University of Southern California Dissertations and Theses

Gender and the bump: An investigation of the reminiscence effect in the Long Beach Longitudinal Study

(USC Thesis Other)

Gender and the bump: An investigation of the reminiscence effect in the Long Beach Longitudinal Study

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content GENDER AND THE BUMP:
AN INVESTIGATION OF THE REMINISCENCE EFFECT
IN THE LONG BEACH LONGITUDINAL STUDY
Copyright 2003
by
Patricia Housen
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
(GERONTOLOGY)
May 2003
Patricia Housen
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
UMI Number: 3103903
UMI
UMI Microform 3103903
Copyright 2003 by ProQuest Information and Learning Company.
All rights reserved. This microform edition is protected against
unauthorized copying under Title 17, United States Code.
ProQuest Information and Learning Company
300 North Zeeb Road
P.O. Box 1346
Ann Arbor, Ml 48106-1346
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
UNIVERSITY OF SOUTHERN CALIFORNIA
THE GRADUATE SCHOOL
UNIVERSITY PARK
LOS ANGELES, CALIFORNIA 90007
This dissertation, written by
Patricia Housen
under the direction of hsz 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
,-D ean -Dean of Graduate Studies
Date ... January _ 9 , _ 2003
DISSERTATION COMMITTEE
Chairperson
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
DEDICATION
To Genaro, truly the love of my life,
who endured coming home to a dark house and cold dinners
longer than I could reasonably expect.
To my parents Josef and Ingeborg,
whose ambitions for me tend to exceed my own
and always included education.
And to my sweet Omalein and Annabelle Molina,
who never asked me when I would be finished
or what I would do afterward.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
ACKNOWLEDGMENTS
I would like to thank five scholars who helped shape not only this
dissertation, but also my experience at the Andrus Center. First, Dr. Elizabeth
Zelinski. She contributed greatly to my intellectual growth as chair of my
dissertation committee, and as my demanding but also indulgent mentor for the past
six years. Her energy and enthusiasm for this research inspired me to continue. I
was also fortunate to have Drs. Eileen Crimmins, Kate Wilber, Dan Kempler, and
Phoebe Liebig on my dissertation and qualifying exam committees. Their keen
insights, constructive criticism and steady encouragement helped me through the
dark days.
A debt of gratitude goes also to my co-workers. Thanks to Dr. Rob
Kennison, for his interest in the project and great suggestions, and to Robin
Engberg, who collaborated with me on the pilot study upon which the dissertation is
based. Thanks also to Marc Simpao, and especially Megan Braziel, for working
through the data with me. Thanks to Deanah Kim, for responding cheerfully to
unreasonable requests for data entry, and also to Laura Giles, who helped present
the findings. Last but not least, thanks go to Dr. Kayan Lewis and Christianne
Lane, who brightened my life as members of our “18-Month-Club.”
Finally, I would like to acknowledge the National Institutes of Health
Multidisciliplinary Training in Gerontology Grant directed by Dr. Vem Bengtson.
The fellowship support allowed me to focus on completing this research.
iii
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
TABLE OF CONTENTS
Dedication........................................................................................................................ ii
Acknowledgments..........................................................................................................iii
List of Tables..................................................................................................................vi
List of Figures.................................................................................................................ix
Abstract.............................................................................................................................x
Chapter I: Introduction.....................................................................................................1
I.A. Background.................................................................................................1
I.B. Research Questions....................................................................................6
I.C. Organization of the Dissertation...............................................................7
Chapter II: Literature Review.........................................................................................9
II.A. Definitions................................................................................................ 9
H.B. Empirical Studies of the Bump..............................................................16
II.B. 1. Discovery of the Bump.............................................................16
II.B.2. The Bump: Robust, Sensitive, Individualized........................22
II. C. Explaining the Bump............................................................................. 32
II. C.l. Cognitive Theories of the Bump...............................................33
II.C.2.Socio-contextual Theories of the Bump.................................. 37
II.D. Hypotheses..............................................................................................42
Chapter III: Research Design........................................................................................45
III.A. Sample................................................................................................... 45
III.B. Measures................................................................................................ 48
III.C. Content Analysis....................................................................................52
III.D. Analysis Procedures............................................................................. 68
Chapter IV: Panel 2 (Experiment 1)............................................................................ 73
IV.A. Descriptive Statistics........................................................................... 73
IV.B. Autobiographical Memory Distribution of the Oldest Old............... 75
IV.C. Autobiographical Memory Distribution and Topic............................79
IV.D. Autobiographical Memory Distribution and Gender.........................82
IV.E. Multivariate Analyses.......................................................................... 89
IV.F. Summary............................................................................................... 96
iv
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Chapter V: Panel 3 (Experiment 2).............................................................................. 99
V.A. Descriptive Statistics............................................................................. 99
V.B. Autobiographical Memory Distribution of the Oldest Old.............. 102
V.C. Autobiographical Memory Distribution and Topic........................... 105
V.D. Autobiographical Memory Distribution and Gender........................108
V.E. Multivariate Analyses..........................................................................114
V.F. Summary.............................................................................................. 121
Chapter VI: Summary and Discussion........................................................................124
VI. A. Review of the Findings...................................................................... 124
VI.B. Theoretical Implications and Future Directions.............................. 136
References.....................................................................................................................143
v
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
LIST OF TABLES
Table II. 1 Taxonomy of Autobiographical and NonAutobiographical
Memory Content...................................................................................10
Table III. 1 Distribution of Memories by Life Period and Content
Domain................................................................................................. 63
Table IV. 1 Means and Standard Deviations of Variables (Participants
Reporting from Ages 0-51)..................................................................74
Table IV.2 Means and Standard Deviations of Variables (Participants
Reporting from Ages 52+)...................................................................74
Table IV.3 Gender Distribution (n=251).................................................................75
Table IV.4 Education
(n=236).................................................................................................. 75
Table IV.5 Age
(n=251).................................................................................................. 75
Table IV.6 Participants by Age Group (n=251).....................................................76
Table IV. 7 Means and Standard Deviations of Variables for Younger and
Older Participants..................................................................................76
Table IV.8 Topics of Autobiographical Memory Protocols (n=251)................... 80
Table IV.9 Means and Standard Deviations of Variables by Topic.................... 81
Table IV.10 Means and Standard Deviations of Variables by Gender.................. 83
Table IV. 11 Means and Standard Deviations by Topic (Men).........................85
Table IV.12 Means and Standard Deviations by Topic (Women)...................... 87
Table IV. 13 Significant Intercorrelations o f the Independent and Dependent
Variables (n=251)................................................................................ 91
Table IV. 14 Gender and Topic as a Function of Autobiographical Memory
Distribution (n=199)............................................................................ 93
vi
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table IV. 15 Means and Standard Deviations of Variables by Life
Period.................................................................................................... 93
Table IV. 16 Autobiographical Memory as a Function of Age, Education,
Gender, Topic, Personality, and Memory (n=199)............................95
Table IV. 17 Multinomial Regression of the Odds of Writing from Three Life
Periods (Model 3).................................................................................96
Table V.l Means and Standard Deviations of Variables (Participants
Reporting from Ages 0-51)................................................................100
Table V.2 Means and Standard Deviations of Variables (Participants
Reporting from Ages 52+)................................................................. 101
Table V.3 Gender Distribution
(n=268)................................................................................................ 101
Table V.4 Education
(n=266)................................................................................................ 101
Table V.5 Age
(n=268).................................................................................................102
Table V.6 Participants by Age Group (n=268)....................................................102
Table V.7 Means and Standard Deviations of Variables for Younger and
Older Participants................................................................................103
Table V.8 Topics of Autobiographical Memory Protocols (n=268)..................105
Table V.9 Means and Standard Deviations of Variables by Topic.................. 106
Table V.10 Means and Standard Deviations of Variables by Gender............... 108
Table V.l 1 Means and Standard Deviations by Topic (Men)............................110
Table V.12 Means and Standard Deviations by Topic (Women)..................... 112
Table V. 13 Significant Intercorrelations of the Independent and Dependent
Variables (n=268)...............................................................................116
vii
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table V.14 Gender and Topic as a Function of Autobiographical Memory
Distribution (n=230)...........................................................................118
Table V. 15 Means and Standard Deviations of Variables by Life Period..........118
Table V. 16 Autobiographical Memory as a Function of Age, Education,
Gender, Topic, Personality, and Memory (n=230).......................... 120
Table V.l 7 Multinomial Regression of the Odds of Writing from Three Life
Periods (Model 3)...............................................................................121
Table VI. 1 Summary of Descriptive Analyses.................................................... 125
Table VI.2 Summary of Multivariate Analyses................................................... 128
Table VI.3 Autobiographical Memory as a Function of Panel, Gender, Topic,
Personality, and Memory (n=429).................................................... 134
Table VIA Multinomial Regression of the Odds of Writing from Three Life
Periods (Model 5)...............................................................................135
v iii
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
LIST OF FIGURES
Figure III. 1 Age of Memories (Participants 55 and Older)................................... 47
Figure III.2 Age of Memories by Life Period (Participants 55 and Older)...........59
Figure III.3 Content Domains (Panel 2)...................................................................61
Figure 131.4 Content Domains (Panel 3)...................................................................62
Figure IV. 1 Distribution of Autobiographical Memory by Age Group.................78
Figure IV.2 Distribution of Autobiographical Memory by Topic......................... 81
Figure IV.3 Distribution of Autobiographical Memory by Gender....................... 84
Figure IV.4 Distribution of Autobiographical Memory by Gender (Male)
and Topic.............................................................................................. 86
Figure IV. 5 Distribution of Autobiographical Memory by Gender (Female)
and Topic.............................................................................................. 88
Figure V. 1 Distribution of Autobiographical Memory by Age Group............... 104
Figure V.2 Distribution of Autobiographical Memory by Topic........................107
Figure V.3 Distribution of Autobiographical Memory by Gender......................109
Figure V.4 Distribution of Autobiographical Memory by Gender (Male)
and Topic............................................................................................ 111
Figure V.5 Distribution of Autobiographical Memory by Gender (Female)
and Topic............................................................................................ 113
ix
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
ABSTRACT
When asked to recollect personal memories from their past, older individuals
disproportionately recall experiences from adolescence and young adulthood. This
cache of recollections, called the bump, has proven tantalizingly robust, extending
across several different methods of data collection. In this study, a free recall
measure was used to investigate whether the bump endures into the ninth decade
and beyond, to further test its resilience to instructional modification, and to
determine if there are gender differences in the distribution. Subsidiary hypotheses
tested the role of life cycle factors (age, education), personality (tender-mindedness
and assertiveness), and memory (episodic and working memory). Guided by
theories proposed to account for the bump, multinomial logistic regression was used
in cross-sectional analyses of data from the 1994-1995 (n=199) and 2000-2002
(n=230) panels of the Long Beach Longitudinal Study. Results were substantially
consistent across the experiments for two of the three primary research questions. A
bump of memories emerged for participants 80 years and older that was statistically
indistinguishable from the bump generated by those 55-79 years of age, and
participants writing about people produced older memories than those writing about
events and decisions. However, while there was evidence of a bump among men
across both panels, among women the phenomenon emerged in only one of the data
sets. Also, there was some indication that older participants may be more likely to
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
produce memories from childhood and the bump than subsequent adulthood, and
that more tender-minded individuals may be more likely to produce memories from
adulthood than the bump.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CHAPTER I: INTRODUCTION
LA. BACKGROUND
Researchers in a number of laboratories have identified a curious
phenomenon. When asked to recollect personal memories from their past, older
individuals disproportionately recall experiences from adolescence and young
adulthood. This anomaly, referred to in the autobiographical memory literature
either as the bump (Rubin & Schulkind, 1997a) or the reminiscence effect
(Fitzgerald, 1996), is defined as the “increase in early memories above what would
be expected by a monotonically decreasing retention function” (Rubin, Wetzler &
Nebes, 1986, p. 208).
As described by Kuhn (1962), paradigms define the problems and methods
used in normal science. They can serve as eye glasses that help researchers focus on
issues hypothesized to be important, and/or function as blinders that obscure
important but unacknowledged phenomena. The discovery of unexpected
anomalies - the bump, for example - can disrupt the course of normal science
conducted in the tradition of a particular paradigm. In this case it has, among other
things, encouraged consideration of “potential filters” such as age, gender and
personality characteristics in the study of autobiographical memory (Webster and
Cappeliez, 1993, pp.80, 84).
The bump was discovered by Rubin, Wetzler and Nebes (1986) in a
reanalysis of data gathered by researchers in several other laboratories over the
1
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
previous decade. Its discovery was unexpected because researchers had anticipated
their data would confirm that autobiographical memory followed principles
exhibited by other memory phenomena, principles with a lineage that stretched back
to the earliest days in the study of memory. Research from the dominant learning
and retention paradigm has consistently shown that the rate of forgetting is most
rapid immediately upon learning, diminishing after longer intervals. And these
expectations were in fact satisfied when young college-aged people were studied.
The distributions of memories from across short life spans did follow a regular
pattern, so that far larger numbers of recent memories than distant memories were
recalled.
The brilliance of Rubin, Wetzler and Nebes (1986) was their willingness to
abandon familiar strategies to pursue hints that for life span samples the data
collected sometimes deviated from the typical pattern. The model they advanced
included three distinct periods. For participants of all ages, there was a recency
component comprised of a large number of memories encoded in the past few years,
as well as a period of childhood amnesia resulting in a dearth of memories from the
earliest years of life. The authors also identified a period of reminiscence, but this
phenomenon only emerged among somewhat older participants. For these
individuals, the many memories derived from about ages 10 to 30 formed a distinct
bump in the distribution.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The finding personal memories from late adolescence and young adulthood
are over-represented was disruptive to the learning and retention paradigm because
it indicated not only biological and maturational factors, but also environmental and
life course factors might be at work. Conway and Rubin (1993) maintained the
phenomenon is sufficiently documented “to show that it cannot be viewed as an
experimental artifact or a curiosity without clinical or sociological importance”
(p. 3). It has been proposed that the bump may represent an “interaction of the
memory system, culturally defined developmental tasks, and culturally derived
forms of representation” (Fitzgerald, 1996, p. 371), an interpretation also
encouraging a more contextual approach.
Yet many, though not all, of the studies of the bump have tended to focus on
cognitive mechanisms. For example, a variety of characteristics of memories in the
bump (e.g. rehearsal, novelty, imagery, meaningfulness, pleasantness) have been
examined to see if they are distinctive in some way. And the effect of a number of
methods used to elicit memories (e.g. cue-word method, important/vivid memory
method), and variations of these methods (e.g. the number and characteristics of
cue-words, instructions regarding recency or the search strategy to be used, wording
of instructions) on how many and how quickly memories are generated has been
studied. While each of these lines of inquiry has contributed to our knowledge
about the bump, less attention has been paid to psychosocial characteristics of the
rememberer that may affect the distribution of autobiographical memory.
3
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
It should be noted at the outset that the autobiographical data used in this
study were derived from an experiment originally designed to test for
developmental changes in language production, and that the analysis was cross-
sectional. The data were collected as part of the 1994-1995 and 2000-2002 panels of
the Long Beach Longitudinal Study, a 25-year-old investigation of memory change
in adults. Most studies of the bump have tended to focus on younger age groups,
and the few that extended into the ninth decade were not specifically focused on
confirming the bump. Because the Long Beach Longitudinal Study includes an
unusually large number of participants 80 years and older, it was possible to test for
age differences in the distribution of autobiographical memory among these oldest-
old.
There is some indication that instructional modifications that are biased
towards older or more recent periods can affect the distribution of autobiographical
memory (Howard and Katz, 1992; Rubin & Schulkind, 1997a). There is also
convincing evidence of individual differences in the distribution of autobiographical
memories (e.g. Rubin, 1982; Rubin, 1989; Rubin & Schulkind, 1997b). Therefore,
the distribution of autobiographical memories was compared for participants writing
about two different categories of topics, and two experiments explored whether
instructional modifications giving participants differing amounts of freedom in
choosing between these topics might affect it. However, the primary goal of this
dissertation was to test for gender variations in the distribution using an individual
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
differences approach, taking social roles, personality and memory ability into
account.
Findings from this study were used to assess several of the theories
highlighted in the work of Rubin (e.g. Rubin, Rahhal & Poon, 1998) and Fitzgerald
(e.g. 1996). The optimal memory theory suggests the making of bump memories is
related to the ebbing and flowing of a variety of cognitive abilities, including
memory. Two more widely held explanations are the rapid change theory and the
self-narrative hypothesis. The rapid change theory is cognitive and mechanistic,
virtually ignoring contextual factors. It proposes memories from adolescence and
young adulthood are favored because they are novel and, especially when they are
encoded in periods of transition, benefit from cognitive mechanisms including
distinctiveness and a lack of proactive interference. The self-narrative hypothesis,
on the other hand, is an organismic and contextual theory. It proposes that the
bump reflects the considerable cognitive effort young people expend to create a
self-narrative based on personal memories that are important to their sense of
identity. Effort for meaning and rehearsal may help explain the retrieval of bump
memories using both the rapid change theory and self-narrative hypothesis
(Fitzgerald, 1996; Rubin, Rahhal & Poon, 1998).
The self-narrative hypothesis rests on Erikson’s theory of personality
development, which does not rule out the effects of social, cultural and historical
factors. Thus it is not unreasonable to anticipate gender differences in the
5
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
distribution of autobiographical memories. Research confirms women
disproportionately serve as kinkeepers and caregivers (Hagestad, 1986;
Montgomery & Datwyler, 1992). Women are also described as being more
communal, and as living more interdependent lives (e.g. Gilligan, 1982). Based on
these findings, hypotheses predicting a flatter distribution of autobiographical
memories across the life span for women could be derived in at least two ways. On
the one hand, the order and/or meaning of Erikson’s stages relating to identity may
differ by gender, and hence the distribution of important autobiographical
memories may cluster differently for men and women. The second explanation is
structural. It suggests women’s communal tendencies, which can be measured
using the personality facet called tender-mindedness (Feingold, 1984), basically
reflect self-perpetuating stereotypes linked to their social roles as caregivers and
kinkeepers. These gendered social roles result in turning points that differ in their
timing and significance for men and women. Since turning points are often novel
events or transitions benefitting from the cognitive mechanisms mentioned above,
these differences may be reflected in the distribution of autobiographical memory.
I.B. RESEARCH QUESTIONS
The bump appears to be quite robust. If participants are instructed not to
report memories from the recent past, research has demonstrated it can emerge in
individuals as young as 36 years of age (Jansari & Parkin, 1996). Little information
is available, however, as to whether the phenomenon extends into the ninth and
6
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
tenth decades of life, a period in which cognitive declines have been identified. The
bump also appears to be relatively immune to the methodology used (e.g. cue word,
free recall), including the sensory properties of cue words (Rubin & Schulkind,
1997c). However, there is some indication that demand characteristics in the form
of experimental instructions can bias the overall distribution of autobiographical
memory (Howes and Katz, 1992; Rubin & Schulkind, 1997a). Finally, few studies
have examined the possibility of gender variations in the bump (see Conway &
Haque,1999; and Rubin, Schulkind, & Rahhal, 1999). These are issues addressed
by the following research questions: 1.) Does the bump endures into the ninth
decade and beyond? 2.) Is the bump immune to instructional modification of free
recall tasks? and 3.) Is there evidence of the bump in both genders?
I.C. ORGANIZATION OF THE DISSERTATION
This dissertation has six chapters. The first has provided a general
introduction to the bump and its significance for the study of autobiographical
memory. Literature on the bump will be reviewed in Chapter II. Chapter III
describes the research design, including the samples, measures, and analysis
procedures used to test the research questions. Chapters IV and V present results of
the analyses for samples drawn from the 1994-1995 (Experiment 1) and 2000-2002
(Experiment 2) panels of the Long Beach Longitudinal Study. Chapter VI will be
used to draw conclusions and explore the implications of the two experiments in
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
regard to instructional modifications. Limitations of this study and
recommendations for future investigations will also be discussed in this
chapter.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CHAPTER II: LITERATURE REVIEW
Researchers in a number of laboratories have identified a curious
phenomenon. When asked to recollect personal memories from their past, older
individuals disproportionately recall experiences from adolescence and young
adulthood. This cache of recollections, called the bump, has proven tantalizingly
robust, extending across several different methods of data collection. Theories
attempting to account for why people would focus on this period of life have
emerged, however relatively few explanatory variables have been confirmed. The
purpose of this chapter was to establish a working definition of the phenomenon
studied, to review empirical studies and theoretical explanations of the bump, and to
present the hypotheses to be tested.
II.A. DEFINITIONS
Brewer (1996, p. 21) was able to compile an extensive list of terms from the
literature, all of them referring to long-term autobiographical memory for facts. This
reflects the nature of the phenomenon being studied, which centers on information
relating to the self, but which overlaps with an individual’s knowledge of his/her
environment. In an effort to classify the different forms of autobiographical
memory, Brewer (1986, p. 29) had previously distinguished memories based on
acquisition condition and form of representation. That is, memories result from
exposure to single or repeated episodes (acquisition condition), and they may or
may not contain imagery (form of representation). In addition to Brewer’s four
9
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
subcategories of personalized autobiographical memories (i.e. personal memory,
generic personal memory, autobiographical fact, and self schema), several
subcategories of depersonalized non-autobiographical memory have also been
described (see Table II. 1, adapted from Webster & Cappeliez, 1993, p. 69). It
should be noted that based on Brewer’s criteria, flashbulb memories - defined by
Kulik and Brown (1977) as detailed, vivid, accurate, and enduring memories
Table II.l. Taxonomy of Autobiographical and NonAutobiographical Memory
Content
Autobiographical Memory
Personal Memory
s,i,p
I remember interviewing for a secretarial
position at ABC Co. headquarters in Los
Angeles on 9/1/1990.
Generic Personal
Memory
R,I,P
I remember bringing my lunch to work in a
brown paper sack, but I don’t remember a
particular lunch.
Autobiographical Fact S,N,
P
I know my parents bought a mattress made by
ABC Co. when I was a child, but I have no
memory of it.
Self Schema R,N,
P
I believe that I am hard working.
NonAutobiographical Memory
Historical Knowledge R,N,D I know the first quarter ends in March.
Historical Knowledge R,I,D I know the annual trade show is in New
York.
Historical Knowledge S,I,D I know ABC Co. changed its logo in 1989.
Historical Knowledge S,N,D I know my boss was bom in Chicago.
S, single episode; R, repeated episode; I, imaginal; N , nonimaginal; P, personalized;
D, depersonalized. Adapted from “Reminiscence and Autobiographical Memory,”
by J.D. Webster & P. Cappeliez, 1993, Developmental Review, 13, p. 69. Copyright
1993 by Academic Press, Inc.
10
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
relating to surprising and consequential events - would also be classified as
personal memories.
These distinctions between contextualized, personalized autobiographical
memories and decontextualized, depersonalized nonautobiographical memories and
their subcategories do not eliminate the problem of overlap. For example,
important personalized autobiographical facts (e.g., the knowledge her first mattress
was made by ABC Co.) and depersonalized historical knowledge (e.g., the
knowledge ABC Co. changed its logo in 1989) may enrich an employee’s retelling
of an event (e.g. the story of her interview for a secretarial position). Still, Brewer’s
taxonomy is a useful tool for researchers in designing experiments and discussing
their results.
In this dissertation, the term autobiographical memory will refer to personal
and generic personal memories, which are similar in that they are both acquired
through personal experience. That is, they are not a retold memory of someone
else’s experience (e.g., a new colleague’s description of her job interview) or factual
information (e.g., knowledge of the fiscal calendar). The subcategories both draw
on imagery. So, in addition to a number of facts about the interview (e.g., its time
and location), the employee’s personal memory may contain images (e.g. the suit
she wore and the interviewer’s desk) that can be retrieved for a retelling. However,
whereas a personal memory is made in a single episode, a generic memory tends to
11
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
encompass information acquired over the course of several episodes (e.g. her
practice of bringing brown bag lunches to work).
