636) did not differ signifi cantly from zero (Cl = 0.0/+.13) . This means chat we are 4 5 unable to conclude (at this time) that when subjects are aroused by one stimulus, it will lead to an increase in the intensity of feelings in response to another stimulus (an other individual). Homogeneity. Although the summary analysis did not show an overall effect for arousal, this result must be viewed in light of the variability in effect sizes across the studies. The Homogeneity Test was significant (Qw = 350.8, d < .001) indicating a large magnitude of variabil ity within the studies. A review of the individual effect sizes reveals substantial variation in direction and magnitude. This variability was anticipated from the inconsis tency of results reported in narrative reviews of the lit erature (Donnerstein et al., 1987; Zilimann, 1984). Conflicting theories propose a number of moderator vari ables to account for these variations. These moderator variables will be examined to determine if they can account for these variations. Further examination of the effects of these variables shall also suggest which theory best accounts for the effects of misatzributed arousal on in creased attraction or dislike for another individual. 4 o Categorical Variables Appendix B defines the categorical variables and the individual levels for each. A few studies contain missing variable values, due to incomplete procedural descriptions. These studies were not included in analyses of these variables. Study orientation. Study orientation affected affect size. There was significant variation between the two categories of study orientation (QB = 10.83, p < .001) . Those studies oriented toward increased attraction found an increase in feelings for the individual (m = - .1857) . Those studies oriented toward increased aggression found a decrease in feelings toward the individual (m = -.0329), that were net significantly different than zero (CI_ = . 12/+ . 05) . Arousal source. The source of the arousal had a sig nificant impact upon the effect size estimates across studies (Qb = 21.05, p < .006). There was also significant variation within categories in which the arousal source was erotica (m = -.0497), exciting films (m = .6176), anxiety/ fear (m = .1008) (Qw = 129.8, Qw = 39.4, Qw = 54.27), re spectively; p < .001) and for exercise (m_= .2649), and humor (m = -.2040) (Qw = 27.53, Qw = 37.45), respectively; p < .02). Within category, variation also approached 47 significance for categories where the arousal source was aggressive films (m = .0624), and aversive noise (m = .0159), (Qw = 26.9, Qw = 5.29, respectively; o < .08). There was no variation within studies when the arousal source was due to an aversive film or dislike due to the presence of only one study. Significant within category variation (Qw) indicates that additional factors besides the source of the arousal contribute to the lack of homogeneity among the effect sizes. Paradigm type. The type of experimental paradigm did not affect the misattribution of arousal. There was no variation in effect sizes between those studies using the original excitation transfer paradigm, the modified excitation transfer paradigm, and the simultaneous paradigm (Qb = 2.35, o > .3). In contrast, the lack of homogeneity within each of the paradigm categories, reflects a signif icant amount of variability among the studies within those categories (Qw = 156.9, Qw = 133.2, Qw = 58.2, all with p > .001), indicating that other factors affect the effect sizes. Subject gender. The gender of experimental subjects did not affect experimental results. There was no signif icant variation between those studies that used male sub jects, those that used female subjects, and those studies 48 that used both (Qb = .902, o > .6) . By contrast, there was significant variation among the studies within each of the three gender categories (Qw = 250.9, d > .001; Qw = 22.3, D > .17; Qw = 55.59, o > .001). Target gender. The correspondence of the subject's gender tc that of the target individual did effect the change in the subject's feelings toward the target (Qb = 13.8, p < .001). Studies in which the target individual and the subject were of the same sex found a decrease in feelings for the target individual (m = -.1952) . Studies in which the target individual was of the opposite sex from the subject found an increase in feelings for the target individual (m = -.1885), !3 = 7.19, c <.00I). However, there was significant variation within the categories (Qw = 153.3, Qw = 94.39, both at o > .001), indicating the presence of other moderating variables. Priming. Whether subjects were primed to like, dis like, or not primed at all, did not affect effect size es timates. There was no variation between the three cate gories (Qb = 4.533, d < .09). However, there was signifi cant variation among the studies wichin each priming cate gory (Qw = 121.3 5, p < .001; Qw = 32.9, o > .003; Qw = 18 6.53, £ > .001). 49 Time between SI and S2. The amount of time between the two potential arousal sources did affect the effect size of a study. There was significant variation between the categories (Qb = 18.657, o < .002). The homogeneity test indicated that studies where the time between the cwo stimuli was over 3 minutes, were homogenous (Qw = 74.5, p < .001; Qw = 126.5, o > .001). Those categories where 51 and 52 were simultaneous, or the time between chem was un der 2 minutes did exhibit significant variation. Those studies where the time was between 2 and 8 minutes were also nonhomogenous (Qw = 57.05, p < .08; Qw = 23.39, o > .003, Qw =42.3, o > .04). This indicates the presence of ocher moderating variables influencing the effect sizes of studies within these categories (Figure 1). There are significant differences between the differ ent levels of the categorical variable. The effect size for simultaneous (Time 1=0) misattribution situations (E = +.2427) is significantly greater than the effect size for experiments where SI and 52 differed by less than one minute (E = -.1229). This effect size where SI and S2 differed by 1 minute or less was also different from those experiments where the time span was more than 1, but less than 2 minutes (E = -.0074), and significantly differenc from these studies where the time span was between 2 and 4 Figure 1. stimulus. .05 o <D m w -.25 Zero 0-2 2-4 4-6 6-8 8 Amount of Time (min.) Effect size as a function of time between the original and subsequent Ui o 51 minutes (E = +.2104), between 4 and 6 minutes (E = -p. 0657), and greater than 5 minutes (E =.0639). These relationships match the pattern proposed by the excitation transfer model. Judged Variables As described earlier, seven variables (arousal inten sity created by the original and secondary stimulus situa tions, valence of the arousal created by the original and secondary stimulus situation, salience of the original stimulus situation, the attenticnaily diverting potential of "he original stimulus, and the reassuring or energizing effects of the secondary stimulus situations)along continu ous dimensions. Multivariate regression analyses were conducted to determine whether these variables caused any of the variations among misattribution effects. Intensity of 51 arousal. The intensity of the arousal produced by the original stimulus had a significant posi tive impact upon effect size (Z = 5.98, o < .001) . Studies in which the level of arousal produced by the original stimuli was judged to be mild or moderate showed a decrease in feeling for the target individual (m = -.1729). Studies in which the level of arousal produced by the original stimuli was judged to be strong or very strong showed an 52 increase in feeling for the target individual (m= t . 1) (Figure 2). Valence of SI arousal. The valence of arousal pro duced by the original stimulus did have a significant im pact upon effect sizes (2 = 1.11, o < .3). Studies in which the valence of the arousal produced by the original stimuli was judged to be positive, showed an average de crease in feeling for the target individual (m = -.0938) . Studies in which the valence of arousal produced by the original stimuli was judged to be negative, showed an average increase in feeling for the target individual (m_= +.0942) (Figure 3) . Salience of SI arousal. Effect sizes varied signifi cantly as a function of judgments of the salience of the original arousal source. The salience of the original arousal stimulus had a positive impact on the size and di rection of effect size (Z = 2.10, p < .04). Studies in which the original stimuli was judged to be moderately or * very salient showed an average increase in feeling for the target individual (m = +.1059). Studies in which the original stimuli was judged to be mildly or not at all, salient showed an average decrease in feeling for the tar get individual (m = -.1059) (Figure 4). Ui .25 .05 w