The two subcategories often overlap within freely recalled autobiographical
memories. In the following passage, for example, a specific memory is identified
and dated by a Long Beach Longitudinal Study participant, then justified as
important because of its consequences, which are generic:
The decision I made 44 years ago to move to California was a major influence
on my life because it led to my meeting my dear friend... She instructed me in
the are of making flower arrangements using tiny beads in all colors strung on
fine wire to fashion creative flower arrangements of all kinds. I received
many awards from them when they were exhibited publicly over the years.
The participant’s blurring of general and specific personal information corresponds
to the proposition that autobiographical memory is hierarchically structured, and
that autobiographical memories represent constructed compilations of knowledge
spanning three cross-indexed levels of knowledge (Conway & Rubin, 1993;
Conway, 1996). It is proposed that autobiographical knowledge about “lifetime
periods” (e.g. the years the employee worked at ABC Co.) contains thematic
knowledge (e.g. the work theme). Knowledge about “general events” corresponds
tolerably well to Brewer’s generic and personal memories categories. Finally,
“event specific knowledge” contains sensory and emotional information relating to
a general event.
The terms general or overgeneral memory, corresponding to what Brewer
refers to as generic personal memory, have been refined to differentiate between
12
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
memories that are categorical and extended (Williams & Dritschel, 1992). Whereas
categorical memories summarize across similar experiences, extended memories
temporally telescope an experience. For example, cued by the word “angry,” one
participant may respond “computerized telephone menus” (categorical memory),
and another may respond “the month it took to get my new computer repaired”
(extended memory).
Many of the researchers interested in autobiographical memory and its
distribution do not study generic personal memories, limiting their work to personal
memories occurring at a particular place during a limited period of time (Williams
& Dritschel, 1992). The instructions for these experiments impose control by asking
for specific memories in which participants were personally involved (e.g. Hyland
& Ackerman, 1988; Jansari & Parkin, 1996; Conway & Haque, 1999). Bekerian
and Dritschel (1992) discourage this approach, citing the importance of considering
other retrieval conditions, especially for the development of autobiographical
memory theory. The author’s concern mirrors Rubin’s (1989b, p. 84) argument that
“control is more often a vice than a virtue,” since it can result in the mis-
classification of important but unexpected effects as error, and since it limits the
generalizability of the results.
Williams and Dritschel (1992) offered evidence that a strict focus on
specific personal memories may ignore the potential importance of generic personal
memories. The authors found that between 16 and 28 percent of the memories
13
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
participants generated in several experiments were general, even though the
instructions had requested specific memories (Williams & Drischel, 1992 citing
Pillemer et al., 1986, Williams & Broadbent, 1986; Williams & Dritschel, 1988;
Williams & Scott, 1988). Similarly, Franklin and Holding (1977) reported that 31
percent of the memories generated in response to cue words were of recurrent
events, and thus difficult to date. In their examination of autobiographical accounts
published in the books A History of Psychology in Autobiography and Models of
Achievement. Mackavey, Malley and Stewart (1991) found that more than two-
thirds of the memories analyzed occurred over relatively long periods of time.
Why some participants generate general personal memories is uncertain, but
potentially very interesting. In a free recall study of college aged-students, Barselou
(1988) reported that only 21 percent of the autobiographical memories reported
were of specific events, and that some participants had difficulty when asked to
generate only specific memories in two follow-up experiments. He suggested this
indicated that the instructions interfered with how the participants were attempting
to access the information. On the one hand, categorical and extended memories
may be a practical way of making sense of the world and conveying out experiences
to others. However, Williams (1996) hypothesized that the production of such
memories might also be caused by a process where individuals cannot get beyond
the level of categorical memories. This “mnemonic interlock” (p. 245) was
attributed to an individual’s need to minimize negative affect, which is achieved by
14
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
aborting the retrieval of more specific memories that may be painful. Alternately,
mnemonic interlock may reflect reduced cognitive capacity, perhaps indexed by
working memory.
While some researchers have focused on the mechanics of autobiographical
memory, including its distribution, others have focused on its function and purpose
in relation to life review and reminiscence behavior (see Webster & Cappeliez,
1993, and Fitzgerald, 1996). This dissertation generally falls into the first camp,
focusing on a phenomenon occurring within the distribution of autobiographical
memories called the bump (Rubin & Schulkind, 1997a) or reminiscence effect
(Fitzgerald, 1996). Adults appear to recall more memories from adolescence and
young adulthood than would be expected if the contents of memory simply reflected
forgetting as a function of time. The discovery of the bump, however, also
encourages hypotheses about contextual factors. Its discovery has enlivened the
discussion of whether autobiographical memories are copies of lived experiences —
some privileged by cognitive mechanisms including novelty and rehearsal;
individualized cognitive constructions — perhaps dependent upon sociohistorical
context as well as psychosocial development; or some combination thereof. The
Long Beach Longitudinal Study provides an unusually large population in which to
study the phenomenon. It also includes more than 200 participants over 80, an age
group in which the existence of the bump has not yet been specifically confirmed.
15
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
II.B. EMPIRICAL STUDIES OF THE BUMP
II.B.l. Discovery of the Bump
The bump may be described as an anomaly, defined by Kuhn (1962) as
unanticipated phenomena that disrupt the course of normal science. Studies
undertaken by Crovitz and Schifftnan (1974), Franklin and Holding (1977), Rubin
(1982), and Fitzgerald and Lawrence (1984) helped set the stage for its accidental
discovery. While they were grounded in the learning and retention paradigm
dominating the study of memory, the design of these experiments also reflected
three important changes in the climate of opinion within the field, which were
identified by Robinson (1986, p. 21).
The first change challenged the adequacy of the experimental approach.
Scientific interest in memory stretches back more than a century to Ebbinghaus and
Galton. Wanting to study the mechanics of memory, but concerned that using
familiar facts as stimuli would confound his experiments, Ebbinghaus introduced
the use of nonsense syllables. Researchers interested in the functional role and
underlying mental processes of memory began to seriously question the ecological
and external validity of Ebbinghaus’ legacy beginning in the 1970s (e.g. Neisser,
1978, 1982; Baddeley, 1989; Bahrick, 1989; Petrinovich, 1989; Rubin, 1989a,b).
Among other things, they were concerned the stimuli used in the laboratory were
practically irrelevant, and that findings from such experiments would not be
generalizable because they do not predict real-world behavior. More specifically,
16
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Bahrick (1989) argued that unless the memory content being studied could be
acquired quickly, was emotionally and motivationally neutral, and would be tested
for retention in short order, it should be studied more naturalistically.
Bahrick’s criteria suggested that autobiographical memory phenomena do
not support a strict experimental approach. The encoding of autobiographical
memories is thought to be a continuous process after very early childhood, and
individual life circumstances and abilities at the time of encoding, storage and
retrieval differ in countless ways. Galton (1879) implicitly came to terms with the
nonlaboratory environments in which autobiographical memories are made. He
devised a method based on cue words to probe his own cache of recollections, and
then described them in terms of how long it took him to elicit associations and how
old the elicited memories were. Galton was 57 at the time of his study, an age
considered advanced for the era, and he was also much older than most participants
in studies of autobiographical memory undertaken in the 1970s. It is interesting that
39 percent of Gabon’s memories came from childhood and young adulthood, 48
percent from manhood, and 15 percent from the most recent period.
Retrospectively it seems ironic that the unusual distribution he reported,
which was probably not a monotonically decreasing function of time, was ignored
for so long.
Crovitz and Schiffinan (1974) rekindled interest in Gabon’s method. They
conducted a more controlled study of autobiographical memory in a sample of
17
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
undergraduates, which limited the potential for encoding and sampling differences
but also limited the generalizability of their results. The authors used cue words
recognized as common to eliminate bias, specifically asked for personal memories,
and treated the ages of the reported memories in what they considered to be a more
meaningful way. This involved using a log function for frequency and time in order
to analyze more remote autobiographical memories, which they found spanned
greater intervals. The finding that the frequency of reported memories declined as
their age increased was later interpreted as a retention function for episodic
memory, primarily reflecting retrieval failure (Rubin, 1982). The studies
undertaken by Franklin and Holding (1977), Rubin (1982), and Fitzgerald and
Lawrence (1984) continued in the cue-word tradition, exploring how the use of
personality measures (i.e. extroversion-introversion), and structurally (e.g. nouns,
verbs, adjectives) and qualitatively (e.g. positive and negative affect) different cues
affected the distribution of autobiographical memory and retrieval latencies.
The second change cited by Robinson (1986, p. 21) was the development of
new theoretical models of memory. Tulving’s (1972) decomposition of the long
term memory system is considered particularly important. The semantic memory
system was hypothesized to store general, decontextualized knowledge of the
world. The episodic memory system was hypothesized to store context-based
information relating to events experienced by an individual, “render(ing) possible
conscious recollection of personal happenings and events from one’s personal past
18
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
and mental projection of anticipated events into one’s subjective future” (Wheeler,
Stuss & Tulving,1997, p. 332). However, Brewer (1986) contends the confusion
about the abstract definition of episodic memory and its actual usage has confused
the study of autobiographical memory. He pointed out that based on Tulving’s
description of episodic memory, it is generally assumed that autobiographical
memory is part of that system (as in Rubin, 1998; Nelson, 1992). Counter
intuitively, Tulving proposed that episodic memory could be tested in the laboratory
using list learning tasks for verbal items, including nonsense syllables and foreign
language vocabulary. He has subsequently acknowledged that most of these tasks
are only minimally episodic in nature (Wheeler, Stuss & Tulving, 1997).
Laboratory-based studies of episodic memory show a great deal tends to be
forgotten shortly after something has been learned. Forgetting curves depicting this
pattern, which levels off over time, are often converted into linear power functions.
Interested to discover whether “episodic” memory in and out of the lab would
exhibit the same frequency distribution, researchers replicated this procedure for
autobiographical events (e.g. Crovitz & Schifftnan, 1974; Rubin, 1982). The
research confirmed similarities, however the logarithmic manipulations later proved
problematic in that they helped to obscure the bump.
The third change cited by Robinson (1986, p. 21) — a new concern with
adult development — was prompted by increases in longevity and the growth of the
elderly population, and the implications of this trend for the society. Between 1900
19
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
and 1991 life expectancy at birth in the United States increased by 29 years, and
between 1900 and 1994 the number of people 65 and older had increased 11 times
to 33.2 million (U.S. Bureau of the Census, 1996). Managing the economic, social
and political implications of this ongoing demographic change will clearly require a
better understanding of the aging process, and how abilities and behavior change
over the lifespan.
Early studies in autobiographical memory used subjects who were young,
typically in their college years (e.g. Crovitz & Schifftnan, 1974; Fitzgerald, 1980;
Fitzgerald, 1981; Robinson, 1976; Rubin, 1980; Rubin, 1982). However, the bump
is generally not found in younger age groups without some modification of
instructions (e.g. Fitzgerald, 1996; Jansari & Parkin, 1996). When Franklin and
Holding (1977) and Fitzgerald and Lawrence (1984) included older participants in
their samples, irregularities began to emerge. Because these experiments were
testing other hypotheses, the anomalies were noted but not explored.
Franklin and Holding (1977) tested the adage that older persons tended to
live in the past. They confirmed that the number of memories reported was a
function of the age of the memory across the sample as a whole. However, the
researchers commented that the relationship seemed “unnecessarily obscure, due to
the procedures conducted” (p. 529). By averaging the medians of individual
distributions for each of the five age groups, they demonstrated that older people did
generate somewhat older memories. Yet frequency polygons of the distributions,
20
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
which appear to show evidence of a bump for most age groups, indicated older
people were also encoding a considerable number of recent memories. Franklin and
Holding concluded the clear tendency towards bimodality and skewness in the data
seriously contradicted a simple linear interpretation (p.530).
Fitzgerald and Lawrence (1984) were interested in whether they could
identify systematic individual differences in autobiographical memory. The
researchers explicitly rejected the use of equal time units because they felt log units
would allow for more direct comparisons of slopes and intercepts between age
groups and prompt types. They found that the power function fit the data of
younger participants. Unexpectedly, they also found similar numbers of 60-year-old
and 20-year-old memories were generated per hour among participants 60-75 years
of age.
An insightful reanalysis of the data gathered by Franklin and Holding
(1977); Fitzgerald and Lawrence (1984); Zola-Morgan, Cohen and Squire (1983);
and new data gathered by Rubin, Wetzler and Nebes (1986) focused attention on
these deviations. In their reanalysis, Rubin and his colleagues (1986) used equal
time units instead of logarithmic scales, which had theretofore compressed the
early years of the participants’ lives. As a result, they identified a bump of
memories stemming from the time the participants were between about ten and 30
years of age, and also helped generate a new research agenda.
21
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
II.B.2. The Bump: Robust, Sensitive, Individualized
Rubin, Wetzler and Nebes (1986) concluded that participants under 30 years
of age did not appear to reminisce, those over 50 certainly did, and that the data was
inconclusive for participants in their 40s. While few studies of the bump have
sampled the oldest old, and none of those reviewed have specifically focused on
confirming the bump in this age group, several experiments have found the bump in
younger individuals under specific conditions. Fitzgerald (1996) reported an
experiment in which individuals between 31 and 46 years of age appeared to show a
strong reminiscence effect. However, as noted by Rubin and Schulkind (1997a),
specifying the bump in this population can be difficult since there is no trough
separating it from the more recent memories that generally predominate. Jansari
and Parkins (1996) tackled the problem by restricting the reporting of memories
from the recent past. In their experiment, cue words were presented to participants
in three age groups, half of whom were told not to include memories from the past
2 Z 'i years. The researchers found that participants of all ages in the no-recency
condition recalled more memories from age 6 to 30, but the effect was particularly
strong for the two younger age groups, resulting in a bump even for those 36 to 40
years of age. Rubin suggested a reanalysis of the data for the unrestricted
participants, eliminating memories less than five years old. The authors found this
procedure produced similar distributions for the recency and no recency groups
across all three age groups.
22
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The over-representation of recent memories has been identified for
individuals of all ages. This retention function is considered one of the stable
components in the distribution of autobiographical memories (Rubin, Wetzler &
Nebes,1986; Rubin & Schulkind, 1997a,b). Certain types of cues, for example
concrete nouns and words high in imagery (e.g. Rubin & Schulkind, 1997c;
Fitzgerald & Lawrence, 1984), do appear to produce older memories than other cues,
for example affect words (e.g. Fitzgerald & Lawrence, 1984). However, none of the
word properties studied by Rubin and Schulkind (1997c) (i.e. imagery,
concreteness, meaningfulness, goodness, emotionality, and frequency) satisfactorily
predicted the generation of bump memories.
The cue word and free recall methods predominate in the empirical literature
on the bump. Even though participants are asked to retrieve autobiographical
memories in these qualitatively different ways, experiments using cue-words have
exposed the bump (e.g. Rubin, Wetzler & Nebes, 1986; Hyland & Ackerman,
1988; Rubin, 1989b; Howes & Katz, 1992; Jansari & Parkin, 1996; Rubin &
Schulkind, 1997a,b; Conway & Hacque, 1999; Rybash, 1999; Rubin, Schulkind &
Rahhal, 1999), as have experiments asking for freely recalled important or vivid
memories (e.g. Fitzgerald, 1988; Mackavey, Malley & Stewart, 1991; Fromholt &
Larsen, 1991; Fromholt, Larsen & Larsen, 1995; Fitzgerald, 1996; Brady-Elnick et
al.,1999; Holmes & Conway, 1999). The same is true for experiments asking
participants to freely recall autobiographical memories they would want included in
23
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
a book about their lives (e.g. Fitzgerald, 1996) or in their life history time line
(Brady Elnick et al., 1999), indicating the bump is a robust effect.
Methodological variations do not eliminate it. For example, cue words used
to prompt autobiographical memories have differed in their affective character (e.g.
Fitzgerald & Lawrence, 1984; Jansari & Parkin, 1996; Conway & Hacque, 1999),
their frequency of use (e.g. Franklin & Holding, 1977; Fitzgerald & Lawrence,
1984), and their meaningfulness and imagery (e.g. Rubin, Wetzler & Nebes, 1986).
The bump emerged whether just a few (e.g. Conway & Hacque, 1999; Fitzgerald,
1988; Fitzgerald, 1996) or many (e.g. Rubin, 1989b; Rubin & Schulkind, 1997b)
autobiographical memories were requested, and whether data were collected orally
(e.g. Fromholt & Larsen, 1991; Fromholt, Larsen & Larsen 1995) or in a written
format (e.g. Fitzgerald, 1988; Mackavey, Malley & Stewart, 1991; Fitzgerald, 1996;
Brady-Elnick et al, 1999; Holmes & Conway, 1999).
However, the peak and location of the bump can vary. The distribution of
autobiographical memories generally, but not always (e.g. Conway & Hacque, 1999)
includes a smaller bump and a much larger recency component using the word cue
method (e.g. Rubin, Wetzler & Nebes, 1986; Hyland & Ackerman, 1988; Rubin &
Schulkind, 1997a,b; Hentges, 1997; Rubin, Schulkind & Rahhal, 1999) than the free
recall method (e.g. Fitzgerald, 1988; Mackavey, Malley & Stewart, 1991; Fromholt
& Larsen, 1991). Free recall experiments (e.g. Fitzgerald, 1988; Fromholt &
Larsen, 1991; Fitzgerald, 1996) also appear to include fewer years in the bump than
24
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
those using the cue word method (e.g. Rubin, Wetzler & Nebes, 1986; Rubin &
Schulkind, 1997a,b). In addition, some studies find evidence of a second bump
occurring later in life (e.g. Fitzgerald, 1988; Conway & Hacque, 1999), perhaps
reflecting period effects (e.g. the assassination of President Kennedy; civil strife in
Bangladesh).
The distribution of autobiographical memories appears to be affected by
dementia and depression. For example, Fromholt and Larsen (1991) did a free
recall experiment using elderly subjects, half of whom were demented, in order to
identify features of autobiographical memory vulnerable to degenerative brain
processes. The researchers found demented participants remembered a higher
proportion of memories of transitional events that resulted in role changes (e.g.
starting school, getting married, having children) than the non-demented controls.
Demented participants retrieved fewer events than the control group, and
participants with more advanced dementia produced less detailed events. While a
bump of memories emerged for both groups, it was less pronounced for demented
participants. In a second free recall experiment, Fromholt, Larsen and Larsen
(1995) examined the distribution of autobiographical memories in depressed,
demented and normal subjects. They found normal subjects recalled significantly
more memories with more details than the other two groups, which did not differ
from one another. There was evidence of a reminiscence bump in all three groups,
25
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
however depressed subjects recalled proportionally more memories from the recent
past, and the memories were more negative.
Researchers have suggested that the distribution of autobiographical
memories is probably affected by how participants are led to sample their memories,
and that carefully designed experiments may artificially constrain the contents of the
store. Autobiographical memory research may be more ecologically valid than
memory studies using Ebbinghaus’ nonsense syllables, however “(n)ot enough
importance has been given...to the inherent difference between reminiscence as
freely expressed and reminiscence as prompted or otherwise framed by the
researcher,” (Webster and Cappeliez, 1993, p. 65).
For example, it is possible that some experimental designs may cause
participants to get “locked in,” with a memory from one period cuing others from
that same period (Howes & Katz, 1992; Jansari & Parkin, 1996, p. 88). Participants
were either cued or asked to spontaneously recall autobiographical memories from
across the life span in an experiment by Howes and Katz (1992). In addition, they
were asked to recall memories in either a forward (remote to recent events) or
backward (recent to remote events) condition. The distribution was affected by the
order instructions in the cued, but not the spontaneous condition, with participants
asked to recall in the forward condition producing more events from childhood.
The authors concluded that while the effect may not be robust, their results suggest
participants can “wear out” (p. 112) in some experimental conditions. Jansari and
26
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Parkin (1996), however, did not find an effect of forward or backward presentation
condition in their study.
Alternately, participants could produce memories from a particular period if
they perceive an experimental preference for those memories. Rubin and Schulkind
(1997a) found the modification of instructions did affect the distribution of
memories reported. Two groups of 20 elderly participants were provided standard
instructions and one additional group of 20 elderly participants was provided what
the researchers called biased instructions. The two groups provided standard
instructions were told they were participating in a memory experiment and cued for
events, whereas the biased group was told they were participating in an
autobiographical memory experiment and was cued for memories. All participants
were encouraged to report episodes from “any point in time in your life” that
occurred “at a very particular place and point in time.” However, whereas the
standard instructions for the two unbiased groups said the point in time might have
occurred “even as recently as this morning,” and that participants “might think of
having gone to the hardware store yesterday,” the corresponding instructions for the
biased group said the point in time might have occurred “even from as far back as
you can remember” and that participants “might remember having gone to a little
country store with your grandfather when you were five.” The researchers found the
distribution skewed towards recency for the groups provided the standard
instructions and towards childhood for the group provided the biased instructions.
27
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
A bump emerged in both conditions, however it appeared to be larger and peak
earlier in the biased group. Unfortunately, not all researchers provide such detailed
accounts of the instructions used, and comparing and contrasting the peaks and
locations of the bump is confounded by this artifact.
There is empirical evidence of considerable individual differences in the
distribution of autobiographical memories. For example, in one experiment
designed to determine whether the common practice of averaging over people was
an artifact that obscured individual differences in their retention functions, Rubin
(1982) cued 10 undergraduates with 326 words and found that the slopes varied
considerably. In a second experiment, Rubin (1989b) compared the distribution of
autobiographical memory for two 70 year olds and found considerable
discrepancies. One subject overwhelmingly favored recent memories, and the other
recalled the largest number of memories from when she was 11-20 years of age.
Rubin and Schulkind (1997b) also found evidence of individual differences in a
third experiment. Five young adults in their twentieth year and five older adults in
their seventieth year were cued for a total of 921 words over the course of nine
sessions. Once again the slopes of the retention functions for the latest 20 years
varied. He noted that in the older group one participant had more memories in her
bump than the four others combined, and while all five participants exhibited a
bump, its location differed, peaking either at age 10-19 or age 20-29. Rubin and
Schulkind (1997b) concluded individual differences are likely to be the norm for
28
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
experiments drawing on different populations. Scholars agree that further
investigation of individual differences is merited, and Fitzgerald (1989) has called
for those investigations include contextual as well as cognitive factors.
The process of identifying mediators of the bump has begun. For example,
Franklin and Holding (1977) and Sperbeck, Whitboume and Hoyer (1986) were
interested in the relationship between personality, response latency and the age of
the reported autobiographical memories. Used the cuing method to prompt subjects
ranging in age from 25 to 74, Franklin et al. (1977) found extroverts responded
slightly faster and that older participants were slightly more introverted. However,
neither latency nor personality had an effect on the average age of the memories.
Sperbeck et al. (1986) used a modified cuing method to prompt subjects 25 to 85
years of age. They found that individuals with a higher openness score on subscales
measuring fantasy, esthetics, feelings, actions, ideas, and values recalled a greater
number of memories. These individuals also recalled more recent events. There
was no correlation between age and experiential openness, suggesting this
personality trait does not explain the increased remoteness of the autobiographical
memories recalled by older participants.
Meanwhile Brady Elnick, Margrett, Fitzgerald, and Labouvie-Vief (1999)
hypothesized that the number of memories about family and relationships from the
bump era would reflect socio-emotional development. The variables examined
measured: 1.) participants’ satisfaction with their current family and family of
29
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
origin; 2.) immature (e.g. passive aggression and acting out), neurotic (e.g.
inhibition and withdrawal) and mature defensive styles (e.g. task-orientation and
humor) in dealing with psychological conflict; 3.) secure, dismissing, preoccupied
and fearful attachment styles; and 4.) more and less integrated descriptions of the
self. The researchers found individuals who reported more memories about family
and relationships scored higher on positive current family climate as did those who
scored lower on immature defensive style. These findings suggest better adjusted
individuals are more likely to recall memories about families and friends from the
period during which their identities were being shaped.
Very little of the research on the distribution of autobiographical memories
has explicitly focussed on gender. However, a broad range of gender-related
differences regarding other aspects of autobiographical memory have been
identified. For example, researchers have generally found that women (and also
first bom or older children) report more memories from early childhood (e.g. Davis,
1999; Friedman & Pines, 1991; Mullen, 1994; Cowan & Davidson, 1984). In regard
to what adults remember, Cowan and Davidson (1984) reported significant gender
differences in the themes recalled from childhood, with men and women
remembering more competence-related and attachment-related events, respectively.
Seidlitz and Diener (1998) found women recalled more positive and negative life
events than men. The work of Brady Elnick and her colleagues (1999) indicated
that women report more significant life events, and that their memories fall into
30
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
somewhat different content domains than similarly aged men. Observed differences
in a study by Mackavey, Malley and Stewart (1991) suggested men’s
autobiographical memories reflected a greater sense of the importance of events in
one’s life. Webster’s (1994) work showed that women were more likely to use
autobiographical memories to create ease of conversation and to keep alive the
memory of a loved one than men.
Several explanations, each of them pointing to the contextual dependence of
autobiographical memory, have been proposed to account for the observed gender-
related differences. One group of studies suggest that families socialize their male
and female children to reminisce differently (e.g. Fivuch, 1998; Buckner & Fivush,
2000), perhaps affecting what adult men and women remember. Fivuch (1998)
found that mothers and fathers elaborated more, provided more feedback, and
discussed emotional aspects of the past more often with their young daughters.
Buckner and Fivush (2000) showed that parents discussed more social events with
girls than boys. It has also been proposed that gender effects are cultural. For
example, MacDonald, Uesiliana and Hayne (2000) found that New Zealand Asian
females report extremely late memories in comparison to New Zealand European
adults, and that women from New Zealand’s European, Maori and Asian cultures all
contained more information than the memories reported by men. Similarly, Wang,
Leichtman and their colleagues extended this concept to firstborn and only children;
31
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
in their 1998 paper they hypothesized that autobiographical memory would reflect
how children are socialized depending on family size and birth order.
Two recent studies by Rubin and his colleagues (1999, 2000) suggest these
gender differences may not extend to the bump and the distribution of
autobiographical memory, however. In their 1999 study, Rubin and his colleagues
asked equal numbers of older men and women, half of whom were 70, the other half
73 years of age, to produce autobiographical memories in response to 124 word
cues, and to list five important memories. Surprisingly, Rubin found a tendency
(not significant) for males to have more memories in the first seven years, but no
other gender effects using either method. In his second paper, which presents a
summary reanalysis of nine studies on the distribution of childhood memories in the
first decade of life, Rubin also found no gender differences, regardless of the
method used by the primary researcher.
II.C. EXPLAINING THE BUMP
A number of theories have been proposed to account for the bump. One of
the goals of this dissertation was to use the pattern of findings to compare and
contrast cognitive and non-cognitive explanations. It has been pointed out that the
theories need not be mutually exclusive (Fitzgerald, 1986; Rubin, Rahhal & Poon,
1998). More specifically, Rubin and his colleagues suggested that the non-cognitive
self-narrative theory may operate through mechanisms identified in the cognitive
32
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
rapid change theory. The theories and their mechanisms may also be relevant to any
cultural or gender based differences that are identified.
II.C.l. Cognitive theories of the bump
Il.C.l.a. Rapid Change
It is novelty that privileges bump memories according to one cognitive
theory. The theory suggests the bump coincides with periods of rapid change during
which young people typically have many novel experiences, and with normative
periods of transition as more permanent adult roles and responsibilities are
established (Rubin, Rahhal & Poon, 1998). Rubin identified several mechanisms
favoring memory for novel events, especially those occurring at the beginning of
periods of stability. Novel events are privileged because they are distinctive,
because individuals expend greater cognitive effort to understand them, and because
there is little proactive interference. In addition, the career and lifestyles decisions
made during the transitional bump period may be echoed in many day-to-day
experiences during subsequent periods of stability, and these reminders are likely to
cue the memories for retrieval and rehearsal on a regular basis. While it is proposed
as an explanation for the bump of autobiographical memories observed during
adolescence and young adulthood, the theory could also be applied to the encoding
of memories in any period of rapid change and transition. For example, additional
years of schooling could shift the location of the bump upward if they cause people
to delay normative events such as marriage and beginning a career (Rubin &
33
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Schulkind, 1997a). Differences in the size or location of the bump reflecting
historical events such as war or women’s suffrage could also be linked to the
novelty of the associated experiences.
Unfortunately, the scales of novelty, uniqueness and rehearsal used to test
the theory are retroactive assessments and suffer from the same weaknesses as other
self-reported measures. Furthermore, empirical evidence as to whether memories
from the bump years differ along these dimensions is mixed. Jansari and Parkin
(1996) asked participants to categorize each autobiographical memory reported for
frequency of spontaneous rehearsal and as a “first” or “non-first” experience. While
they found no significant effects of life period on rehearsal, more first-time or
unique experiences were recalled from early life than from midlife. They concluded
that the first-time memories from early life accounted for the bump, and that the
lower number of first-time memories reported from midlife accounted for the
asymptote that followed. However, when Rubin and Schulkind (1997a) asked
participants to rate the autobiographical memories reported on seven-point scales
for frequency of rehearsal, novelty, significance, and vividness, they reported that
bump memories did not consistently differ from memories from other life periods.
The authors concluded that other unspecified biological or environmental factors
may cause memories from the bump to be encoded in a way that makes them easier
to retrieve. Similarly, Conway and Hacque (1999) found that memories from the
34
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
bump were neither better rehearsed nor more novel than memories from other life
periods.
II.C.l.b. Cognitive Abilities
A second theory suggests that the encoding, storage and retrieval of
autobiographical memories reflects the ebbing and flowing of cognitive abilities
across the life span. According to this theory, more autobiographical memories
should be available for reporting from periods of peak cognitive efficiency than
from periods characterized by decline.
Episodic and working memory have been linked to the distribution of
autobiographical memory (e.g. Rubin, 1998; Nelson, 1992; Jansari & Parkin, 1996).
From a structural perspective, episodic memory is thought to encode, store and
retrieve autobiographical memories. From a resource perspective, the central
executive component of working memory is thought to mediate the cyclical retrieval
of autobiographical memories (Conway & Rubin, 1993). These constructs both
exhibit age differences favoring younger people when tested in the laboratory
(Zelinski & Bumight, 1997; Nyberg, Backman, Emgrund, Olofsson & Nilsson,
1996; Salthouse, 1991; Salthouse & Babcock, 1991; Hultsch, Hertzog, Dixon, &
Small, 1998). Age-related atrophy affecting the frontal lobes appears to be a
significant factor in explaining these types of age differences (Schacter, 1996).
Neurophysiological data from a positron emission tomography study (Conway,
Turk, Miller, Logan, Nebes, Meltzer & Becker, 2000) and using an
35
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
electroencephalogram (Conway, Pleydell-Pearce, & Whitecross, 2001) suggest that
the frontal lobes also play an important role in the construction and integration of
autobiographical memories.
Rubin, Rahhal and Poon (1998) argue that while a theory based on cognitive
abilities could account for the rapid increase in the number of autobiographical
memories reported during the bump years when many cognitive abilities reach their
peak, the number of autobiographical memories reported appears to drop off faster
than do most cognitive abilities in the years that follow. However, there is also a
great deal of individual variability in performance on the memory measures
(Zelinski & Bumight, 1997; Daneman, 1987). In addition, women tend to perform
at somewhat higher levels than men on episodic memory tasks, a difference that
may extend into old age (Zelinski, Gilewski & Schaie, 1993; Herlitz, Nilsson &
Backman, 1997; Herlitz, Airaksinen & Nordstrom, 1999), and education is related
to working memory, with higher levels predicting better performance (Hultsch,
Hertzog, Dixon & Small, 1998). Thus the cognitive abilities theory could gain
some support if individuals of all ages who do well on memory tasks, younger
participants, women, and those with higher levels of education report a
disproportionate number of autobiographical memories from more recent periods.
36
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
II.C.2. Socio-contextual theories of the bump
II.C.2.a. Self-narrative
A third theory is based on the importance of a select group of memories to
the individual’s identity, and to what Brewer (1986) might call the individual’s self
schema. According to this theory, forging a sense of self is a normative
developmental task that results in the privileging of certain memories. Fitzgerald
(1988, 1996) essentially agreed with Erikson that identity formation is typically a
task of late adolescence and early adulthood. He suggested individuals tend to
construct a unique experientially-based self-narrative during this period. The
process, which entails revisiting memories and assessing their importance to the self
concept, is a difficult one. Therefore, once the narrative is established, it tends to be
quite stable, and the memories selected continue to benefit from rehearsal.
Erikson’s theory left room for cultural differences in trait emphases, stage
lengths and the accommodations made for developmental readiness (Hoare, 2002).
However, it has been criticized for underestimating cultural variations in the sense
of self that could affect the order and/or meaning of the stages. In Erikson’s theory,
the identity formed during adolescence is a product of a constellation of
independently reached decisions. Meanwhile the intimacy vs. isolation, and
generativity vs. stagnation stages, which emphasize the individual’s connection with
others, follow and depend upon successful resolution of the identity vs. confusion
stage. Markus and Kitayama’s (1991) work suggests this model may be more
37
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
applicable to Western cultures. It involves a construal of self that:
...requires constructing oneself as an individual whose behavior is
organized and made meaningful primarily by reference to one’s own
internal repertoire of thoughts, feelings, and action, rather than by
reference to the thoughts, feelings and actions of others, (p. 226)
On the other hand, in East Asian societies there is an emphasis on relationships and
group memberships based on the “normative imperative” of maintaining
interdependence (p. 227). Achieving this construal of self requires:
...seeing oneself as part of an encompassing social relationship and
recognizing that one’s behavior is determined, contingent on, and, to a
large extent organized by what the actor perceives to be the thoughts,
feelings, and actions of others, (p. 227)
Markus and Kitayama (1991) propose these divergent approaches towards the
bounderies of the self have a variety of behavioral consequences. They could also
affect the distribution of autobiographical memories. More specifically, the bump
might occur later or appear to be flatter if the sense of self is linked not to one’s
unique occupational, political and religious identity independent of others, but to
intimate relationships.
Culturally and historically linked variations in the peak and location of the
bump have been identified. Benson, Jarvi, Arai, Thiebar, Frye, and Goracke
McDonald (1992) found a bump when they asked Japanese and midwestem
American subjects for 10 vivid memories. While the bump peaked in the second
decade for Americans, for Japanese participants it peaked in the third decade.
Generational differences in the bump based on shared cultural experiences affecting
38
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
identity formation are also possible. While studying the autobiographical memories
of young and older Bangladeshi participants, Conway and Hacque (1999) found
evidence of not one, but two bumps. In response to 15 common cue words, the
older group reported a much smaller bump from the period of adolescence and
young adulthood, and an unexpected bump of memories from when these
participants were between about 35 and 55 years of age. The latter corresponded to
the beginning of what became a violent national independence movement, and what
was referred to as a defining moment for a group of individuals.
Markus and Kitayama (1991) argue the distinction between independent and
interdependent self construal represents “one of the most general and overarching
schemata” (p. 230), and that it has implications for gender differences as well.
Bakan’s (1966) agency and communion, the “two fundamental modalities of living
forms,” correspond to Markus and Kitayama’s constructs. Bakan described the
modalities as follows:
Agency manifests itself in self-protection, self-assertion, and self
expansion; communion manifests itself in the sense of being at one with
other organisms. Agency manifests itself in the formation of separations;
communion in the lack of separations. Agency manifests itself in
isolation, alienation, and aloneness; communion in contact, openness, and
union. Agency manifests itself in the urge to master; communion in
noncontractual cooperation. Agency manifests itself in the repression of
thought, feeling and impulse; communion in the lack and removal of
repression, (p. 15)
Bakan concluded that whereas agency is characteristically masculine, communion is
characteristically feminine. It has been suggested that these modalities can be
39
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
measured using the personality facets called assertiveness and tender-mindedness,
respectively (Feingold, 1994).
Bakan’s distinction was supported by Gilligan’s (1982) finding that the
women she studied did not choose to describe themselves in terms of their
occupations, but in terms of their relation to others and responsibilities towards
others. Similarly, Ruth and Vilkko (1996) found that women’s explanations of how
their lives evolved show a “relational self,” often referring to “we” instead of “I,”
and often describing themselves as objects rather than subjects in their narratives
(p. 170). Gilligan concluded that “the sequential ordering of identity and intimacy
in transition from adolescence to adulthood better fits the development of men than
it does the development of women” (p. 163). The stability and continuity of
relationships could lead to gender differences in the distribution of memories, with
those who define themselves primarily in terms of their ongoing, evolving bonds
and social responsibilities being more likely to generate a more even distribution of
memories from across the lifespan.
An alternate approach, which looks to structural explanations for behavior,
can be used to arrive at similar predictions regarding gender-based differences in the
distribution of autobiographical memory. Eagly argues that social roles, and not
prior socialization or biology, are the most important and proximal predictors of
behavior (1987). Social roles became increasingly gendered as a result of
industrialization and urbanization in the nineteenth and twentieth centuries, with
40
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
men taking paid jobs outside the home in order to support their families, and
women staying at home to look after the household (Day Hulbert, 1993). The
women’s movements of the twentieth century had a tremendous impact on the
choices available to women, however sex roles continue to differ. Today higher
proportions of women still provide care for children, spouses, parents, in-laws, and
friends (Montgomery & Datwyler, 1992). Women also continue to serve as
kinkeepers, maintaining communication and monitoring family relationships
(Hagestad, 1986). Furthermore, they are taught to expect that these roles will be
subject to change (Sales, 1978). For Eagly, agency and communion represent
stereotypes that reinforce these social roles and personality traits, and hence the
status quo.
Moen’s (1995) summary of the literature pointed out some ways in which
responsibilities associated with women’s social roles may result in a different
pattern of turning points. For example, because of their caregiving and kinkeeping
responsibilities, women are more likely to move in and out of the labor force than
men, whose work role, once established, is more stable. As a result, turning points
associated with work and retirement may differ and/or not have the same
significance. Women may also focus more attention on establishing non-family role
identities after their children leave home, leading them to transition back into jobs
or school during adulthood. Also because of their work histories and the pivotal
importance of the domestic role, the turning points of widowhood and divorce may
41
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
have different implications for the economic and psychological well-being of
women. If turning points are privileged by novelty, effort for meaning and
rehearsal, these gender differences should be reflected in the distribution of
autobiographical memory. That is, based on the discontinuities inherent in their
social roles, women might be expected to generate a considerably flatter
distribution, especially during adulthood.
II.D. HYPOTHESES
The central questions guiding the research are: 1.) Does the bump endure
into the ninth decade and beyond? 2.) Is the bump immune to instructional
modifications of free recall tasks? and 3.) Is there evidence of the bump in both
genders? Based on the literature, no differences in the distribution of memories
distinguishing the oldest old from younger participants, or those writing about
people or events/decisions were expected. However, it was anticipated that women
would generate a flatter distribution of memories than men. More specifically, the
following relationships were hypothesized:
1.) Participants 80 years and older generate a bump of memories corresponding
to late adolescence and young adulthood;
2.) Younger and older participants are equally likely to report memories from
the bump years;
3.) Participants writing about individuals who have been important in their lives
generate the largest proportion of memories from the bump years;
4.) Participants writing about events/decisions that have been important in their
lives generate the largest proportion of memories from the bump years;
42
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
5.) Participants writing about people and events/decisions are equally likely to
report memories from the bump years;
6.) Men generate the largest proportion of memories from the bump years;
7.) Men writing about individuals who have been important in their lives
generate the largest proportion of memories from the bump years;
8.) Men writing about events/decisions that have been important in their lives
generate the largest proportion of memories from the bump years;
9.) Women produce similar proportions of memories from childhood, the bump
and adulthood;
10.) Women writing about people who have been important in their lives
produce similar proportions of memories from childhood, the bump and
adulthood;
11.) Women writing about events/decisions that have been important in their
lives produce similar proportions of memories from childhood, the bump
and adulthood;
12.) Men are more likely than women to report memories from the bump years;
13.) Men writing about individuals who were important in their lives are more
likely to generate memories from the bump years than women writing on
that topic;
14.) Men writing about events/decisions that were important in their lives are
more likely to report memories from the bump years than women writing on
that topic;
Also based on the literature, it was anticipated that more assertive
individuals and those with more education would be more likely to generate
memories from the bump. Meanwhile tender-minded individuals and those with
better memory abilities were expected to produce more memories from childhood
and/or adulthood. The following subsidiary hypotheses were tested:
43
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
15.) Those who have more education are more likely to produce memories from
the bump;
16.) Those who are more tender-minded are more likely to produce memories
from childhood and adulthood;
17.) Those who are more assertive are more likely to produce memories from the
bump;
18.) Those who score well on episodic memory measures are more likely to
report memories from adulthood;
19.) Those who score well on working memory measures are more likely to
report memories from adulthood.
44
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CHAPTER III. RESEARCH DESIGN
III.A. SAMPLE
Data in this dissertation come from the 1994-1995 and 2000-2002 panels of
the Long Beach Longitudinal Study. The study is designed to assess memory
change in adults, identify mechanisms of change with individual differences
models, and extend models of change to a large sample of the oldest-old (Zelinski,
1999). The study sample includes adults of all ages, predominantly residing in
California’s Los Angeles, Orange and San Bernardino counties. Subjects tested in
1994-1995 were randomly selected from membership rosters of the Family Health
Plan (FHP), a health maintenance organization based in Southern California until it
was sold. Subjects tested in 2000-2002 were subsequently recruited from a program
serving seniors in Long Beach, through newspaper advertisements and a direct mail
campaign.
Participants who joined the study in 1994 completed a battery of paper-and-
pencil memory and intelligence tests in two 2-3 hour sessions. The protocol was
extended to include several computerized reasoning tasks for the 2000 testing, so
that subjects completed an additional 3 hour session. Participants who joined the
study in 1994 were predominantly tested in a small group format at Southern
California FHP facilities. Participants joining in 2000 were also tested in small
group formats. FHP facilities were no longer available, however, so that the
sessions were conducted in area hospitals and libraries equipped with appropriate
45
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
facilities. Homebound participants in both panels were accommodated in private
sessions held at their convenience in their homes. Finally, subjects in both panels
were asked to complete a mail-in packet of questionnaires designed to gather health
behaviors, physical activity and personal data, and measure personality,
autobiographical and self-reported memory.
Autobiographical memory data for this study were available for 845 subjects
aged 32-95. Of these, 465 were tested in 1994-1995 (Experiment 1) and 380 were
tested in 2000 (Experiment 2). Absent experimental restriction of memory
sampling, the bump reliably emerges at some point in the sixth decade of life
(Rubin, Wetzler & Nebes, 1986), and participants less than 55 years of age were
excluded. Sixty-six participants in Experiment 1 and 22 participants in Experiment
2 fell into this category. The responses of another 70 Experiment 1 participants and
31 Experiment 2 participants were impossible to code, primarily because the
memories were too general to confidently locate them within a single life period.
Approximately 68 percent (n=224) of the participants from Experiment 1
provided the specific age of the memory they reported. Research assistants were
able to contact another 22.5 percent (n=74) by telephone. These participants were
asked to date the memory they reported after the protocols were read back to them.
The files of 9.4 percent of the participants (n=31) provided information about the
age of the memories. Ultimately these procedures allowed for the precise dating of
329 protocols, 251 of whom reported memories from the bump years (18-34) and
46
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
the periods immediately surrounding it (i.e. 0-17 and 35-51). Experiment 2
participants were specifically asked to identify the age of the memory they reported.
As a result, a higher percentage provided this information. Approximately 81
S Exp. 1
(n=329)
B Exp.2
(n=327)
Figure III.1
Age of M em ories (Participants 55 and O lder)
0 \ OS CO t-*- * —I w-i
m m •* ? t j -
Age at lim e of Memory
percent (n=265) of the participants reported the age of the memory in their protocol,
and 62 of the participants (19 percent) who did not were successfully contacted. In
all, 327 of the protocols were dated, and of these 268 participants reported from
childhood, the bump and adulthood. Figure III. 1, which illustrates the overall
distribution of the autobiographical memories, indicates the existence of a bump of
memories peaking at 18 years of age in both Experiments.
47
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
III.B.l. Autobiographical Memory
A measure of autobiographical memory was included in the mail-in packet
of questionnaires in 1994-1995 and 2000-2002. In Experiment 1, participants were
asked to compose an autobiographical paragraph on a lined sheet of paper. The task
did not require the paragraphs be dated. The instructions, which offered
participants a choice of topics, were as follows:
We would like to know a little about your life. On the lines below,
please write a paragraph about a decision, event or person that
influenced your life. Choose one of the topics and tell us about the
decision, event or person, and how your life was influenced. You do
not need to fill the entire page - a few sentences will do.
The measure was designed to test for developmental changes in language
production, and analyses of the data gathered in 1994-1995 suggested participants
were approaching the task differently based on the topic they chose. Instructions
were changed in Experiment 2 in order to study whether topics would produce
different distributions of autobiographical memories, and whether random
assignment to a topic would affect the bump. As before, participants were asked to
compose an autobiographical paragraph on a lined sheet of paper. However, this
time they were also asked to date the memory they described, and were presented
one of two versions. Instructions, which limited the participants’ choice of topic,
were as follows:
48
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Version A: We would like to know a little about your life. On the
lines below, please write a paragraph about either 1) a decision you made
that had a major influence, or 2) an event that influenced your life. Choose
one of the topics and tell us about the decision or event, and how your life
was influenced. You do not need to fill the entire page - a few sentences
will do.
Version B: We would like to know a little about your life. On the
lines below, please write a paragraph about a person who had a major
influence on your life. Tell us about the person, and how they influenced
your life. You do not need to fill the entire page - a few sentences will do.
A coding manual was designed for the purpose of analyzing the responses
(See Section III.C.2.).
III.B.2. Personality.
Participants completed the NEO-PIR, a self-reported measure assessing
personality (Costa & McCrae, 1992). The NEO-PIR measures 30 facets defining the
Big Five personality traits (Neuroticism, Extroversion, Openness, Agreeableness,
and Conscientiousness), including assertiveness, a facet of Extraversion, and tender
mindedness, a facet of Agreeableness. These two facets are assessed with eight
items each as follows:
49
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
E4 Assertiveness:
I am dominant, forceful, and assertive
I sometimes fail to assert myself as
much as I should.
I have often been a leader of groups I
have belonged to.
In meetings, I usually let others do the
talking.
Other people often look to me to make
decisions.
I would rather go my own way than be
a leader of others.
In conversations, I tend to do most of
the talking.
I don’t find it easy to take charge of a
situation.
A6 Tender-mindedness:
Political leaders need to be more aware
of the human side of their policies.
I’m hard-headed and tough-minded in
my attitudes.
We can never do too much for the poor
and elderly.
I have no sympathy for panhandlers.
Human need should always take
priority over economic considerations.
I believe all human beings are worthy
of respect.
I have sympathy for others less
fortunate than me.