Abstract (if available)

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Rationalizing risk: sexual behavior of gay male couples

PDF

The relationship between alcoholism and crime: autonomic and neurpsychological factors

PDF

How casual attribution influences behavioral actions taken by self-aware individuals

PDF

The Müller-Lyer illusion: a new variant, some old and new results

PDF

An investigation of the pseudohomophone effect: where and when it occurs

PDF

Violent environments and their effects on children

PDF

Children's immediate reactions to interparental conflict

PDF

A qualitative study on the relationship of future orientation and daily occupations of adolescents in a psychiatric setting

PDF

Predictors of pretend play in Korean-American and Anglo-American preschool children

PDF

Public art in corporate downtown Los Angeles

PDF

The application of Sarbin's theory of emotions as narrative emplotments to stories of two men diagnosed with cancer

PDF

The Cloisters Cross: a re-examination of date and style

PDF

"It was in the air": Adolph Gottlieb, the pictographs, New York, and the Zeitgeist of the 1940s

PDF

The use of occupational therapists or interdisciplinary teams in the evaluation of assistive technology needs of children with severe physical disabilities in Orange County schools

PDF

Structural equation modeling in educational psychology

PDF

Fine motor skills of two- to three-year-old drug exposed children

PDF

Work for the masters degree

PDF

Formative influences in the life of Olive Schreiner, Victorian feminist and freethinkier

PDF

On the limits of not being scripted: video-making and discursive prositioning in coastal southeast Sulawesi

PDF

Japanese pronunciation by English speaking students and English-Korean speaking students: contrastive and error analysis

Asset Metadata

Creator Greenwood, Kristyn Elaine (author)

Core Title Meta-analysis on the misattribution of arousal

School Graduate School

Degree Master of Arts

Degree Program Psychology

Degree Conferral Date 1995-05

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

Tag OAI-PMH Harvest,psychology, general,psychology, social

Format masters theses (aat)

Language English

Contributor Digitized by ProQuest (provenance)

Advisor Miller, Herman (committee chair), Raine, Adrian (committee member), Read, Stephen J. (committee member)

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c18-8363

Unique identifier UC11357412

Identifier 1378415.pdf (filename),usctheses-c18-8363 (legacy record id)

Legacy Identifier 1378415-0.pdf

Dmrecord 8363

Document Type Thesis

Format masters theses (aat)

Rights Greenwood, Kristyn Elaine

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