I would rather be known as “merciful”
than as “just.”_____________________
Assertiveness and tender-mindedness are used to measure, respectively, the agentic
and communal orientations to life as proposed by Bakan (1966). The agentic, or
instrumental orientation is considered characteristically masculine. The communal,
or expressive orientation is characteristically feminine. Assertiveness and tender-
mindedness are nearly pure measures of the agentic and communal orientations
(Feingold, 1994). As described by Costa and McCrae (1992), individuals who score
high on assertiveness are dominant, forceful, and socially ascendant as compared to
low scorers, who prefer others do the talking and tend to keep in the background.
Individuals who score high on the tender-mindedness scale have sympathy and
concern for others, are moved by their needs and believe social policies should
serve those needs. Low scorers, who consider themselves realists, are more
hardheaded and prefer making decisions based upon logic.
50
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
III.B.3. List Recall.
In this classic test of episodic memory, subjects were asked to study two
typed lists for 3.5 minutes each, one per paper-and-pencil testing session. Each list
contained 20 concrete high-frequency one- or two-syllable English nouns.
Session 1.
home, flag, bird, ocean, dirt, woman,
exam, kettle, tank, painter, lemon, jury,
star, money, alcohol, vest, iron, rattle,
garden, church____________________
Session 2.
air, film, wagon, hall, yam, iris, barrel,
jacket, rice, depot, letter, map,
emperor, novel, parade, satin, gate,
organ, tape, umbrella______________
They were asked to free recall the word lists in writing approximately 20 minutes
later. There was no time limit for recall, which was scored as the proportion of
correctly recalled items.
III.B.4. Syntax Memory.
In this test of working memory, subjects were asked to listen carefully as a
series of sentences was read to them. After each sentence had been read,
participants were first asked to rate the sentence as either grammatical, somewhat
grammatical, or ungrammatical. They were then asked to reproduce the sentence in
writing. Grammatically correct sentences were constructed to be more or less
difficult to process, depending upon particle placement. For example, the first
version of the following sentence, in which the particle is placed immediately
following the verb with which it is associated, is considered easier to process than
the second, in which the particle is placed at the end of the sentence:
51
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Version 1: The worried new father
picked UP the loudly protesting baby
who was wet from her bath.
Version 2: The worried new father
picket the loudly protesting baby who
was wet from her bath UP.
The proportion of correctly recalled propositions was determined for each sentence,
and two variables were computed, one averaging the easy, and one averaging the
difficult sentence scores.
III.C. CONTENT ANALYSIS
III.C.l. Introduction to the Data
Participants were asked to share significant personal memories. In response,
they composed passages ranging from one word or phrase retorts (e.g. “Myself,”
“No one,” or “I don’t do this sort of thing”) to more or less well organized, multi
paged essays (e.g. the 348-word tale of a husband and wife who started a new
business after losing everything they owned to a fire sparked by a barbecue).
Perhaps because of the nature of the sample and task - a comparatively well-
educated and healthy population of volunteers donating a day or two of their time to
the study of memory - there were relatively few personal disclosures of traumatic
events (3 and 1 percent in Experiments 1 and 2, respectively). These paragraphs
(e.g. about the pregnant teenager who attempted an abortion, the man who lent a
relative money and ended up poor in old age, the destitute mother who was talked
into giving her children up for adoption, the schoolboy who was made to clean
latrines at his school because he was Native American, the mother who took a
52
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
mentally disturbed adult son back into her home and suffered the consequences)
stood out as the exceptions.
Transitions (e.g. “In 19411 was working as a draftsman for the South
Dakota Planning Board. I had a wife and two small children. Pearl Harbor came
along and it changed everyone’s plans...”) - often related to education (7 and 3
percent in Experiments 1 and 2, respectively), work (9 and 12 percent), residence
(11 percent in both experiments), war (9 and 7 percent), home life (8 and 12
percent), religion/ spirituality (5 and 3 percent) - and figures from those transitions
(e.g. “He was a man that I worked with while in a remote oil field camp in
Venezuela. I had been out of engineering school for just a short time and he helped
me adjust to practical engineering...”) were prominent themes.
Many participants (31 and 42 percent in Experiments 1 and 2, respectively)
wrote memories about or relating to people, “My father...”, “My mother...”, “My
wife...”, “My husband...”, “My children...”, or “My friend...” Often these
paragraphs mentioned wisdom and kindness, or provided examples of love and
loyalty (e.g. “The person who has made a major influence on my life is my wife of
69 years. She has brought a new life of love and companionship, and understanding
to m e...”).
A smaller number told stories of loss by death (e.g. “It was the death of my
husband in 1953..6 and 3 percent in Experiments 1 and 2, respectively) and
divorce (e.g. “Divorced when I was 30. .2 percent in both data sets). However, a
53
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
resilient tone characterized most of these paragraphs. The widow, for example,
rejected a life of dead-end jobs that limited time with her children and “went back to
college and obtained my degree...” The divorcee remarried nine years later and
concluded, “We have had a wonderful life for the past 38 years.”
III.C.2. Content Analysis Procedures
The sampling units for the autobiographical memories are the written
paragraphs submitted by participants who joined the Long Beach Longitudinal
Study in 1994 or 2000. As previously described, participants in Experiment 1 were
asked to submit a memory about a person, event or decision that had a major
influence on their lives. Each discrete memory from the past or present comprised a
recording unit. Task instructions changed in Experiment 2, with participants
specifically assigned to write about either a person, or an event/decision. In
addition, Experiment 2 participants were specifically asked to provide the age of the
memory, not a requirement in Experiment 1.
The two research assistants coding the autobiographical paragraphs followed
the guidelines described below, an expanded version of those developed for a pilot
study presented at the scientific meeting of the Gerontological Society of America
in 2000. While some of the examples used were drawn directly from paragraphs
submitted by participants, others were crafted to illustrate particular issues thought
likely to arise. The research assistants worked independently using scoring sheets.
Reliability was calculated as the percentage of agreement between coders on two
54
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
percent of randomly selected protocols (n=12). Percent agreement for the
identification of topic (i.e. person or event/decision) was 100 percent, for age of
memory 91 percent, and for content domain 75 percent. Coding differences were
resolved through discussion.
III.C.2.a. Identifying the Recording Units
The coders’ first task was to identify the recording units to be analyzed and
enter the identifying text into the data file verbatim. Both “flagged” and
“unflagged” recording units were analyzed for this study. In flagged memories, the
participants specifically indicated which person or event/decision they were writing
about. For example, a subject might begin a paragraph by saying, “The most
important person in my life was my father...” In other cases, however, the
recording unit was not flagged. For example,
I was 15 when I met Mr. Wright. He was the creative writing teacher
at my high school and he taught me the basics of how to craft stories.
He was always available to critique my work.
Here the participant included information about a person (the teacher) and an event
(meeting the teacher). The research assistants were asked to read such passages
closely, and to ask themselves what person or event/decision the participant was
suggesting exerted influence on his or her life. As part of this process, the coders
indicated whether the recording unit referred to a person or event/decision. This was
only an issue in Experiment 1 protocols, because participants had been allowed to
55
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
choose between topics. In the unflagged example, a careful reading suggests the
participant sees the person rather than the event as having a major influence.
Some participants did not follow directions and wrote about more than one
recording unit. Fifty-three participants provided two memories, 17 provided three
memories, and 4 provided four memories in Experiment 1. The corresponding
numbers in Experiment 2 were 64, 23 and 3. While all of the additional memories
were scored, this dissertation is based on the first recording unit mentioned. The 20
Experiment 1 and the dozen Experiment 2 paragraphs that were returned blank or
contained more than four recording units were not scored. Research assistants were
also instructed to ignore brief unflagged mentions of one sentence or less about
extraneous subjects. In the following passage, for example, the subject focuses on
one parent but mentions the other, almost as an afterthought:
My father was very important. He decided to teach us right from wrong
by example. He truly lived his life using lessons learned from the Bible,
explaining his decisions to us in that way. My mother helped him.
The final statement is not flagged and would not be counted as an recording unit.
The research assistants were also instructed that multiple recording units within a
single paragraph should be entirely independent. This is the case in the following
paragraph:
My four children are a great influence on my life. I also have to include
the day I married my wife (1933) as a pivotal moment.
The correct coding would include two recording units, one about a person, the other
about an event/decision. In some instances, participants reported on a memory and
56
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
then expanded on that theme. The research assistants were instructed to identify the
originating source of the recording unit, not developments resulting from it. For
example,
Following surgery after an on the job injury, I was forced into early
retirement in 1970. Psychologically I was not ready for that, so I became
very depressed.. .1 enrolled in a clay sculpture class.. .That is the secret;
get involved. Another important thing that happened to me more recently
has been the death of my wife. I am lucky to have children and friends to
see me through.
In this case, the statement contained one unflagged and one flagged recording unit.
The unflagged recording unit can be described as event-based decision - the
participant’s work-related injury initiated a chain of events - and would be coded
for the originating event. The flagged recording unit concerns the death of his wife,
an entirely independent event. Because it was the second memory mentioned, it
would be coded but not analyzed.
III.C.2.b. Determining Age
For recording units to be scored, a participant’s age at the time of the
memory was mentioned in or at the conclusion of the paragraph, was determined by
contacting the subject, coincided with datable historical events, or referred to an
event (e.g. marriage, childbirth, medical procedures), the age of which could be
deduced from other questionnaires in the participant’s file. Using these procedures,
a total 329 protocols from Experiment 1 and 327 protocols from Experiment 2 were
analyzed.
57
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
A series of categories - reflecting the approximate age of the subject at the
time the memory was encoded - was constructed. The categories, generally
corresponding to childhood and young adolescence (0-17), young adulthood
(18-34), adulthood (34-64), and older adulthood (65+), are empirically and
theoretically based. They allowed the research assistants to code recording units
spanning ranges of years using content analysis. The 18-34 period also overlaps late
adolescence and young adulthood, which the self-narrative theory suggests
corresponds to the development of a stable identity. Figure III.2 illustrates the
distribution of memories by category. Like Figure III. 1, this figure indicates a surge
of memories peaking during the bump years in both panels.
Mackavey, Malley and Stewart (1991), who also used content analysis
to study autobiographical memory, derived a somewhat different set of categories:
early childhood (less than 12), early adolescence (13-17), college years (18-21),
early adulthood (22-35), middle adulthood (36-50), and later adulthood (more than
50). The differences between the two systems reflect sample and data
characteristics. Mackavey and his colleagues relied on the written autobiographical
memories of uniformly and very highly educated, professional psychologists. The
Long Beach Longitudinal Study sample was more diverse. Only about a third of the
participants had college degrees. As a result, distinguishing between the college
years and early adulthood made little sense. In addition, the psychologists were
cued to produce longer autobiographies that were intended for publication. The
amount of detail provided in response apparently made it easier to differentiate
58
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
between early childhood and adolescence than in the Long Beach Longitudinal
Study.
V
W )
e Q
Figure III.2
Age of M;men»ries by Life Period (Participants 55 and older)
gE xp.l
(n=327)
0-17 18-35 35-64 65+
Life Period
Memories spanning two or more life periods were not included in
this study. Thus the following statement, which refers generally to the teenage
years, breaching two life periods, would be omitted: “I got married young, while I
was still in my teens.” Approximately equal proportions of memories spanning a
number of years within a single life period were reported in Experiment 1 (12.8
percent) and Experiment 2 (12.2 percent). For memories spanning even-number
ranges of years, the research assistants calculated the middle and subtracted one-half
year. For example, one participant wrote that, “My high school gym coach had a big
influence on me, encouraging me to do my best regardless of the odds.” High
59
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
school generally encompasses grades 9-12, and typically corresponds to ages 14
through 17. In this case, the mid point is 15.5 years of age, from which a half-year is
subtracted. For odd-number ranges, the research assistants simply calculated the
absolute middle year.
Finally, where no other clues were provided, recording units written
in the present tense were coded as referring to the current time and entered as the
participant’s current age. For example:
My wife is a wonderful woman. We always have a lovely time
together. Most recently we went on a trip to Italy. She is a great
companion.
This procedure was used in 24 (7.3 percent) Experiment 1 and 10 (3.1 percent)
Experiment 2 cases.
III.C.2.C. Content Domains
This study was designed to address research questions relating to the
distribution, rather than the content of autobiographical memories. However,
research assistants were asked to code the recording units for content based on a
system developed by Brady Elnick (1999). After analyzing a random sampling of
protocols, Brady Elnick and her colleagues settled on six broad categories ( i.e.
family/relationships, education/work, residence, social activities, health, and
miscellaneous). In the Long Beach Longitudinal Study, these were expanded into
13 categories as follows: family, friends/acquaintances, education, work, residence,
social activities, health/lifestyle, military activity/war, family life (including meeting
60
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
and marrying spouse, and birth of children), divorce, death, religion/spirituality, and
trauma/major national events. This data is described below. Statistical analyses are
planned for future publications.
Figures III.3 and III.4 illustrate the overall distribution of memories by
content domain. The largest proportion of participants in both panels reported
memories relating to their families (22 and 25 percent in Experiments 1 and 2,
respectively), and the same sizable proportion wrote about where they lived or
moved (11 percent each). A noticeably larger proportion of Experiment 2
participants wrote about friends (9 and 17 percent) and family life (8 and 12
percent), while more participants in Experiment 1 wrote about education
experiences (7 and 3 percent) and their health (6 and 2 percent).
F igure III.3
C o n te n t D om ains (E x p e rim e n t!)
M ilitary
9%
H ealth
6%
Socializing
3%
Residence
11%
D ivorce Death
2% 6%
Education
7%
Friends
9%
Religion
5%
Family Life
8 %
W ork
Family
22%
61
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure III.4
C o n ten t D om ains (Exp.2)
Socializing
2%
Health
2% Family Life
Education
3%
Worl
12 ° / c
M ilitary
Residence
1 1 %
Friends
17%
Traum a
1%
Divorce
2%
Religion
Death
3%
3%
25%
Table III. 1 describes the distribution of content domains by life periods. The
first number in each cell represents what percentage of the total number of
memories reported in each life period were derived from each content domain.
These data do not explain why participants chose to report from one life period and
not another. However, the distribution of memories does appear to reflect what
issues were normatively important during each period. For example, relatively large
proportions of the memories from childhood and adulthood were written about
family, life periods during which the individual is arguably most likely to rely on
relatives for social support. Larger proportions of memories from childhood than
the bump were written about friends, the bump representing a period during which
62
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table III.l Distribution of Memories by Life Period and Content Domain
Experiment 1 Experiment 2
Content Child Bump Adult 52+ Child Bump Adult 52+
Domain: n=71 n=118
I I
n=78 n=85 n=140 n=43 n=59
Family 50.7* 9.3 8.1 29.5 48.2 15.7 4.7 25.4
48.0 14.7 6.7 30.7 51.3 27.5 2.5 18.8
Friends 12.7 5.1 11.3 7.7 29.4 10.0 18.6 11.9
32.1 21.4 25.0 21.4 46.3 25.9 14.8 13.0
Education 5.6 11.9 6.5 1.3 3.5 5.7
— —
17.4 60.9 17.4 4.3 27.3 72.7 - -
Work 2.8 5.9 11.3 16.7 2.4 11.4 16.3 23.7
6.9 24.1 24.1 44.8 5.1 41.0 17.9 35.9
Residence 1.4 17.8 16.1 5.1 4.7 11.4 25.6 6.8
2.8 58.3 27.8 11.1 11.4 45.7 31.4 11.4
Socializing 4.2 1.7 3.2 5.1
—
2.1
—
5.1
27.3 18.2 18.2 36.4 - 50.0 - 50.0
Health 1.4 1.7 6.5 15.4 1.2
_
7.0 6.8
5.3 10.5 21.1 63.2 12.5 - 37.5 50.0
Military 5.6 20.3
— _
1.2 14.3 4.7
—
14.3 85.7 “ - 4.3 87.0 8.7 -
Family Life 2.8 15.3 4.8 2.6 2.4 23.6 2.3 5.1
8.0 72.0 12.0 8.0 5.1 84.6 2.6 7.7
Divorce
_
1.7 9.7
_ _ _
.7 9.3 3.4
- 25.0 75.0 - - 14.0 57.1 28.6
Death 8.5 1.7 12.9 6.4 2.4 1.4 7.0 6.8
28.6 9.5 38.1 23.8 18.2 18.2 27.3 36.4
Religion 2.8 4.2 8.1 6.4 4.7 1.4 4.7 5.1
11.8 29.4 29.4 29.4 36.4 18.2 18.2 27.3
Trauma 1.4 3.4 1.6 3.8
_
2.1
_ _
11.1 44.4 11.1 3.33 - 100. - -
*The first number in each cell represents what percentage of the total number of
memories reported in each life period were derived from each content domain. The
second number in each cell reports what percentage of the total number of
memories written about a specific content domain fell into each life period.
63
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
individuals are generally preoccupied with establishing families and homes of their
own. Finally, comparatively larger numbers of the memories from adulthood
related to work, perhaps reflecting changing goals and interests as children mature
and leave home.
The content of memories within the life periods also generally corresponds
with Erikson’s psychosocial theory of development. According to this theory,
relationships with parents and peers should be particularly important for children
and adolescents as trust, autonomy, initiative, and identity are learned. The ability
to form sexual and social partnerships becomes important in young adulthood as
individuals learn how to forge intimate sexual and social relationships. And the
sense of generativity, reflected in the desire to contribute to society as a parent,
worker or perhaps volunteer, can be a positive outcome of adulthood.
The second number in each cell of the table reports what percentage of the
total number of memories written about a specific content domain fell into each life
period. These figures indicate that the distributions can be peaked toward one life
period or another, or more evenly spread across the life periods. For example, the
distribution of family memories was clearly peaked toward childhood. That is,
when people wrote about family, they were writing from the childhood life period
about half the time. Memories written about education, military experiences, and
family life overwhelmingly came from the bump period, and memories about
divorce from the period following the bump. Meanwhile, the distributions of
64
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
memories about friends, socializing, death, and religion were relatively evenly
dispersed across the four life periods.
III.C.3. Examples of Coding Procedure
Approximately one percent of the total number of protocols analyzed for this
study was sampled using the SPSS Select Cases: Random Sample procedure. These
six cases will be used to illustrate how the coding guidelines described above were
applied. The paragraphs were transcribed exactly as submitted, including spelling,
grammatical and stylistic errors. A star is used to replace words that were illegible,
and the names of private individuals were indicated by ellipses.
Example 1. (86-year-old woman, 13 years schooling)
In 1945 my husband, a Navy Lt., died suddenly leaving me with 3 boys
ages 11 and 8yrs and a 6 wk old baby. That event certainly changed my
life “Big Time.” I went back to Oklahoma where both set of parents
lived - No training for any kind of work. A former teacher helped me
get a job in a Ladies Ready to Wear - A new Dr came to town and hired
me for his one girl office - He trained me in everything including ☆ +
Lab. - 1 was with him 3 yrs. Then put the boys in the car and headed
four Calif - 1 later ecame a nurse and the boys all turned out to be great
citizens!!!
Research assistants identified the originating source of the recording unit - the death
of the participant’s husband - a flagged event triggering a cascade of other events
and decisions. Because a date (1945) was provided, it was possible to calculate the
participant’s age (31). The content of the memory was coded as death of spouse.
65
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Example 2. (81-year-old man, 13 years schooling)
In 1954 I made a decision to come to California to go into a
business with my friend. A food processing venture
completely foreign to me - Very successful. #2 I was injured in
auto accident as a passenger. Age 61. I had a brain concussion
+ was in a coma 4 V z days. I came out of it with a definitely
new personality. Previous to the accident was very bashful and
returned ☆ very vocal and self assured.
This participant reported two recording units. However, only the first - the decision
to come to California - was analyzed for this study. Once again, the participant
provided the date of the memory (1954), allowing the research assistants to
calculate his age (41). The content of the paragraph was coded as work.
Example 3. (75-year-old woman, 13 years schooling)
I m et... in 1979. She told me about Yoga and meditation
a program that came on T.V., Instructor was Richer Hittleman.
It has helped me physically as well as mentally.
As in Example 1, the research assistants’ first task was to identify the originating
source of the recording unit. The participant from Experiment 2 was assigned to
write about a person, and chose to focus on the friend who directed her towards an
enduring new interest. The participant indicated the year (1979), which allowed for
the calculation of her age (59). The content of this autobiographical memory was
coded as people, other than family.
66
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Example 4. (79-year-old man, 16 years schooling)
I went into the insurance business because a fellow would not
take my no for an answer. The insurance business proved
excellent for me. My parents who made sure I learned the
value of money, and never to go into debt, to this day I do not
use credit cards. With very few exceptions, I pay cash. My
wife who worked right along side with me and proved that she
could reach all her goals.
Three recording units concerning people (colleague, parents, wife) were identified
in this protocol. Because no ages were specified, this participant was contacted by
telephone. After having the recording units read back, he said the memories
referred to when he was 39- 40, 9-12, and 24-79 years old, respectively. The lower
value of the short range (39) was entered for the first recording unit, and the content
of the memory was coded as people/other than family related.
Example 5. (75-year-old woman, 14 years schooling)
I always wanted to get married and have a family. I always
loved babies and longed for one. I was so thrilled when I got
pregnant + said I wanted a little girl - 1 had her when I was 23
+ 1 often told people thru the years she was exactly what I
wanted - if I could have ordered her from a catalog - She was
so lovely + good + smart + kind to everyone. People looked up
to her + ask her for help for so many things, but she was always
there for her Dad and me too. When I lost her 6 weeks ago my
world fell apart. I’m so alone now, all my family are gone. I
feel so alone - So this test is not me, as I usually am - at all. I
still have one very important person to me, my son, but he lives
in Hawaii + 1 don’t see him often, but we love each other very
much + talk quite often on the phone.
This participant from Experiment 2 highlighted the early years of her daughter’s
life. However, in the space provided at the bottom of the protocol, she explicitly
67
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
indicated she was reporting on a memory from the current year, the content of
which was coded as the death of child.
Example 6. (73-year-old man, 16 years of schooling)
In 1946,1 was studying engineering at Caltech. I had an
appointment to West Point and was told to report to the
military academy during that summer. Since World War II had
just concluded, I decided that it was not advantageous to spend
my life in the military. Therefore, I enlisted in the Army for
eighteen months. They sent me to Korea as part of the
occupation forces. And after my tour of duty, I returned to
Caltech & completed my engineering studies in 1950. The
Korean War started about the same time. I always remember
that the decision I made in 1946 had profound effects.
This paragraph begins with a reference to a decision made in 1946.
However, in the space provided at the bottom of the protocol, they indicated his
memory comes from between 19 and 23 years of age. The research assistants
followed the rule for short, odd-number ranges of years to arrive at age 21, and they
coded the content as work.
III.D. Analysis Procedures
Experiments 1 and 2 are presented in Chapters IV and V, respectively. The
bump, which can be obscured by recency effects (Jansari & Parkin, 1996), may or
may not exist in adults before the sixth decade of life (Rubin, Wetzler & Nebes,
1986). In order to study an age group in which the bump has been confirmed, and
to compare and contrast the distribution of memories from childhood, the bump and
subsequent adulthood across cohorts of that age group absent of recency effects,
only participants over 55 years of age reporting memories from childhood
68
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(ages 0-17), the bump (ages 18-34) and adulthood (35-51) were included.
Frequency tables and clustered bar charts were used to describe the data.
The following hypotheses were tested using crosstabulation tables and y2 tests to
determine whether the proportion of participants writing from the bump period
(ages 18-34) was statistically independent of the proportion writing about
surrounding life periods (ages 0-17 and 35-51):
Hypothesis 1. Participants 80 years and older generate a bump of memories
corresponding to late adolescence and young adulthood;
Hypothesis 3. Participants writing about individuals who have been important in
their lives generate the largest proportion of memories from the bump years;
Hypothesis 4. Participants writing about events/decisions that have been important
in their lives generate the largest proportion of memories from the bump years;
Hypothesis 6. Men generate the largest proportion of memories from the bump
years.
Hypothesis 7. Men writing about individuals who have been important in their lives
generate the largest proportion of memories from the bump years;
Hypothesis 8. Men writing about events/decisions that have been important in their
lives generate the largest proportion of memories from the bump years;
Hypothesis 9. Women produce similar proportions of memories from childhood, the
bump and adulthood;
Hypothesis 10. Women writing about people who have been important in their lives
produce similar proportions of memories from childhood, the bump and adulthood;
Hypothesis 11. Women writing about events/decisions that have been important in
their lives produce similar proportions o f memories from childhood, the bump and
adulthood.
69
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Similarly, crosstabulation tables and x2 tests were used to test the following
hypotheses, which compared and contrasted subsamples of participants based upon
their age, the topic on which they wrote, and their gender:
Hypothesis 2. Younger and older participants are equally likely to report memories
from the bump years
Hypothesis 5. Participants writing about people and events/decisions are equally
likely to report memories from the bump years;
Hypothesis 12. Men are more likely than women to report memories from the bump
years;
Hypothesis 13. Men writing about individuals who were important in their lives are
more likely to generate memories from the bump years than women writing on that
topic;
Hypothesis 14. Men writing about events/decisions that were important in their
lives are more likely to generate memories from the bump years than women
writing on that topic.
Next, bivariate correlation matrices and multinomial logistic regression were
used to test Hypotheses 1-14 and subsidiary Hypotheses 15-19, which relate to life
cycle factors, personality and memory:
Hypothesis 15. Those who have more education are more likely to produce
memories from the bump.
Hypothesis 16. Those who are more tender-minded are more likely to produce
memories from childhood and adulthood.
Hypothesis 17. Those who are more assertive will be more likely to produce
memories from the bump.
Hypothesis 18. Those who score well on episodic memory measures will be more
likely to report memories from adulthood.
70
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Hypothesis 19. Those who score well on working memory measures will be more
likely to report memories from adulthood.
The multivariate analyses were limited to cases with complete data. Using
SPSS NOMREG, a main effects multinomial logistic regression model was
specified to predict membership in one of the three categories (i.e. memories
reported from childhood, young adulthood or adulthood). A modified sequential
approach was taken to find the most parsimonious model. Predictions were based
on the dichotomous factors topic and sex, and the continuous covariates age,
education, tender-mindedness, assertiveness, episodic memory, and working
memory. Five models were specified. The background measures age and education
were entered first (Model 1), followed by gender (Model 2), topic (Model 3), the
two personality facets (Model 4), and finally the memory measures (Model 5).
A final model was chosen using the log likelihood % 2 test described by
Tabachnick and Fidell (2001). Model 1 is assessed for goodness of fit by comparing
the constant and predictors specified in Model 1 to a constant only model.
Improvements to the fit made in subsequent models are tested by calculating the % 2
differences between them, with reliable improvements indicated at p<.05.
Goodness of fit is also examined using the deviance criterion, which tests for
differences between observed and expected frequencies. Here nonsignificant
differences indicate that the models specified satisfactorily replicate the data.
Nagelkerke’s R2 was used in multinomial logistic regression as an approximate
counterpart of R2 in linear regression. Finally, classification tables were examined.
71
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The percent correctly predicted overall, as w ell as the percent correctly predicted for
each life period were considered.
Once the best fitting model was identified, individual predictors were
evaluated using the likelihood ratio test, which compares model fit with and without
them, and the Wald test, a measure based on the standard error of each coefficient.
A criterion of p<.05 was used for these two tests of the variables.
72
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CHAPTER IV: EXPERIMENT 1
Participants were relatively unconstrained in this free recall experiment.
They were asked to compose an autobiographical paragraph on a lined sheet of
paper included in a mail-in packet of questionnaires. The task did not require the
paragraphs be dated, and it allowed them to choose between two topics. The
instructions were as follows:
We would like to know a little about your life. On the lines below, please
write a paragraph about a decision, event or person that influenced your life.
Choose one of the topics and tell us about the decision, event or person, and
how your life was influenced. You do not need to fill the entire page - a
few sentences will do.
Content analysis procedures described in Chapter III were used to date the
protocols.
IV.A. DESCRIPTIVE STATISTICS
The autobiographical memories of 329 participants who were at least 55
years of age when they joined the Long Beach Longitudinal Study in 1994-1995
were scored. Of these subjects, 251 reported memories from the life periods of
primary concern: 28.3 percent from childhood (ages 0-17), 47 percent from the
bump years (ages 18-34), which correspond to late adolescence and early adulthood,
and 24.7 percent from subsequent adulthood (ages 35-51). Tables IV. 1 through
IV. 5 describe this subsample in terms of the independent variables of interest.
Overall, similar numbers of men and women were represented. More than two-
thirds of the participants were between 65 and 84 years of age at the time of testing.
73
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The largest proportion (40.7 percent) had completed some high school, and of this
number 82.3 percent had earned their high school diplomas. Independent-samples
T tests indicated that participants reporting autobiographical memories from ages
0-51 had a small educational advantage over those reporting from later years
(t=2.368, p=.019), but this was the only mean level difference distinguishing them
(Tables IV. 1 and IV.2).
Table IV. 1 Means and Standard Deviations of Variables (Participants Reporting
From Ages 0-51)_____________________________________________________
Scale Mean SD n
Age In years 73.04 9.02 251
Education In years 13.92 2.96 236
Assertiveness 0-32 15.45 4.8 221
Tender mindedness 0-32 20.4 3.67 221
List Recall Proportion (0-1) 0.5425 0.1845 247
Working Memory Proportion (0-1) 0.6111 0.1588 231
Table IV.2 Means and Standard Deviations of Variables (Participants Reporting
From Ages 52+)_____________________________________________________
Scale Mean SD n
Age In years 72.1 7.68 78
Education In years 13.14 2.33 78
Assertiveness 0-32 14.53 4.33 68
Tender mindedness 0-32 21.06 3.35 68
List Recall Proportion (0-
1)
0.5643 0.1845 77
Working Memory Proportion (0-
1)
0.6111 0.1588 73
74
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table IV. 3 Gender Distribul.ion (n=251)
Number Percent
Men 129 51.4
Women 122 48.6
Table IY.4 Education (n=236)
Number Percent
0-8 years 7 3
9-12 years 96 40.7
13-14 years 49 20.8
15-16 years 41 17.4
17+ years 43 18.2
Table IV.5 Age(n=251)
Number Percent
55-64 44 17.5
65-74 101 40.2
75-84 76 30.3
85+ 30 12
IV.B. AUTOBIOGRAPHICAL MEMORY DISTRIBUTION OF
THE OLDEST OLD
Few studies have had sample sizes allowing for an explicit investigation of
whether the bump exists among the oldest old. However, the Long Beach
Longitudinal Study includes an unusually large number o f participants in their ninth
and tenth decades of life. Autobiographical memory data was available for 73
individuals 80 and older, representing 29.1 of the subsample (Table IV.6). A
75
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
number of mean level age differences distinguished older and younger participants
in terms of the variables of interest (Table IV.7). As might be expected, the older
cohort was more tender-minded (t=-2.649, p=.009), and the younger cohort had
more education (t=2.757, p=.006), was marginally more assertive (t=l .718, p=.087),
and performed better on the episodic (t=6.306, p=.000) and working memory
(t=4.593, p=.000) tasks.
Number Percent
<80 178 70.9
>80 73 29.1
Table IV.7 Means and Standard Deviations of Variables for Younger and Older
Participants
Age <80 Age>79
Mean SD Mean SD
Age 68.42 6 84.3 3.57
Education 14.23 2.82 13.05 3.17
Assertiveness 15.79 4.9 14.56 4.47
Tender-minded 20 3.78 21.44 3.15
List Recall 0.5862 0.1769 0.4342 0.1574
Working Memory 0.6384 0.1527 0.5334 0.1511
The x2 test was used to examine the distribution of autobiographical
memories from childhood preceding the bump, the bump years, and adulthood
following the bump. More specifically, the procedure was used to compare the
76
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
observed and expected frequencies of autobiographical memories reported from
these adjoining life periods. First the childhood period was compared with the
bump period, and then the bump period with the adulthood period. This was done
in order to appraise whether life periods contained the same or different proportions
of values, corresponding to insignificant and significant % 2 values. It was assumed
that significant differences would indicate the existence of the bump. Where
identified, the direction of statistical differences was confirmed by visual inspection
of the data.
Hypothesis 1. Participants 80 years and older generate a bump of memories
corresponding to late adolescence and young adulthood.
Figure IV. 1 displays the distribution of memories for participants over 80
years of age. In this experiment, 32.9 percent, 46.6 percent, and 20.5 percent of the
autobiographical memories reported were derived from the childhood, bump and
adulthood life periods, respectively. While a bump was visually discemable, the % 2
procedure provided only partial evidence of a bump. The goodness of fit test
indicated significant differences in the proportions of autobiographical memories
reported from the young adulthood and adulthood life periods (y2 (l, n=49)=7.367,
p=.007), but not between childhood and young adulthood (% 2( \ , n = 58)= 1.724,
p=.189)
77
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure IV.l Distribution of Autobiographical Memory by Age O oup
0-17 18-34 35-51
Life Period
■ Total (n=251)
■ 804- (n=73)
□ <80 (n=178
Hypothesis 2. Younger (less than 80 years of age) and older (more than 79 years of
age) participants are equally likely to report memories from the bump years.
Figure IV.l also displays the distribution of memories for the 178 younger
participants. Of these, 26.4 percent, 47.2 percent, and 26.4 percent of the memories
reported were derived from the childhood, bump and adulthood life periods,
respectively. Once again, the % 2 procedure was used to compare the observed and
expected frequencies of autobiographical memories reported from adjoining life
periods. A bump of memories was clearly indicated. There was a significant
difference in the proportions of reported autobiographical memories from the
childhood and bump life periods (x2(l, n=131)=10.450, p=.001). The proportions
also differed between the bump and adulthood (x2(l, n=131)=T0.450, p=.001).
78
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
While the bump appears more peaked in the younger cohort, the proportions
of younger and older participants writing from each of the life periods did not differ
statistically. There was no relationship between the proportions of younger and
older individuals writing from childhood (% 2 (1, n=251)=1.069, p=.301), young
adulthood (x2 (l, n=251)=.008, p=.929), or adulthood (x2(l, n=251)=.955, p=.329).
IV.C. AUTOBIOGRAPHICAL MEMORY DISTRIBUTION AND
TOPIC
There is evidence that differences in the types of cues used and differences
in how participants are instructed to complete an autobiographical memory task
affects the distribution of autobiographical memories (Rubin & Schulkind, 1997c;
Fitzgerald & Lawrence, 1984; Howes & Katz, 1992; Rubin & Schulkind, 1997a).
The autobiographical memory task used in Experiment 1, which asked participants
to write about persons, events or decisions important in their lives, provided an
opportunity to extend this line of research.
Less than one-third (29.5 percent) of the participants chose to write about
events or decisions (Table IV. 8). While about one in three of the women (32.8
percent) but only about one in four of the men (26.4 percent) chose to write about a
person, this difference was not statistically significant (x2 (l, n=251)=1.247,
p-264). However, several mean level differences distinguished those writing on
the different topics (Table IV.9). On average, those writing about people were
79
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table IV. 8 Topics of Autobiographical Memory Protocols (n=251)
Number Percent
Person 74 29.5
Event/Decision 177 70.5
Table IV.9 Means and Standard Deviations of Variables by Topic
Person Event/
Decision
Mean SD Mean SD
Age 71.28 9.29 73.77 8.82
Education 14.28 2.83 13.76 3.01
Assertiveness 16.46 4.65 15.03 4.82
T ender-mindedness 20.15 3.3 20.5 3.82
List Recall 0.5799 0.1749 0.5274 0.1867
Working Memory 0.6219 0.1501 0.6065 0.1627
younger (t=-2.007, p=.046), more assertive (t=2.030, p=.044), and scored higher on
tests of episodic memory (t=2.037, p=.043).
Hypothesis 3. Participants writing about individuals who have been important in
their lives generate the largest proportion of memories from the bump years.
The x2 test indicated significant differences in the proportions of reported
autobiographical memories between the childhood and bump life periods (x2 (l,
n=63)=l 1.571, p=.001), but not between the bump and adulthood (x2 (l,
n=29)=1.690, p=.194). The distribution of memories for participants choosing to
write about important people in their lives was clearly skewed towards childhood
(Figure IV.2). The largest proportion of memories (60.8 percent) came from this
80
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
period. Another quarter (24.3 percent) came from the bump, and 14.9 percent were
derived from adulthood.
Figure IV. 2 Distribution of Autobiographical Vfcmory by Iopic
0-17 18-34 35-51
Life Period
■ Total (n=251)
■ People(n=74)
□ Event/Decision
(n=177)
Hypothesis 4. Participants writing about events/decisions that have been important
in their lives generate the largest proportion of memories from the bump years.
Figure IV.2 also displays the distribution of memories for participants
choosing to write about important events or decisions. Of these participants, 14.7
percent, 56.5 percent, and 28.8 percent wrote from childhood, the bump, or
adulthood. The y2 procedure confirmed a bump of memories from late adolescence
and early adulthood, indicating significant differences in the proportions of
reported autobiographical memories between childhood and bump life periods
(% 2 (1, n=126)=43.460, p=.000), and between the bump and adulthood (x2 (l,
n=151)= 15.901, p=.000).
81
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Hypothesis 5. Participants writing about people and events/decisions are equally
likely to report memories from the bump years.
The proportion of memories reported from the three life periods studied
differed depending on whether participants chose to write about persons or
events/decisions. Those writing about people were more likely to report from
childhood (% 2 (1, n=251)=:54.721, p=.000), and less likely to report from young
adulthood (x2 (H n=251)=21.683, p=000) and adulthood (x2 (l, n=251)=5.459,
p=.019) than those writing about events/decisions.
IV.D. AUTOBIOGRAPHICAL MEMORY DISTRIBUTION AND
GENDER
Researchers have identified gender-related differences regarding several
aspects of autobiographical memory (e.g. Davis, 1999; Cowan & Davidson, 1984;
Seidlitz & Diener, 1998; Mackavey, Malley & Steward, 1991; Webster, 1994).
However, there is little evidence of gender-related differences in the distribution of
autobiographical memory (e.g. Rubin, Schulkind & Rahhal, 1999, Rubin, 2000).
The Long Beach Longitudinal Study autobiographical memory experiment was used
to expand on these findings. Approximately equal numbers of men and women were
included in the subsample (Table IV.3). There were several significant mean level
gender differences in terms of the variables of interest (Table IV. 10). First, the
women scored as being more tender-minded (t= -2.886, p=.004). They also
performed better than the men on tests of immediate recall (t= -4.239, p=.000).
82
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table IV.10 Means and Standard Deviations of Variables by Gender
Male Female
Mean SD Mean SD
Age 73.47 8.17 72.59 9.85
Education 14.2 2.92 13.61 2.98
Assertiveness 15.76 4.87 15.12 4.73
Tender
mindedness
19.72 3.79 21.12 3.4
List Recall 0.4961 0.1727 0.5924 0.1845
Working Memory 0.5957 0.1519 0.6269 0.1647
Hypothesis 6. Men generate the largest proportion of memories from the bump
years.
Figure IV. 3 displays the overall distribution of memories for the men
participating in this study. Of the memories they reported, 24 percent, 59.7 percent,
and 16.3 percent came from childhood, the bump and adulthood, respectively. The
X2 procedure indicated significant differences in the proportion of memories
reported between childhood and the bump (x2 (l, n=108)=19.593, p=.000), and
between the bump and adulthood (x2 (l, n=98)=32.0, p=.000).
83
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure IV. 3 Distribution of Autobiographical Memory by Gender
■ Women
Life Period
Hypothesis 7. Men writing about individuals who have been important in their lives
generate the largest proportion of memories from the bump years.
Hypothesis 8. Men writing about events/decisions that have been important in their
lives generate the largest proportion of memories from the bump years.
Thirty-four (26.4 percent) of the 129 men included in the subsample chose
to write about an important person in their life. These participants significantly
differed from men writing about important events/decisions in several ways (Table
IV.l 1). On average, they were younger (t=-2.439, p=.016), had more education
(t=2.742, p= .007), were more assertive (t=2.031, p= .045), and performed better on
the tests of immediate recall (t=2.893, p=.004).
84
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table IV. 11 Means and Standard Deviations by To pic (Men)
People Event/
Decision
Mean SD Mean SD
Age 70.59 7.89 74.49 8.06
Education 15.36 2.89 13.78 2.82
Assertiveness 17.26 4.38 15.2 4.95
Tender-
mindedness
19.13 3.5 19.94 3.9
List Recall 0.5689 0.1741 0.4708 0.1657
Working Memory 0.63 0.1497 0.5828 0.1517
Of the memories men writing about people reported, 52.9 percent, 38.2
percent, and 8.8 percent came from childhood, the bump and adulthood,
respectively (Figure IV.4). The distribution is clearly skewed towards childhood.
The x2 procedure indicated significant differences for the comparison of bump and
adulthood memories (x2 (l, n=16)-6.250, p=.012), but not for childhood and bump
memories (x2 (l, n=31)=.806,p=369).
The more familiar pattern emerged among the men who chose to write about
events/decisions, with 13.7 percent, 67.4 percent, and 18.9 percent writing from
childhood, the bump and adulthood, respectively (Figure IV.4). Here the x2
procedure confirmed the bump, indicating significant differences between
childhood and the bump periods (x2 (l, n=77)=33.779, p=.000), and the bump and
adulthood periods (x2 (l, n=82)=25.805, p=.000).
85
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure IV. 4
Distribution of Autobiographical IVtmory by Gender (IVfele) and Topic
80
70 4
Total (n=129) People (n=34) E/D (n=95)
Life Period
Hypothesis 9. Women produce similar proportions of memories from childhood,
the bump and adulthood.
Figure IV.3 displays the overall distribution of memories for the women
participating in this study. Of the memories they reported, 32.8 percent, 33.6
percent, and 33.6 percent came from childhood , the bump and adulthood,
respectively. The % 2 procedure indicated no significant differences between
childhood and the bump (% 2 (1, n=81)=.012, p=.912) or the bump and adulthood
(X 2 (l,n=82)=.000,p=1.0).
Hypothesis 10. Women writing about people who have been important in their
lives produce similar proportions of memories from childhood, the bump and
adulthood.
Hypothesis 11. Women writing about events/decisions that have been important in
their lives produce similar proportions of memories from childhood, the bump and
adulthood.
86
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Forty (32.8 percent) of the 122 women included in the subsample chose to
write about an important person in their life. These participants did not
significantly differ from women writing about events/decisions (Table IV. 12).
Table IV. 12 Means and Standard Deviations by To die (Women)
People Event/
Decision
Mean SD Mean SD
Age 71.88 10.41 72.94 9.62
Education 13.34 2.44 13.74 3.23
Assertiveness 15.74 4.83 14.84 4.7
Tender
mindedness
21.09 2.85 21.14 3.65
List Recall 0.5895 0.1773 0.5938 0.1889
Working Memory 0.615 0.152 0.6329 0.1714
Women writing about people reported 67.5 percent, 12.5 percent, and 20
percent of their memories from childhood, the bump and adulthood, respectively
(Figure IV.5). The distribution is clearly skewed towards childhood. The % 2
procedure indicated significant differences between childhood and the bump (x2 (l,
n=32)=15.125, p=.000), but not between the bump and adulthood (x2 (l,
n=13)=.692, p=.405).
Visually, a more common pattern emerged for women choosing to write
about event/decisions, with 15.8 percent, 43.9 percent, and 40.2 percent reporting
from childhood, the bump and adulthood, respectively (Figure IV.5). In this case
87
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure IV.5
Distribution of Autobiographical Mfemory by Gender (Female) and Topic
80
70
60
50
40
41
0
i-
0 1
Ph 30 -
20 -
10
0
10-17
118-34
□ 35-51
Total (n=122) People (n=40)
Life Period
E/D (n=82)
the x2 procedure indicated significant differences between childhood and the bump
(X2( l, n=49)=10.796, p=.001), but not between the bump and adulthood (x2 (l,
n=69)=. 130, p=. 718).
Hypothesis 12. Men are more likely than women to report memories from the
bump years.
Overall, the proportion of memories reported from the bump (x2 (l,
n=251)=17.124, p=000) and adulthood (x2 (l, n=251): =10.121, p=001) differed by
gender, with men reporting proportionally more memories from the bump than
women (59.6 percent vs. 33.6 percent) and women reporting proportionally more
memories from adulthood than men (33.6 percent vs. 16.3 percent). Men and
women did not differ in the proportion of memories stemming from childhood
(x2 (l> n=251)=2.370, p=124).
88
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Hypothesis 13. Men writing about individuals who were important in their lives are
more likely to generate memories from the bump years than women writing on that
topic.
Men and women writing about people did not differ in the proportion of
memories stemming from childhood (x2 (l, n=74)=1.635, p=.201) and adulthood
(X 2( l , n=74)=1.814, p=. 178). However, there was a difference in the proportion of
memories from the bump (x2 (l, n=74)=6.613, p=.010), with men reporting more
memories from this period than women (38.2 percent vs. 12.5 percent).
Hypothesis 14. Men writing about events/decisions that were important in their
lives are more likely to report memories from the bump years than women writing
on that topic.
Men and women writing about events/decisions did not differ in the
proportion of memories stemming from childhood (x2 (l, n=T77)=.165, p=.684),
however there was a difference for the bump (x2 (l, n=177)=9.860, p=.002) and
adulthood (x2 ( l , n=177)=9.732, p=.002). Men wrote more memories from the
bump than women (67.4 percent vs. 43.9 percent), and women wrote more
memories from adulthood than men (40.2 percent vs. 18.9 percent).
IV.E. MULTIVARIATE ANALYSES
The descriptive and nonparametric analyses described above provide
evidence that the bump endures into the ninth and tenth decades of life, and that it
varies by gender and the method used. These findings are interesting, contributing
to our understanding of the bump. However, they reveal nothing about additional
individual differences that may help to explain the distribution.
89
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
A bivariate correlation matrix (Table IV. 13) was computed to explore the
viability not only of the primary, but also the subsidiary hypotheses, which relate to
life cycle factors, personality and memory:
Hypothesis 15. Those who have more education are more likely to produce
memories from the bump.
Hypothesis 16. Those who are more tender-minded are more likely to produce
memories from childhood and adulthood.
Hypothesis 17. Those who are more assertive are more likely to produce memories
from the bump.
Hypothesis 18. Those who score well on episodic memory measures are more likely
to report memories from adulthood.
Hypothesis 19. Those who score well on working memory measures are more likely
to report memories from adulthood.
As anticipated, both topic and gender were significant correlates. Those
who chose to write about people were more likely to report autobiographical
memories from childhood (r= -.467, p=.000), and those who chose to write about
events/decisions were more likely to write about the bump period (r=.294, p=.000)
and adulthood (r=.147, p=.019). Significant correlations also indicated that women
were less likely to write from young adulthood (r= -.261, p=.000) and more likely to
write from adulthood (r=.201, p=.001) than men. There were no associations
between age and the period from which participants reported.
The matrix provided evidence for only one of the subsidiary hypotheses. As
expected, greater tender-mindedness, considered a characteristically feminine
measure of the communal orientation (Bakan, 1966; Feingold, 1994), was
90
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
R eproduced with permission of the copyright ow ner. Further reproduction prohibited without p erm ission .
Table IV. 13
Significant Intercorrelations of the Independent and Dependent Variables (n=251)
1 2 3 4 5 6 7 8 9 10 11
1. Topic -
2. Age .126* -
3. Sex --
4. Education -.195** -
5. Assertiveness -136*
. 1 9 9 ** 289* * *
-
6. Tender-mindedness
191**
--
7. List Recall -.129*
.420***
261***
.280*** .232** !
8. W orking M emory
382***
386* * * 332*** 44 6 * * * ; „
9. Childhood
467* * *
--
10. Young
adulthood
2 9 4 ***
.261***
-.145*
592* * *
\
11. Adulthood .147* .2 0 1 **
.360***
5 3 9 *** j
*p<.05, **<.01, ** p<.001
associated with fewer reports from the bump period (r= -.145, p=.031). Greater
tender-mindedness was also positively associated with being female (r=.191,
p=.004). It was marginally correlated with being older (r=.l 17, p=.082) and with
lower levels of education (r= -.121, p=.077).
Autobiographical memories reported by the participants may originate in
childhood, the bump, or adulthood life periods. As this dependent variable is
categorical, multinomial logistic regression was chosen as the most appropriate
multivariate method. Analysis was limited to cases with complete data (n=T99).
Comparisons of the proportions of memories reported from childhood and young
adulthood n=199)=l0.390, p=.001), and from young adulthood and adulthood
(X2( l, n=199)=l 9.042, p=.000) indicated this restriction did not affect the existence
of the bump.
Tables IV. 14 and IV. 15 describe the other variables used in the analysis.
There was a mean level difference in tender-mindedness, with those reporting from
adulthood scoring higher than those reporting from the bump years (t—2.323,
p=.022). In addition, those who wrote from adulthood scored marginally higher on
the episodic memory measure than did those who wrote from childhood (t=-1.916,
p=.058).
92
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table IV. 14 Gender and Topic as a Function of Autobiographical Memory
Distribution (n=199)
Childhood
(0-17)
n (percent)
Bump
(18-35)
n (percent)
Adulthood
(36-51)
n (percent)
Sex
Male 25 (43.9%) 66 (68) 11 (24.4)
Female 32 (56.1) 31 (32) 34 (75.6)
Topic
Person 36 (63.2) 17(17.5) 9(20)
E/D 21 (36.8) 80 (82.5) 36 (80)
Table IV. 15 Means and Standard Deviations of Variables by Life Period
Childhood (0-17)
Mean SD
Bump (18-34)
Mean SD
Adulthood (35-51)
Mean SD
Age 72.81 9.56 72.37 8.18 70.67 7.8
Education 13.88 2.78 14.11 2.91 13.56 2.78
Tender-
mindedness
20.6 3.8 19.68 3.6 21.16 3.33
Assertiveness 16.37 4.42 15.19 5.16 15.29 4.94
Episodic Memory 0.5294 0.1801 0.5562 0.1682 0.6011 0.197
Working
Memory
0.6245 0.1469 0.6164 0.1664 0.6123 0.1609
Using SPSS NOMREG, five main effects models were specified to predict
membership in one of the three categories of the dependent variable (Table IV. 16).
Age and education were entered first (Model 1), followed by gender (Model 2),
topic (Model 3), the two personality facets (Model 4), and finally the memory
measures (Model 5). Model 3, which accounted for about one third of the observed
93
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
variance (Nagelkerke R2 =.324), emerged as the best fitting model using the
deviance criterion and the log likelihood %2 test.
A nonsignificant deviance criterion (p=.783) indicated that the frequencies
predicted by the model adequately replicated the observed frequencies at the various
levels of outcome. Table IV. 16 shows that the background factors age and
education did not reliably improve the fit of Model 1 as compared to a constant only
model (x2 (4, n=199)=3.607, p=.462). Subsequent comparisons of the log-
likelihood ratios for models with and without gender (Model 2) and topic (Model 3)
showed reliable improvements. However, the addition of the personality facets in
Model 4 and the memory measures in Model 5 did not.
Of the 199 memories reported, 28.6 percent came from childhood, 48.7
percent from the bump and 23 percent from adulthood. Model 3 correctly classified
59.6 percent of the childhood memories, 69.1 percent of the bump memories and
42.2 percent of the memories from adulthood. Whereas the addition of gender
noticeably improved the classification of cases into the adulthood life period in
Model 2 (Model 1=0 percent; Model 2= 44.4 percent), the addition of topic clearly
improved classification into the childhood life period in Model 3 (Model 2=28.1
percent; Model 3=59.6 percent).
Comparisons with and without each predictor using the likelihood ratio test
showed that both gender and topic reliably distinguished between outcomes. Table
IV. 17 presents the odds ratios and confidence intervals comparing those writing
94
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Reproduced with permission of the copyright ow ner. Further reproduction prohibited without p erm ission .
Table IV.16 Autobiographical Memory as a Function of Age, Education, Gender, Topic, Personality, and Memory (n=199)
f to Remove
Model 1 Model 2 Model 3 Model 4 Model 5
Background Factors
Age
Education
2.380
1.823
1.377
.718
2.942
1.041
3.052
1.025
1.539
.362
Gender 24.233*** 24.158*** 20.950*** 21.216***
Tonic 38.449*** 36.936*** 36.475***
Personality Facets
Tender-minded
Assertiveness
2.279
1.901
2.299
2.628
Memory Measures
Immediate Recall
Working Memory
4.153
1.727
M odel Statistics
Log-likelihood
chi square test
Deviance criterion
Nagelkerke R2
Classification
-2LL=343.489,
df=4, / = 3.607,
n.s.
p=.115
.021
48.7%
% 2a =24.234,
dfA=2,p<.001
p=.171
.149
53.8%
Z2a=38.499,
d/A=2, p<.001
p=. 783
.324
60.3%
X2 a=4.278,
dfA=4, n.s.
p=.924
.341
60.8%
% 2 a=5.524,
dfA=4, n.s.
p=.933
.363
59.3%
*p<.05; **p<.01; ***p< 001
M 3
Table IV. 17 Multinomial Regression of the Odds of Writing from Three Life
Periods (Model 3)_________________________________________________
Child vs. Bumn
B Odds Ratio
(St. Err.) (95% Cl)
Adult vs. Burnt)
B Odds Ratio
(St. Err) (95% Cl)
Child-vs. Adulthood
B Odds Ratio
(St. Err) (95% Cl)
Topic
-Person
-E/D (Ref)
2.197 9.001***
(.406) (4.1-20.0)
2.022 7.553***
(.480) (2.9-19.3)
Sex
-Male
— Female (Ref)
-1.049 .350**
(.390) (.16-.752)
-1.843 .158***
(.414) (-. 1 -.36)
-
Note: CI=Confidence Interval
*p<.05; **p<.01; ***p<.001
about childhood or adulthood as opposed to the bump, and about childhood as
opposed to adulthood. Controlling for the other variables, those choosing to write
about people were nine times more likely to report from childhood as compared to
the bump (eb =9.001, p<.001), and more than seven times as likely to report from
childhood as compared to adulthood (e b = ::7.553, p<.001). Men were more than two-
thirds less likely to report from childhood than the bump (eb =.350, p<.01), and they
were even less likely to report on adulthood (eb = .158, p<.001).
IV. F. SUMMARY
A bump of memories stemming from late adolescence and early adulthood
was identified in Experiment 1. It was more peaked for participants 55 to 79 years
of age than for those 80 and older. As hypothesized, however, there was no
96
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
statistical difference in the proportion of memories reported from childhood, the
bump or adulthood between the younger and older groups.
Contrary to the predictions, the distribution of memories differed depending
upon whether participants chose to write about people or events/decisions. That is,
those choosing to write about people were more likely to report from childhood than
participants writing about events or decisions, who tended to focus on memories
from the bump years.
The patterns were different for men and women. As predicted, the largest
proportion of memories reported by men overall came from the bump period,
whereas for women the distribution of memories was flatter across the life periods
studied. Both men and women who chose to write about people produced the
largest proportion of memories from childhood. However, whereas 38.2 percent of
the men writing on this topic reported memories from the bump years, only 12.5
percent of the women wrote from this period. Furthermore, whereas men writing
about events or decisions generated a clear bump of memories, women writing on
this topic produced similar proportions of memories from the bump and adulthood
periods.
Multivariate models confirmed these findings. Those writing about people
were much more likely to generate memories about childhood (as compared to the
bump or adulthood) than those writing about events/decisions, even with gender,
age and education held constant. And men were less likely to report from childhood
97
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
or adulthood (as compared to the bump) than women, with topic, age and education
held constant. Contrary to the hypotheses, neither the personality facets nor the
memory measures emerged as predictors of autobiographical memory.
98
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CHAPTER V: EXPERIMENT 2
Participants in Experiment 1 were allowed to choose to write about a person,
event or decision important in their lives. One of the goals of Chapter 5 was to
explore whether the distribution of autobiographical memory changes when that
choice is constrained. Participants in Experiment 2 were asked to date the memories
they described, and were assigned either to Version A or B. The instructions were
as follows:
Version A: We would like to know a little about your life. On the lines
below, please write a paragraph about either 1) a decision you made that
had a major influence, or 2) an event that influenced your life. Choose
one of the topics and tell us about the decision or event, and how your
life was influenced. You do not need to fill the entire page - a few
sentences will do.
Version B: We would like to know a little about your life. On the lines
below, please write a paragraph about a person who had a major influence
on your life. Tell us about the person, and how they influenced your life.
You do not need to fill the entire page - a few sentences will do.
As in Experiment 1, the content analysis procedures described in Chapter III were
used to date the protocols.
V.A. DESCRIPTIVE STATISTICS
The autobiographical memories reported by 327 participants aged 55 and
older in 2000-2002 were precisely dated using content analysis. Of these, 268
reported memories from the life periods of interest: 31.7 percent from childhood
(0-17), 52.2 percent from the bump years (18-34), which correspond to late
adolescence and young adulthood, and 16 percent from adulthood (ages 35-51).
99
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Tables V.l-5 describe this subsample in terms of the independent variables of
interest. Slightly more than half (54.9 percent) of the sample was female, and 157
(58.6 percent) were between 75 and 84 years of age at the time of testing. The
largest proportion (33.6 percent) of participants had completed some high school,
and of these 86 percent had earned their high school diplomas. Independent-
samples T tests revealed no significant mean level differences between those
reporting autobiographical memories from between ages 0-51 (Table V.l) and those
reporting autobiographical memories from 52+ (Table V.2).
Table V.l Means and Standard Deviations of Variables (Participants Reporting
From Ages 0-51)
Scale Mean SD n
Age In years 76.62 7.65 268
Education In years 14.35 2.92 256
Assertiveness 0-32 15.66 5.13 252
T ender-mindedness 0-32 21.18 3.49 252
List Recall Proportion (0-1) .5122 .1885 268
Working Memory Proportion (0-1) .5110 .1787 255
100
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table V.2 Means and Standard Deviations of Variables (Participants Reporting
From Ages 52+)____________________________________________________
Scale Mean SD n
Age In years 76.39 7.51 59
Education In years 14.34 2.34 56
Assertiveness 0-32 15.62 5.28 53
Tender-mindedness 0-32 21.55 3.35 53
List Recall Proportion (0-1) .5178 .1886 59
Working Memory Proportion (0-1) .4794 .1999 58
Table V.3 Gender Distribution (n-268)
Number Percent
Men 121 45.1
Women 147 54.9
Table V.4 Education (n=256 )
Number Percent
0-8 years 3 1.2
9-12 years 86 33.6
13-14 years 58 22.7
15-16 years 52 20.3
17+ years 67 22.3
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table V.5 Age (n=268)
Number Percent
55-64 25 9.3
65-74 52 19.4
75-84 157 58.6
85+ 34 12.7
V.B. AUTOBIOGRAPHICAL MEMORY DISTRIBUTION OF
THE OLDEST OLD
Researchers have shown that, given the right conditions, individuals in their
third and fourth decade exhibit a bump of memories (Jansari & Parkins, 1996).
However, the existence of a bump among those in their ninth and tenth decades has
not been confirmed. Autobiographical memory data was available for 99
individuals 80 and older, representing 36.9 percent of the Long Beach Longitudinal
Study Panel 3 subsample (Table V.6). As in Chapter IV, there were several
significant mean level age differences distinguishing older and younger participants
(Table V.7). Whereas the older cohort was more tender-minded (t=-2.454,
p =.015), the younger cohort performed better on both the episodic (t=4.052,
p=.000) and working memory (t=3.756, p=.000) tasks.
Number Percent
<80 169 63.1
>80 99 36.9
102
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table V.7 Means and Standard Deviations of Variables for Younger and Older
Participants________________________________________________________
Age <80 Age>79
Mean SD Mean SD
Age 72.51 6.43 83.65 3.08
Education 14.44 2.91 14.18 2.94
Assertiveness 16.04 5.15 15 5.05
Tender-
mindedness
20.77 3.44 21.88 3.49
List Recall .5469 .1848 .4529 .1808
Working Memory .5420 .1750 .4569 .1729
Autobiographical memories from childhood, the bump and adulthood were
examined to determine whether there is evidence of the reminiscence effect. As in
Chapter IV, the % 2 procedure was used to compare the observed and expected
frequencies of autobiographical memories reported from adjoining life periods.
Hypothesis 1. Participants 80 years and older generate a bump of memories
corresponding to late adolescence and young adulthood.
Figure V.l displays the distribution of autobiographical memories for
participants over 80 years of age. In this experiment, 31.1 percent, 55.6 percent,
and 13.1 percent of the memories reported are derived from childhood, the bump,
and adulthood, respectively. Confirming the existence of the bump, the goodness of
fit test indicated significant differences in the proportions of autobiographical
memories reported between childhood and the bump periods (x2(l, n=86)=6.698,
103
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure V.l Distribution of Autobiographical Memory by Age Group
60
■ Total (n=268)
■ 80+ (n=99)
□ <80 (n=l 69)
0-17 18-34 35-51
Life Period
p=.01), as well as between the bump and adulthood periods (x2(l, n=68)=25.941,
p=.000).
Hypothesis 2. Younger (less than 80 years of age) and older (more than 79 years of
age) participants are equally likely to report memories from the bump years.
Figure V.l also illustrates the overall distribution of memories for
participants less than 80 years of age. In a sample of 169, 32 percent, 50.3 percent,
and 17.8 percent of the memories reported were derived from the childhood, young
adulthood and adulthood life periods, respectively. Once again, the % 2 procedure
was used to compare the observed and expected frequencies of autobiographical
memories reported from adjoining life periods. The goodness of fit test clearly
indicated the existence of a bump. There was a significant difference in the
proportions of reported autobiographical memories for the comparison of
104
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
frequencies from the young adulthood and adulthood life periods (% 2(1,
n=139)=6.914, p=.009). In addition, there was a difference between the young
adulthood and adulthood life periods (% 2(1, n=l 15)=26.304, p=.000).
The proportion of younger and older participants writing from each of the
life periods did not differ. Statistically there was no relationship between the
proportions of younger and older individuals writing from childhood (x2 (1,
n=268)=012, p=.914), young adulthood (x2(l, n=268): : : : : .692, p=.405), or adulthood
(X2(l> n=268)=.989, p=320).
V.C. AUTOBIOGRAPHICAL MEMORY DISTRIBUTION AND
TOPIC
The autobiographical memory task used in Experiment 2 instructed
participants to write about persons or events/decisions that were important in their
lives. More than one third of the participants (Table V.8) were assigned to write
about people who had been important in their lives. Overall, 41.3 percent of the
men and 37.4 percent of the women were assigned to this topic. No mean level
differences distinguished those writing about people and events/decisions (Table
V.9).
Table V.8 Topics of Autobiographical Memory Protocols (n=268)______________
Number Percent
Person 105 39.2
Event/Decision 163 60.8
105
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table V.9 Means and Standard Deviations of Variables by Topic
Person Event/
Decision
Mean SD Mean SD
Age 77.08 7.35 76.33 7.84
Education 14.48 2.90 14.26 2.94
Assertiveness 15.84 4.95 15.54 5.26
Tender-mindedness 21.30 3.39 21.11 3.57
List Recall .5057 .1833 .5163 .1923
Working Memory .5351 .1789 .4954 .1774
Hypothesis 3. Participants writing about individuals who have been important in
their lives generate the largest proportion of memories from the bump years.
Autobiographical memories from childhood, the bump years and adulthood
were examined to determine whether there is evidence that the reminiscence effect
is influenced by topic. The distribution of memories for participants assigned to
write about important people was clearly skewed towards childhood (Figure V.2).
The largest proportion of memories (59 percent) came from this period. Almost a
third (31.4 percent) came from the bump, and the final 9.5 percent from adulthood.
The goodness of fit tests used indicated significant differences in the proportions of
reported autobiographical memories between the childhood and bump periods
(X 2(l, n=95)=8.853, p=.003), and between the bump and adulthood
(X 2(l, n=43)=l2.302, p= 000).
106
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure V. 2 D istribution of Autobiographical Memory by Topic
■ Total (n=268)
■ People(n=105)
□ Event/Decision(n=l 63)
0 1 7 18-34 35-51
Life Period
Hypothesis 4. Participants writing about events/decisions that have been important
in their lives generate the largest proportion of memories from the bump years
Figure V.2 also displays the distribution of memories for participants
assigned to write about important events or decisions. Of these participants, 14.1
percent, 65.6 percent and 20.2 percent wrote from childhood, the bump or
adulthood. The goodness of fit tests confirmed significant differences in the
proportions of reported autobiographical memories between the childhood and
bump life periods (% 2 (1, n=130)=54.277, p=.000), and between the bump and
adulthood (x2 (1, n=140)=39.114, p=.000).
Hypothesis 5. Participants writing about people and events/decisions are equally
likely to report memories from the bump years.
Assignment to a topic affected the proportion of memories reported from the
three life periods. Those writing about people were more likely to write from
107
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
childhood (x2( 1, n=268)=59.546, p=.000), and less likely to report from young
adulthood (y2 (1, n=268)=29.966, p=.000), or adulthood (x2 (1, n=268)=5.450,
p=.020) than those writing about events/decisions.
V.D. AUTOBIOGRAPHICAL MEMORY DISTRIBUTION AND
GENDER
Research exploring gender as it relates to the distribution of
autobiographical memory is limited. The Long Beach Longitudinal Study study
provides and opportunity to expand this line of inquiry. Slightly more than half of
those tested in 2000-2002 were female (Table V.3). At the mean level (Table
V.10), these women had less education (t=2.658, p=.008) than men, but performed
better on the measures of episodic memory (t=-4.956, p=.000). They were also less
assertive (t=1.996, p=.047) and more tender-minded (t=-3.366, p=.001).
Table V.10 Means and Standard Deviations of Variables by Gender____________
Male Female
Mean SD Mean SD
Age 77.26 6.92 76.1 8.19
Education 14.88 2.94 13.91 2.83
Assertiveness 16.35 5 15.07 5.18
Tender-mindedness 20.40 3.67 21.85 3.20
List Recall .4519 .1741 .5618 .1860
Working Memory .4929 .1805 .5258 .1765
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Hypothesis 6. Men generate the largest proportion of memories from the bump
years.
Figure V. 3 Distribution of Autobiographical Memory by Gender
60 -
0
a
■ Men
■ Women
0-17 18-34 35-51
Life Period
Figure V.3 shows the overall distribution of autobiographical memories for
the men tested. Of the 121 memories reported, 28.1 percent, 57.9 percent and 14
percent came from childhood, the bump and adulthood, respectively. The % 2
procedure indicated significant differences in the proportions of memories reported
between childhood and the bump (y2 (1 ,n=104)=l 2.462, p=.000), as well as
between the bump and adulthood (x2(l, n=87)=32.287, p=.000).
Hypothesis 7. Men writing about individuals who have been important in their lives
generate the largest proportion of memories from the bump years.
Hypothesis 8. Men writing about events/decisions that have been important in their
lives generate the largest proportion of memories from the bump years.
109
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Of the 121 men participating, 50 (41.3 percent) were assigned to write about
an important person. They performed significantly better than men writing about
important events/decisions on tests of working memory (t=2.132, p=.035), the only
mean level difference between the two groups (Table V.l 1).
Men writing about people reported 52 percent, 36 percent, and 12 percent
from childhood, the bump and adulthood, respectively (Figure V.4). The
distribution appears to be skewed towards childhood. However, while the % 2
procedure indicated a significant difference for the comparison of the bump and
adulthood periods (% 2 (l,n=24)=6.0, p=. 014), the comparison of the childhood and
bump periods was not statistically significant (y2 (1, n=44)=1.455, p=.228).
Table V. 11 Means and Standard Deviations by Topic (Men)___________________
People Event/
Decision
Mean SD Mean SD
Age 78.18 5.71 76.61 7.62
Education 14.69 3.21 15.02 2.75
Assertiveness 16.42 4.09 16.31 5.59
Tender-
mindedness
20.38 3.71 20.41 3.66
List Recall .4250 .1623 .4708 .1806
Working Memory .5354 .1605 .4635 .1887
Among those men assigned to the topic of events/decisions, 11.3 percent,
73.2 percent, and 15.5 percent wrote on childhood, the bump and adulthood,
respectively. The % 2 procedure confirmed the bump, showing significant
110
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
figure V. 4
D istribution o f A utobiographical IVfemory by G ender (TVfale) and Topic
80 t
70 \
BO-17
■ 18-34
□ 35-51
Total (n=121) People (n=50)
life Period
E D (n=71)
differences between childhood and the bump periods (x2 (l,n=60)=32.267, p=000),
and the bump and adulthood (x2 (l,n=63)=26.683, p=.000).
Hypothesis 9. Women produce similar proportions of memories from childhood,
the bump and adulthood.
The overall distribution of memories for women participating in Panel 3 is
displayed in Figure V.3. Of the 147 memories reported, 34.7 percent, 47.6 percent,
and 17.7 percent came from childhood, young adulthood, and adulthood,
respectively. The x2 procedure indicated significant differences between the bump
and adulthood (x2 (l ,n=96)=20.167, p=.000), however the comparison of the
childhood and bump periods was only marginal (x2 (l,n=121)=2.983, p=.084).
Hypothesis 10. Women writing about people who have been important in their lives
produce similar proportions of memories from childhood, the bump and adulthood.
Ill
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Hypothesis 11. Women writing about events/decisions that have been important in
their lives produce similar proportions of memories from childhood, the bump and
adulthood.
Of the women participating in this experiment, 55 (37.4 percent) were
assigned to write about an important person from their lives. They did not
significantly differ from women writing about events/decisions in terms of the
variables of interest (Table V.12).
Table V.12 Means and Standard Deviations by Topic (Women)________________
People Event/
Decision
Mean SD Mean SD
Age 76.07 8.50 76.12 8.04
Education 14.27 2.59 13.71 2.96
Assertiveness 15.31 5.62 14.92 4.92
Tender-
mindedness
22.15 2.85 21.67 3.41
List Recall .5791 .1710 .5515 .1946
Working Memory .5349 .1953 .5203 .1649
In this experiment, 65.5 percent, 27.3 percent and 7.3 percent of the women
assigned to write about important people reported from childhood, the bump, and
adulthood, respectively. The skew of the distribution towards childhood was
unmistakable. Both the x2 comparison of childhood with the bump life period
(X2 (l,n=51)=8.647, p=.003), and the bump with adulthood life period
(X2 (l,n=T9)=6.368, p=.012) were statistically significant.
112
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Women assigned to the topic of events/decisions produced a bump of
memories, with 16.3 percent writing from childhood, 59.8 percent from the bump,
and 23.9 percent from adulthood (Figure V.5). The % 2procedure confirmed the
Figure V. 5
Distribution of Autobiographical Memory by Gender (Female) and Topic
70
Total (n=l 47) People (n=55) E©(n=92)
Life Period
bump, both in comparisons of childhood with bump memories (x2 (l ,n=70)=22.857,
p=.000), and bump with adulthood memories (x2 (l,n=77)=14.143, p=.000).
Hypothesis 12. Men are more likely than women to report memories from the
bump years.
Men and women did not differ in the proportion of memories stemming
from the childhood (% 2 (1, n=268)=1.333, p=.248) or adulthood (x2 (l, n=268)=.652,
p=.419) life periods. However, there was a marginal difference in the bump years
113
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(X2 (l, n=268)=2.785, p=.095), with men reporting proportionally more memories
from the bump than women (57.9 percent vs. 47.6 percent).
Hypothesis 13. Men writing about individuals who were important in their lives are
more likely to generate memories from the bump years than women writing on that
topic.
Men and women did not differ in the proportion of memories stemming
from childhood (x2 (l, n=105)=1.961, p~ 161), the bump (x2 (l, n=105)=.926,
p=.336), or adulthood (x2 (l, n=105)=.679, p=.410).
Hypothesis 14. Men writing about events/decisions that were important in their
lives are more likely to report memories from the bump years than women writing
on that topic.
Men and women did not differ in the proportion of memories stemming
from the childhood (x2 (L n=163): = :.839, p=.360) or adulthood (x2 (l, n=163)=1.760,
p=.185) life periods. However, once again there was a marginal difference in the
bump years (x2 (l, n=163)=3.218, p=.073), with men reporting proportionally more
memories from the bump than women (73.2 percent vs. 59.8 percent).
V. E. MULTIVARIATE ANALYSES
As in Chapter IV, the principal and subsidiary hypotheses were explored by
computing a bivariate correlation matrix (Table V.13). The principal hypotheses
predicted that the distribution of autobiographical memories would be consistent for
the oldest old and across topics, but that it would be influenced by gender.
Subsidiary hypotheses predicted associations between the distribution of
autobiographical memory and educational attainment, personality characteristics
related to gender, and memory performance:
114
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Hypothesis 15. Those who have more education are more likely to produce
memories from the bump.
Hypothesis 16. Those who are more tender-minded are more likely to produce
memories from childhood and adulthood.
Hypothesis 17. Those who are more assertive are more likely to produce memories
from the bump.
Hypothesis 18. Those who score well on episodic memory measures are more likely
to report memories from adulthood.
Hypothesis 19. Those who score well on working memory measures are more likely
to report memories from adulthood.
Both topic and gender correlated with the autobiographical memory
variables. More specifically, there was a strong association between writing about
people and reporting autobiographical memories from childhood (r= -.471, p=.000).
Those who were assigned to write about events/decisions were more likely to write
about the bump period (r=.334, p=.000) and adulthood (r=. 143, p=.020). In regard
to gender, women were marginally less likely to write from young adulthood
(r= -.102, p=.096) than men. The data from Experiment 2 also indicated an
association between age and the period from which participants reported. Increased
age was marginally correlated with writing from the bump period (r=.108, p=.076)
and inversely associated with writing from adulthood (r=-.125, p=.041).
The matrix provided marginal evidence for only two of the subsidiary
hypotheses. As in Experiment 1, higher scores for tender-mindedness were
associated with fewer reports from the bump period (r= -.107, p=.089), and also
with more reports from adulthood (r=. 106, p=.092). A higher level of tender-
mindedness was significantly correlated with increased age (r= 177, p=.005) and
115
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
R eproduced with permission of the copyright ow ner. Further reproduction prohibited without p erm ission .
Table V.13 Significant Intercorrelations of the Independent and Dependent Variables (n=268)
1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11.
T D
=. 1. Topic
C Q
31 ...............
O 2. Age --
p 3. Sex -
m ...............
§. 4. Education -.129* -.165** -
C D ...............
j g 5. Assertiveness -.125* .153* -
- 5
B . 6. Tender-minded
C
177**
.208** -
& ...............
o 7. List Recall
■ o
“ 5 ...............
373***
.201** --
o
5; 8. Working Memory
C D
Q . ............... .
349***
249*** 404***
-
1 9. Childhood
o 471***
c
--
o > 10. Young adulthood .334***
|
723***
--
§ 11. Adulthood .143*
i
N J
L / i
*
.132*
.298***
457***
-
*p<.05,** p<.01, *** p<.001
i
Os
being female (r=.208, p=.001), and it was marginally inversely associated with
assertiveness (r=-.107, p =.092). There was also a correlation between higher
working memory performance and reports from adulthood (r=.132, p =.035).
The dependent variable used in the following analyses classifies
autobiographical memories as stemming from one of three life periods. Therefore,
multinomial logistic regression, which can be used to model three or more
categories of nominal outcome, was chosen as an appropriate multivariate test of the
primary and subsidiary hypotheses. Analysis was limited to cases with complete
data (n=230). Comparisons of the frequencies of memories reported from childhood
or young adulthood (% \l, n=193)=16.834, p=.000), and from young adulthood or
adulthood (% 2(1, n=162)=47.802, p=.000) indicated this restriction did not affect the
existence of the bump.
Tables V.14 and V.15 describe the variables used in the analyses in terms of
these categories. In both tables, the first column refers to those reporting from
childhood, the second those reporting from the bump, and the third those reporting
from adulthood. There was a mean level difference in age, with those reporting
from the bump significantly older than those reporting from adulthood (t=2.051,
p=.042). In addition, there was a mean level difference in working memory
performance favoring those reporting from adulthood as compared to the bump
(t=-2.094, p = .038).
117
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
V.14 Gender and Topic as a Function of Autobiographical Memory Distribution
(n=230)_______________ _____________________________________________
Childhood (0-
17)
n (percent)
Bump (18-35)
n (percent)
Adulthood(36-
51)
n (percent)
Sex
Male 29 (42.6%) 63 (50.4) 15 (40.5)
Female 39 (57.4) 62 (49.6) 22 (59.5)
Topic
Person 51(75) 31 (24.8) 8(21.6)
E/D 17 (25) 94 (75.2) 29 (78.4)
V.15 Means and Standard Deviations of Variables by Life Period
Childhood (0-17)
Mean SD
Bump (18-34)
Mean SD
Adulthood (35-51)
Mean SD
Age 76.22 7.68 77.10 7.82 74.11 7.74
Education 14.46 2.81 14.46 2.90 14.27 2.31
Tender-
mindedness
21.18 3.45 20.83 3.65 21.81 3.02
Assertiveness 16.04 5.35 15.44 4.54 15.95 6.66
Episodic Memory .5375 .1953 .5035 .1870 .5601 .1641
Working
Memory
.5357 .1852 .4970 .1653 .5620 .1681
As in Experiment 1, five models were specified to predict membership in
one of the three categories of the dependent variable (Table V.16). Age and
education were entered first (Model 1), followed by gender (Model 2), topic (Model
3), the two personality facets (Model 4), and finally the memory measures (Model
5). The log likelihood % 2 test and the deviance criterion indicated that Model 3,
118
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
which accounted about a quarter of the observed variance (Nagelkerke R2 =.265),
was the strongest model.
Neither age nor education reliably improved the fit of Model 1 as compared
to a constant only model (x2 (4, n=230)=4.493, p=.343), and adding gender did not
significantly contribute to the fit of Model 2 (x2(2, n=230)=1.296, p>.05). The
only significant refinement resulted from the inclusion of topic in Model 3 (x2 (2,
n=230)=53.668, p<001). Adding this variable also resulted in the single largest
increase to the Nagelkerke R2 statistic (Model 2 =.029, Model 3 =.265), and there
was a good fit using the deviance criterion (p=.854).
More than half of the 230 memories reported were generated from the bump
period, 29 percent from childhood and 16 percent from adulthood. Model 3
successfully predicted 76 percent of the bump memories, 73.5 percent of those from
childhood, and none of the memories from adulthood. The inclusion of topic
clearly accounted for the increased ability to predict childhood memories, however
overall there was only an 8.7 percent improvement over blind estimation into the
most frequently reported life period, the bump.
Table V.17 presents the odds ratios and confidence intervals comparing
those writing about childhood or adulthood as opposed to the bump. Topic
predicted life period in two of the comparisons. That is, as compared to those
writing about events/decisions, those assigned to write about a person were 9.7
times more likely to write from childhood than the bump, and 10.7 times more
119
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
R eproduced with permission of the copyright ow ner. Further reproduction prohibited without p erm ission .
Table V.16 Autobiographical Memory as a Function of Age, Education, Gender, Topic, Personality, and Memory (n=230)
y 2 to Remove
Model 1 Model 2 Model 3 Model 4 Model 5
Background Factors
Age
Education
4.345
.369
3.997
.230
4.181
.180
5.897
.288
2.533
.866
Gender 1.296 1.889 1.575 .769
Tonic 53.668*** 53.877*** 53.247***
Personality Facets
T ender-mindedness
Assertiveness
3.841
.929
4.215
.891
Memory Measures
Immediate Recall
Working Memory
.615
1.841
Model Statistics
Log-likelihood
Z2 test
Deviance criterion
Nagelkerke R2
Classification
-2LL=355.136,
df=4,%2 =4.493,
n.s.
p = 1 9 0
.022
54.3%
X1 * = 1 .2 9 6 ,
dfA=2, n.s.
p=.354
.029
54.3%
Z 2 a =53.668,
dfA=2, p<.001
p=.854
.265
63%
Z 2 a = 4 -6 3 8 ,
dfA=4, n.s.
p = 9 7 1
.283
62.6%
Z 2a = 2 .9 6 7 ,
dfA=4, n.s.
p=.974
.294
63.9%
*p<.05; **p<.01; ***p<.001
N )
O
V.17 Multinomial Regression of the Odds of Writing from Three Life
Periods(Model 3)___________________________________________
Childhood vs. Bump
B Odds Ratio
(St. Err) (95% Cl)
Adulthood vs. Bump
B Odds Ratio
(St. Err) (95% Cl)
Child-vs.Adult-hood
B Odds Ratio
(St. Err) (95% Cl)
Topic
-Person
-E/D
(Ref)
2.274 9.718***
(.356) (4.8-19.5)
2.367
(.492)
10.66***
(4.1-28)
Age -.05
(.023)
.955*
(.912-
.999)
Note: CI=Confidence Interval
*p<.05, **p<.01; ***p<.001
likely to write from childhood than adulthood. In addition, each additional year of
life slightly reduced the likelihood of writing from adulthood as compared to the
bump.
V.F. SUMMARY
One of the goals of this chapter was to test whether the distribution of
autobiographical memories would be affected if participants were constrained
regarding the topic of their paragraph. Thus while participants in Experiment 1
were allowed to choose to write about a person or event/decision, participants in
Experiment 2 were assigned to one or the other topic.
As anticipated, the existence of a bump of memories stemming from late
adolescence and early adulthood was identified in Experiment 2. However, whereas
the bump of memories was less pronounced for those 80 and older than for those
121
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
55-79 years of age in Experiment 1, it was practically identical for the two age
groups in Experiment 2.
As was the case in Experiment 1, the topic unexpectedly affected the
generation of memories. Contrary to the predictions, participants writing about
people were more likely to report from childhood than those writing about events or
decisions, who tended to generate memories from the bump years. Still, almost a
third of those writing about people produced memories from the bump years.
Also as anticipated, there was evidence of a bump of memories among the
male participants overall. Those writing about people, while showing a clear skew
towards childhood, produced more than a third of their memories from the bump
period (36 percent). Contrary to the hypotheses and the findings in Experiment 1,
there was considerable evidence of a bump of memories among women as well,
especially among those writing about events or decisions. However, there was a
marginal difference between the proportion of bump memories generated by men
and women writing on this topic (p=.073). That is, a slightly higher proportion of
men (73.2 percent) than women (59.8 percent) produced memories from this period.
These findings were reflected in the multivariate analyses. In the model
chosen, only topic reliably improved the fit. Those writing about people (as
compared to those writing about events/decisions) were much more likely to write
about childhood than the bump or adulthood, with gender, age and education held
122
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
constant. Age also emerged as an individual predictor, with older individuals being
slightly less likely to report from adulthood as compared to the bump.
123
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CHAPTER VI: SUMMARY AND DISCUSSION
The study of the bump, which has identified and described an interesting
anomaly, is moving in a theoretical direction emphasizing explanation. The
objectives of this dissertation - to determine whether a bump of memories emerged
for participants 80 and older, whether the anomaly was affected by instructional
modifications, and whether gender impacted its dimensions - reflect that trend.
Two experiments were conducted, allowing participants differing amounts of
freedom in choosing a topic from which to generate memories, and the results were
compared to assess possible selection effects. Subsidiary hypotheses tested the role
of life cycle factors, personality and memory in the distribution of autobiographical
memory. The existence of such individual differences would have implications for
theoretical discussions of the bump, and multivariate models were developed for
this purpose.
VI.A. REVIEW OF THE FINDINGS
While scholars have suggested that designs prompting multiple memories
may have a cloning effect, where memories from one period cue other memories
from that period, there was clear evidence of a bump in the two samples studied,
even though participants were prompted to free recall just a single memory. The
results were substantially consistent for two of the three primary research questions
across both experiments (see also Tables VI. 1 and VI.2).
124
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
A bump of memories from late adolescence and young adulthood emerged
for both age groups in Experiments 1 and 2, suggesting the distribution of
autobiographical memory may be stable into the ninth and tenth decades of life
among healthy individuals.
Table VI. 1 Summary of Descriptive Analyses
Hypotheses Experiment 1 Experiment 2
1. Participants 80 years and older generate
a bump of memories corresponding to late
adolescence and young adulthood.
X |X Partially
accepted
Provides some
support for
bump in oldest
old.
✓ Accepted
Provides
support for
bump in oldest
old.
2. Younger and older participants are
equally likely to report memories from the
bump years.
i / Accepted
Provides support
for similarity of
distribution
across the two
age groups.
/ Accepted
Provides
support for
similarity of
distribution
across the two
age groups.
3. Participants writing about individuals
who have been important in their lives
generate the largest proportion of memories
from the bump years.
X Rejected
Indicates a surge
of memories
from childhood.
X Rejected
Indicates surge
of memory
from
childhood.
4. Participants writing about
events/decisions that have been important
in their lives generate the largest proportion
of memories from the bump years.
✓ Accepted
Provides support
for bump.
✓ Accepted
Provides
support for
bump.
125
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table VI. 1 (Cont.) Hypotheses Experiment 1 Experiment 2
5. Participants writing about people and
events/decisions are equally likely to report
memories from the bump years.
X Rejected
Provides
evidence that
distribution is
affected by topic.
X Rejected
Provides
evidence that
distribution is
affected by
topic.
6. Men generate the largest proportion of ✓ Accepted ✓ Accepted
memories from the bump years. Provides support Provides
for bump. support for
bump.
7. Men writing about individuals who have X Rejected X Rejected
been important in their lives generate the Indicates a surge Indicates a
largest proportion of memories from the of memories surge of
bump years. from childhood. memories from
childhood.
8. Men writing about events/decisions that ✓Accepted ✓Accepted
have been important in their lives generate Provides support Provides
the largest proportion of memories from the for bump. support for
bump years. bump.
9. Women produce similar proportions of ✓ Accepted X ✓ Partially
memories from childhood, the bump and Provides no rejected
adulthood. support for Provides some
bump. support for
bump.
10. Women writing about people who have X ✓ Partially X Rejected
important in their lives produce similar rejected Provides
proportions of memories from childhood, Indicates a surge evidence of
the bump and adulthood. of memories surge of
from childhood. memory from
childhood.
126
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table VI. 1 (Cont.) Hypotheses Experiment 1 Experiment 2
11. Women writing about events/decisions
that have been important in their lives
produce similar proportions of memories
from childhood, the bump and adulthood.
X » X Partially
rejected
Provides some
support for
bump.
X Rejected
Provides
support for
bump.
12. Men are more likely than women to » X Accepted X Rejected
report memories from the bump years. Support for Only marginal
gender support for
difference. gender
difference.
13. Men writing about individuals who ✓ Accepted X Rejected
were important in their lives are more likely Support for No support for
to generate memories from the bump years gender gender
than women writing on that topic. difference. difference.
14. Men writing about events/decisions •X Accepted X Rejected
that were important in their lives are more Support for Only marginal
likely to report memories from the bump gender support for
years than women writing on that topic. difference. gender
difference.
Approximately half of the participants 80 and older generated memories from
young adulthood (46.6 percent and 55.6 percent, respectively). Similar proportions
(47.2 percent and 50.3 percent) of the younger participants reported from this
period. Although the bump was more pronounced in Experiment 2, especially
among the oldest old, the differences between experiments were not statistically
significant.
127
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table VI.2 Summary o f Multivariate Analyses
Hypothesis/
V ariable
Memory Ability Theory Self Narrative/Rapid
Change Theories
1-3 Age ✓Younger participants
exhibit more recent
memories (Experiment 2
and combined data sets
only)
4-14 Gender ✓Women exhibit more
recent memories
(Experiment 1 and
combined data sets)
✓ Women’s memories
more even across lifespan
(Experiment 1 and
combined data sets)
15 Education X More education, more
recent memories
X More education, more
bump memories
16 Commune
✓ More communal, fewer
bump memories
(Combined data sets only)
17 Agency
X More agentic, more
bump memories
18 Episodic
Memory
(EM)
X Better EM, more recent
memories
19 Working
Memory
(WM)
X Better WM, more recent
memories
The findings were also substantially consistent across panels in regard to the
effect of topic. Whereas participants clearly generated a bump when they wrote
about events or decisions, they did not when they wrote about people (see Figures
IV.2 and V.2). Approximately 60 percent of the memories relating to people were
generated from childhood (Experiment 1 = 60.8 percent; Experiment 2 = 59
128
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
percent). The skew was apparent in both experiments, and it was confirmed in the
multivariate models. That is, while holding other variables constant, individuals
writing about people were much more likely to produce memories from childhood
than from the bump or subsequent adulthood (Tables IV.17 and V.17). While these
results were not anticipated, they follow research suggesting showing that the
distribution of autobiographical memory can reflect properties of the prompts used.
In their studies, Rubin & Schulkind (1997c) and Fitzgerald & Lawrence (1984)
found that concrete nouns and words high in imagery generated older memories
than affect words. It is possible the prompt to write about an important person is
relatively concrete and imaginal, drawing up stories about familiar people and faces.
Or the cue may encourage the recollection of more extended memories linked to
earlier life periods. Experiments could be designed to test these and other
explanations.
There were clear differences between the two experiments in regard to
gender. Men in both experiments exhibited almost identical distributions. There
was evidence of a bump, especially for men writing about events/decisions. While
men writing about people reported more memories from childhood, they still
produced a considerable number of memories from the bump years (38.2 percent in
Experiment 1 and 36 percent in Experiment 2). The patterns are quite different
between experiments for female participants, however. While there was evidence
of a bump in Experiment 2, it was entirely absent in Experiment 1, with women
129
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
reporting almost identical proportions of memories from the three life periods. A
closer inspection shows that women in Experiment 1 writing about people wrote a
smaller proportion of memories from the bump years than did women from
Experiment 2 (12.5 percent and 27.3 percent, respectively)^2 (2, n=95)= 5.385,
p = .068). Women in Experiment 1 writing about events/decisions showed some
evidence of a bump, however they produced a considerably larger proportion of
memories from adulthood than did their counterparts from Experiment 2
(40.2 percent vs. 23.9 percent) (x2 (2, n=174)=5.754, p = .056). These marginally
significant differences between experiments were reflected in the multivariate
models. As compared to women, men from Experiment 1 were significantly less
likely to report from childhood or adulthood than the bump. Gender was not a
predictor, however, in Experiment 2.
Several explanations might account for differences between the experiments.
First, the autobiographical memory measure could be unreliable. The same group
of people might show different distributions if they were tested several times, or
different groups of people with similar attributes might generate different patterns
of memories. This explanation seems unlikely since both panels show similar
distributions of memories overall, as well as by topic, and since the overall
distribution is strongly supported in the autobiographical memory literature on the
bump.
130
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
It is also possible that historic events affected the internal validity of the
study. The data examined were gathered in 1994 and 2000. Many political,
economic and cultural events captured the nation’s attention during this seven year
period. Most prominently, some of the first wave of testing coincided with the
fiftieth anniversary of World War II’s end. The many books, plays, television series,
and major motion picture productions that ensued are evidence of continuing
interest in this milestone. Many participants in the study experienced the war as
young adults. Public focus may have caused them to retrieve autobiographical
memories from this period.
Participants in the two experiments might have differed in systematic ways.
However, this explanation also seems unlikely since a test of the covariance
matrices for the nine observed variables of interest indicated no significant
differences between the two data sets fi/2 (36, n=429)=34.533, p= 538).
Finally, the comparison might be confounded if the changes in task
instructions led to selection bias. In 1994-1995, participants self selected whether
to generate memories about people, or events/decisions. However, in 2000-2002
they were randomly assigned to a topic. It is possible that choosing to write about
one topic or another reflects preferences, perhaps based on individual differences,
and that random assignment to topic usurps those preferences. If this were the case,
then there should be significant differences in the proportions of participants from
131
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Experiments 1 and 2 writing on the topics, and significant individual differences
between those choosing and assigned to them.
There was some evidence to support this explanation. A significantly higher
proportion of the participants in Experiment 2 (39.2 percent) were assigned to write
about people than chose to do so in Experiment 1 (29.5 percent) n=519)=
5.394, p = .02), and participants in Experiment 2 generated a larger number of bump
memories whether they wrote about people or events/decisions (See Figures IV.2
and V.2).
Differences between experiments appear to be linked to gender. There was
a significant difference among the proportions of men choosing (26.4 percent) and
assigned (41.3 percent) to write about people (yj( 1, n=250) = 6.268, p = .012).
Assigning men to a topic appeared to limit selection effects. While the men
choosing to write about people in Experiment 1 were younger, had more education,
were more assertive, and performed better on immediate recall than those writing
about event or decisions, only working memory ability distinguished men
participating in Experiment 2, with those assigned to write about people scoring
higher. However, the distribution of memories were relatively similar for men in
both experiments regardless of topic (See Figures IV.4 and V.4). These results
suggest that while men may have preferences in regard to topics for reminiscing,
constraining their choices does not affect the distribution of autobiographical
memories.
132
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Conversely, there was no statistical difference in the proportions of women
choosing (32.8 percent) or assigned (37.4 percent) to write about people, and there
were no mean level differences distinguishing women writing about people or
events/decisions in either data set, suggesting selection effects were not at work.
However, women assigned to write about people produced more than twice the
proportion of memories from the bump as did the women who chose this topic
freely (See Figures IV.5 and V.5). These results suggest that some women may
generate different distributions of autobiographical memories in response to
different prompt types. If this is the case, however, the characteristics
distinguishing the women were not captured.
The two data sets were merged in an effort to appraise the various
explanations. A variable called panel was created to distinguish between those
tested in 1994-1995 (1) and 2000-2002 (2). It was anticipated that this variable
would capture variance associated with the effects of sampling, selection and/or
history. As in Chapters IV and V, a modified sequential approach was taken using
the SPSS NOMREG multinomial logistic regression procedure. Using complete
data, six models were specified (Table VI.3). Panel was entered first (Model 1),
followed by age and education (Model 2),gender (Model 3), topic (Model 4), the
two personality facets (Model 5), and finally the memory measures (Model 6).
Including the variable called panel in Model 1 did not significantly improve
on a constant only model (% 2 (dfi^, n=429)= 3.050, p=.218), suggesting that
133
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
R eproduced with permission of the copyright ow ner. Further reproduction prohibited without p erm ission .
Table VI. 3 Autobiograp lical Memory as a Function of Panel, Gender, Topic, Personality, and Memory (n=429)
y 2 to Remove
M odel 1 M odel 2 M odel 3 M odel 4 Model 5 M odel 6
Panel 3.050 .892 1.729 2.768 3.259 2.777
Background Factors
A ge
Education
5.965
2.003
4.246
.691
4.643
.737
5.964
.898
3.807
1.016
Gender 17.421*** 18.058*** 14.715** 12.799**
Tonic 90.485*** 89.442*** 88.525***
Personality Facets
Tender-mindedness
Assertiveness
5.440
2.510
5.439
2.461
M emory Measures
Immediate Recall
Working Memory
1.245
.311
M odel Statistics
Log-likelihood
chi square test
-2LL=21.718
df=2, x l=3.050
% 2 a=7.059
dfA=4, n.s.
X *a=17.422,
dfA=2, p<.001
X2 a =90.485
dfA=2, p<.001
Z2 a =7.865
dfA=4,p<.01
£ * = 1 .4 1 4
dfA=4, n.s.
Deviance criterion
Nagelkerke R2
Classification
P=-
.008
51.7%
p= .087
.027
51.7%
p= .160
.072
52.2%
p=.892
.277
61.5%
p=.991
.293
61.1%
p=.991
.295
60.6%
*p<.05; **p<.01; ***p<.001
CO
4^
Table VI.4 Multinomial Regression of the Odds of Writing about Childhood or
Adulthood Versus Writing about the Bump (Model 5)_____________________
Child vs. Bumn
B Odds Ratio
(St. Err.)(95% Cl)
Adult-vs. Bump
B Odds Ratio
(St. Err) (95% Cl)
Child- vs.Adult
B Odds Ratio
(St. Err) (95% Cl)
Topic
-Person
-E/D (Ref)
2.187 8.906***
(.263) (5.3-14.9)
2.155
(.341)
8.628***
(4.4-16.8)
Sex
-Male
-Female (Ref)
-.678 .508*
(.268) (.30-.858)
-.978
(.288)
.376**
(.21-
.66)
Age -.038
(.017)
.962*
(.93-
.99)
.040
(.019)
1.041*
(1-0-1.1)
Tender-
mindedness
.089
(.040)
1.094*
(1.0-
1.2)
Note: CI=Confidence Interval
*p<.05; **p<.01; ***p<.001
sampling, selection and history did not play an important role in the differences
found between experiments. Comparisons of Model 2 with Model 1 indicated that
age and education did not contribute to predicting autobiographical memory either
(% 2 (df=4, n=429)=7.059, p<.05). However, the addition of gender in Model 3
(X2 (df=2, n=429)= 17.422, pc.OOl) and topic in Model 4 (% 2 (df=2, n=429)=90.485,
p<.001) did improve the fit. Furthermore, the frequencies predicted by Model 4
reasonably replicated the observed frequencies using the deviance criterion
(p=.892). While a comparison of Model 4 and Model 5 log-likelihood ratios
showed that the addition of personality facets did not improve fit (% 2 (df=4,
135
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
n=429)=7.865, p<.l), tests of individual predictors included in Model 5 correspond
to several theoretically based predictions and should be considered.
Table VI.4 presents the Model 5 odds ratios and confidence intervals
comparing those writing about childhood or adulthood as opposed to the bump, and
about childhood as opposed to adulthood. Those writing about important
individuals were almost nine times more likely to report from childhood than the
bump (eb -8.906, p<.001), and from childhood than adulthood (eb =8.628, p<.001).
Gender emerged as an independent predictor in comparisons of childhood with the
bump (eb = :.508, p<.001), and adulthood with the bump (eb =.376, p<.001). In both
instances, men were more likely to report from young adulthood than women. In
addition to these findings, more tender-minded individuals were more likely to
report from adulthood than the bump period (eb =1.094, p<.05).
VI.B THEORETICAL IMPLICATIONS AND FUTURE
DIRECTIONS
One of the theories relating to the bump proposes that the making of
autobiographical memories is associated with the development of cognitive abilities
across the lifespan. It suggests that the tendency to report from young adulthood
hinges on greater encoding capacity during that life period, or on declining encoding
and retrieval capacities as people age. If this were true, younger participants,
participants who performed well on the memory measures, women, and those with
136
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
more education should produce more memories from adulthood, the most recent
period studied in this dissertation.
Support for the theory was mixed. There was evidence of a bump for both
younger and older participants, who did not statistically differ from each other in
their production of memories from this period. However, age emerged as an
independent predictor when it was used as a continuous variable in the multivariate
models specified for Experiment 2 and the combined data sets. In the combined
data sets, each addition year of life increased a participant’s likelihood of writing
from childhood and the bump as compared to adulthood. On the other hand, there
was no evidence that declining episodic or working memory abilities predicted
older memories. As the theory might predict, there was some evidence women
produced more recent memories in a comparison of the bump and adulthood period.
However, they also produced more memories from childhood. Finally, participants
with more education were no more likely to produce memories from a particular
period than those with less education.
The memory ability hypothesis is just one explanation for the tendency of
older individuals to be more “past oriented” (Sperbeck, Whitboume & Hoyer,
1986,171). Others include the cognitive rapid change theory and the contextual self
narrative hypothesis. The first suggests that memories from the bump predominate
because they are novel, because they tend to be encoded during periods of
transition, and because they benefit from more frequent rehearsal and retrieval
137
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(Rubin, 1998); the second that the bump reflects the increased cognitive effort
required as young people create self narratives based on important memories
(Fitzgerald, 1996). Devising hypotheses that distinguish between the two theories is
difficult, and the researchers allow that these explanations may be linked
(Fitzgerald, 1986; Rubin, Rahhal & Poon, 1998). That is, the self narrative
hypothesis may function through the mechanisms of the rapid change theory.
The theories were examined by testing for gender differences in the
distribution of autobiographical memory. It was hypothesized that men would be
more likely to report from the bump period than women, whose self narratives
might be thought of as works in progress linked to the experiences of others, and
whose life transitions may therefore be spread more evenly across the life course.
There was some evidence supporting this hypothesis in Panel 1 and in the combined
data sets. However, the data did not support the expectation that those with higher
levels of education, an indicator of the timing of normative events or alternately of
occupational status, would exhibit more bump memories. The hypothesis was also
extended to the communal and agentic “modalities” of being (Bakan, 1966), which
relate to an individual’s orientation toward social linkages. Women have
traditionally been expected to behave more communally and men more agentically
(Eagly, 1987; Bakan, 1966). These traits were captured by the NEO-PIR facet
called tender-mindedness and assertiveness, respectively. Data partially supported
the expectation that those who are more tender-minded would report memories from
138
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
across the lifespan in the combined data sets, however there was no evidence that
those who are more assertive reported more memories from the bump period.
Future tests of the theories could benefit from cohort analysis. Demographic
changes, changes in the structure of the family, educational improvement, and the
feminist movement have affected women’s social roles (Zollinger Giele, 1993).
Over the course of the last century, historical events such as World War II and the
Great Depression, changing economic conditions encouraging dual earner
households, and a growing demand for teachers, office and health care workers have
caused larger proportions of women to enter the workforce. The life course
perspective proposes such historic events and trends can change patterns of
transitions (Elder & Johnson, 2000). Differences in the autobiographical memory
distributions of younger and older cohorts of women, suggesting autobiographical
memory is affected by social roles and societal expectations, would provide indirect
support for the self narrative/rapid change theories.
The distinction between specific and general memories could be used to
further examine the strong relationship between topic and autobiographical memory
distribution found in the data. Does the distribution of memories differ by topic
because the cues encourage the generation of more or less specific personal
memories? Is the specificity of personal memories associated with the life periods
from which they are generated? And could age, gender, education, and the
139
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
personality and memory measures be used to predict the generation o f more or less
specific memories?
Finally, the differences in results found between the two experiments
indicate that sampling and selection issues should also be investigated. While the
variable called panel did not contribute to prediction in the multivariate analysis of
the combined data sets, more deliberate experiments could be undertaken. For
example, autobiographical memory data gathered for Panel 2 in 1997 and 2000
could be compared to determine whether participants assigned to report on either
people or events/decisions generate memories consistently over time. The Panel 2
data gathered in 1994 should not be included in this analysis since the instructions
for the task were changed in 1997. In order to determine whether participants are
consistent in choosing between topics, the original instructions might be used
during the next wave of testing, allowing for a comparison of Panels 2 and 4. It
would also be interesting to compare data from this experiment with data gathered
for a traditional word cue experiment to determine whether reporting is consistent
across the two methods.
Perhaps the most important contribution of this dissertation is the attempt to
develop linkages between theory and hypotheses predicting gender differences in
the distribution of autobiographical memory. The psychology of aging has been
criticized for studying gender only in passing (Sinnott & Shifren, 2001). The
authors suggest that while gender is often included as a variable, it tends not to be a
140
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
main focus, and hence little effort is made to explain why and how it should be
important. As a result, findings relating to gender appear peripheral and
uninteresting. This analysis certainly applies to the study of the distribution of
autobiographical memory. Very few studies of the bump have explicitly focused on
gender. Those that have were part of larger, well established research programs
primarily testing cognitive theories. Many of the same criticisms could also be
leveled at this dissertation, which is based on an instrument designed to test for
developmental changes in language production.
If scientists are interested in discovering whether there are gender
differences in the bump, they will have to plan studies specifically for that purpose.
For example, it would be interesting to test whether the bump contains more
memories relevant to the self than other life periods, whether having memories
relevant to the self is predicted by an independent self construal, and whether the
independent self construal is linked to gender. Accomplishing this would require
good measures of independent and interdependent self construal, perhaps by
appraising the closeness and significance of relationships, and a precise method for
analyzing the content of memories and their relevance to the self. Women tended to
report a somewhat larger number of significant life events, and their memories
from their bump years fell into different content domains than similarly aged men
(Brady Elnick et al., 1999). This line of research should pursued, perhaps using
hypotheses derived from social role theory. For example, it would be interesting to
141
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
determine whether gendered social roles, as measured by caregiving and kinkeeping
activities, are reflected in the timing of turning points and the content of the
memories reported.
Gerontology is striving to become an interdisciplinary field, to enrich the
study of aging by acknowledging the importance not only of biology and
development, but also of cultural and historical factors relating to the life course.
This dissertation reflects those ambitions by including variables conceptually linked
to social roles. The results of the research are not entirely consistent across the two
experiments, and hypotheses relating to education, assertiveness and memory had to
be rejected outright. In part this may reflect that the measure used to tap
autobiographical memory was originally designed for a different purpose.
Nevertheless, there was some evidence suggesting that the distribution of
autobiographical memory may differ for men and women, and for individuals based
on their orientation towards others. These findings follow research regarding other
aspects of autobiographical memory, and they have theoretical implications for our
understanding of the cause and purpose of the bump.
142
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
REFERENCES
Baddeley, A. (1989). Finding the bloody horse. In L.W. Poon, D.C. Rubin, &
B.A.Wilson (Eds.), Everyday cognition in adulthood and late life (pp. 104-
115). New York: Cambridge University Press.
Bahrick, H. P.(1989). The laboratory and ecology: Supplementary sources of data
for memory research. In L.W. Poon, D.C. Rubin, & B.A.Wilson (Eds.),
Everyday cognition in adulthood and late life (pp.73-83). New York:
Cambridge University Press.
Bakan, D. (1966). The duality o f human existence: An essay on psychology and
religion. Chicago: Rand McNally and Company.
Bekerian, D. A., & Dritschel, B. H., (1992). Autobiographical remembering: An
integrative approach. In M. A. Conway, D.C. Rubin, H. Spinnler, & W.A.
Wagenaar (Eds.), Theoretical perspectives on autobiographical memory (pp.
135-150). Dordrecht: Kluwer Academic Publishers.
Benson, K.A., Jarvi, S.D., Arai,Y., Thielbar, P. R.S., Fry, K. J., & Goracke
McDonald, B. L. (1992). Socio-historical context and autobiographical
memories: Variations in the reminiscence phenomenon. In M. A. Conway,
D. C. Rubin, H. Spinnler, & W. Wagenaar (Eds.), Theoretical perspectives
on autobiographical memory (pp. 313-322). Dordrecht: Kluwer Academic
Publishers.
Boring, E., & Linzey, G. (1967). A history o f psychology in autobiography (Vol. 5).
New York: Appleton-Century-Crofts.
Brady Elnick, A., Margrett, J. A., Fitzgerald, J. M., & Labouvie-Vief, G. (1999).
Benchmark memories in adulthood: Central domains and predictors of their
frequency. Journal o f Adult Development, 6, 45-59.
Brewer, W. F. (1986). What is Autobiographical Memory. In D.C. Rubin (Ed),
Autobiographical memory (pp. 25-49). Cambridge: Cambridge University
Press.
Brewer, W. F. (1988). Memory for randomly sampled autobiographical events. In
U. Neisser, & E. Winograd (Eds.), Remembering reconsidered: Ecological
and traditional approaches to the study o f memory (pp. 193-243).
Cambridge: Cambridge University Press.
143
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Brewer, W. F. (1996). What is recollective memory? In D. C. Rubin (Ed.),
Remembering our past: Studies in autobiographical memory (pp. 19-66).
New York: Cambridge University Press.
Brittlebank, A. D., Scott, J., Williams, J. M. G., & Ferrier, I. N. (1993).
Autobiographical memory in depression: State or trait marker. British
Journal o f Psychiatry, 162, 118-121.
Buckner, J. P., & Fivush, R. (2000). Gendered themes in family reminiscing.
Memory, 8, 401-412
Brown, R., & Kulik, J. (1977). Flashbulb memories. Cognition, 5, 73-99.
Crovitz, H. F., & Schiffinan, H. (1977). Frequency of episodic memories as a
function of their age. Bulletin o f the Psychonomic Society, 4, 517-518.
Coltrane, S. (2000). Research on household labor: Modeling and measuring the
social embeddedness of routine family work. Journal o f Marriage and the
Family, 62, 1208-1233.
Conway, M. A. (1996). Autobiographical knowledge and autobiographical
memories. In D. C. Rubin (Ed.), Remembering our past: Studies in
autobiographical memory (pp. 67-93). Cambridge: Cambridge University
Press.
Conway, M. A.,& Haque, S. (1999). Overshadowing the reminiscence bump:
Memories of a struggle for independence. Journal o f Adult Development, 6,
35-44.
Conway, M. A., Pleydell-Pearce, C. W., & Whitecross, S. E. (2001). The
neuroanatomy of autobiographical memory: A slow cortical potential study
of autobiographical memory retrieval. Journal o f Memory and Language,
45, 493-524.
Conway, M. A., & Rubin, D. C. (1993). The structure of autobiographical memory.
In A. F. Collins, S. E. Gathercole, M. A. Conway, & P. E. Morris (Eds.),
Theories o f Memory (pp. 103-137). Hillsdale: Lawrence Erlbaum
Associates, Publishers.
Conway, M. A., Turk, D. J., Miller, S.L, Logan, J., Nebes, R.D., Meltzer, C.C., &
Becker, J.T. (1998). The neuroanatomical basis of autobiographical
memory. Unpublished.
144
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Costa, P. T., & McCrae, R. R. (1992). Revised NEO Personality Inventory (NEO
PI-R) and NEO Five-Factor Inventory (NEO-FFI). Odessa, FL:
Psychological Assessment Resources, Inc.
Cowan, N., & Davidson, G. (1984). Salient childhood memories. Journal o f Genetic
Psychology, 145,101-107.
Dannefer, D. (1984). Adult development and social theory: A paradigmatic
reappraisal. American Sociological Review, 49, 100-116.
Davis, P. J. (1999). Gender differences in autobiographical memory for childhood
emotional experiences. Journal o f Personality and Social Psychology, 76,
498-510.
Day Hulbert, K. (1993). Reflections on the lives of educated women. In K. Day
Hulbert & D. Tickton Schuster (Eds.), Women's lives through time:
Education American women o f the twentieth century (pp. 417-443). San
Francisco: Jossey-Bass Publishers.
Eagly, A. H. (1987). Sex differences in social behavior: A social-role interpretation.
Hillsdale: Lawrence Erlbaum Associates, Publishers.
Elder, G. H., & Kirkpatrick Johnson, M. (2002). The life course and aging:
Challenges, lessons, and new directions. In R.A. Settersten (Ed.), Invitation
to the life course: Toward new understanding in later life, Part II, Chapter
2, (pp. 49-81). Amityville, NY: Baywood.
Feingold, A. (1994). Gender differences in personality: A meta-analysis.
Psychological Bulletin ,116:429-456.
Fitzgerald, J. M. (1980). Sampling autobiographical memory reports in adolescents.
Developmental Psychology, 16, 675-676.
Fitzgerald, J.M. (1981). Autobiographical memory reports in adolescence.
Canadian Journal o f Psychology, 35, 69-73.
Fitzgerald, J. M. (1986). Autobiographical memory: A developmental perspective.
In D.C. Rubin (Ed.), Autobiographical memory (pp. 122-133). Cambridge:
Cambridge University Press.
Fitzgerald, J. M. (1988). Vivid memories and the reminiscence phenomenon: The
role of the self narrative. Human Development, 31, 261-273.
145
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Fitzgerald, J. M. (1992). Autobiographical memory and conceptualizations of the
self. In M.A.Conway, D.C. Rubin, H. Spinnler, & W.A. Wagenaar (Eds.),
Theoretical perspectives on autobiographical memory (pp. 99-114). Boston:
Kluwer Academic.
Fitzgerald, J. M. (1996). Intersecting meanings of reminiscence in adult
development and aging. In D.C. Rubin (Ed.), Remembering our past:
Studies in autobiographical memory (pp. 360-383). Cambridge: Cambridge
University Press.
Fitzgerald, J. M., & Lawrence, R. (1984). Autobiographical memory across the life
span. Journal o f Gerontology, 39, 692-699.
Fivuch, R. (1998). Gendered narratives: Elaboration, structure, and emotion in
parent-child reminiscing across the preschool years. In C.P. Thompson
(Ed.), Autobiographical memory: Theoretical and applied perspectives (pp.
79-103). Mahway, NJ: Lawrence Erlbaum Associates.
Franklin, H. C., & Holding, D. H. (1977). Personal memories at different ages.
Quarterly Journal O f Experimental Psychology, 29, 527-532.
Friedman, A., & Pines, A. (1991). Sex differences in gender-related childhood
memories. Sex Roles, 25, 25-32.
Fromholt, P., & Larsen, S. F. (1991). Autobiographical memory in normal aging
and primary degenerative dementia. Journal o f Gerontology: Psychological
Sciences, 46, 85-91.
Fromholt, P., Larsen, P., & Larsen, S. F. (1995). Effect of late-onset depression and
recovery on autobiographical memory. Journal o f Gerontology:
Psychological Sciences, 50, 74-81.
Galton, F. (1879). Psychometric experiments. Brain, 2,149-162.
Gilligan, C. (1982). In a different voice: Psychological theory and women's
development. Cambridge, MA: Harvard University Press.
Hagestad, G. O. (1986). Sex and gender in the aging society. In A. Pifer, & L.
Bronte (Eds.), Our aging society (pp. 141-160). New York: W.W.Norton &
Company.
146
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Hentges, B. (1997). The reminiscence bump: The distribution and characteristics of
autobiographical memories. Unpublished Dissertation, University of
Houston, Houston, TX.
Herlitz, A., Airaksinen, E., & Nordstrom, E. (1999). Sex differences in episodic
memory: The impact of verbal and visuospatial ability. Neuropsychology,
13, 590-597.
Herlitz, A., Nilsson, L. G., & Backman, L. (1997). Gender differences in episodic
memory. Memory and Cognition, 25(6), 801-811.
Hoare, C. H. (2002). Erikson on development in adulthood: New insights from the
unpublished papers. Oxford: Oxford University Press.
Holmes, A., & Conway, M. A. (1999). Generation identity and the reminiscence
bump: Memory for public and private events. Journal o f Adult Development,
6,21-34.
Hosmer, D. W., & Lemeshow, S. (2000). Applied logistic regression. USA: John
Wiley & Sons, Inc.
Howes, J. L., & Katz, A. N. (1992). Remote memory: Recalling autobiographical
and public events across the lifespan. Canadian Journal o f Psychology, 46,
92-116.
Hulbert, K. D. (1993). Reflections on the lives of educated women. In K.D. Hulbert,
& D.T. Schuster (Eds.), Women's lives through time: Educated American
women o f the 20th century (pp. 417-444). San Francisco: Jossey-Bass
Publishers.
Hultsch, D. F., Hertzog, C., Dixon, R. A., & Small, B. J. (1998). Memory change in
the aged. New York: Cambridge University Press.
Hyland, D. T., & Ackerman, A. M. (1988). Reminiscence and autobiographical
memory in the study of the personal past. Journal o f Gerontology:
Psycholocial Sciences, 43, 35-39.
Jansari, A.,& Parkin, A. J. (1996). Things that go bump in your life: Explaining the
reminiscence bump in autobiographical memory. Psychology and Aging, 11,
85-91.
147
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Kuhn, T. S. (1996). The structure o f scientific revolutions (3rd ed.). Chicago:
University of Chicago Press.
Linzey, G. (1974). A history o f psychology in autobiography (Vol. 6). Englewood
Cliffs, NJ: Prentice-Hall.
Linzey, G. (1979). A history o f psychology in autobiography (Vol. 7). San
Francisco: Freeman.
MacDonald, S., Uesiliana, K., & Hayne, H. (2000). Cross-cultural and gender
differences in childhood amnesia. Memory, 8, 365-376.
Mackavey, W. R., Malley, J. E., & Stewart, A. J. (1991). Remembering
autobiographically consequential experiences: Content analysis of
psychologists' accounts of their lives. Psychology and Aging, 6, 50-59.
Markus, H. R., & Kitayama, S. (1991). Culture and the self: Implications for
cognition, emotion and motivation. Psychological Review, 98, 224-253.
Moen, P. (1995). Gender, age, and the life course. In R. H. Binstock, & L. K.
George (Eds.), Handbook o f aging and the social sciences (4th ed., pp.171-
187). San Diego: Academic Press.
Montgomery, R. J. V., & Datwyler, M. M. (1992). Women and men in the
caregiving role. In J. Hendricks, & L. Glasse (Eds.), Gender and aging (pp.
59-68). Amityville, NY: Baywood Publishing Company, Inc.
Mullen, M. K. (1994). Earliest recollections of childhood: A demographic analysis.
Cognition, 52, 55-79.
Neisser, U. (1982). Memory: What are the important questions? In
M.M.Gruneberg, P.E. Morris, & R.N. Sykes (Eds.), Practical aspects of
memory (pp.3-24). London: Academic Press.
Neisser, U. (1982) Memory observed: Remembering in natural contexts. San
Francisco: W.H. Freeman.
Nyberg, L., Backman, L., Emgrund, K., Olofsson, U., & Nilsson, L. G. (1986). Age
differences in episodic memory. Journal o f Gerontology: Psychological
Sciences, 5 IB, 234-240.
148
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
O'Connell, A., & Russo, N. F. (Eds.). (1983). Models o f achievement: Reflections o f
eminent women in psychology. New York: Columbia University Press.
Petrinovich, L. (1989). Representative design and the quality of generalization. In
L.W. Poon, D.C. Rubin, & B.A. Wilson (Eds.), Everyday cognition in
adulthood and late life (pp. 11-24). New York: Cambridge University Press.
Robinson, J. A. (1976). Sampling autobiographical memory. Cognitive Psychology,
8, 578-595.
Robinson, J. A. (1986). Autobiographical memory: A historical prologue. In D.C.
Rubin (Ed.), Autobiographical memory (pp. 19-24). Cambridge: Cambridge
University Press.
Rubin, D. C. (1980). 51 properties of 125 words: A unit analysis of verbal behavior.
Journal o f Verbal Learning and Verbal Behavior, 19, 736-755.
Rubin, D. C. (1982). On the retention function for autobiographical memory.
Journal o f Verbal Learning and Verbal Behavior, 21,21-38.
Rubin, D. C. (1986). Introduction. In D.C. Rubin (Ed.), Autobiographical memory,
(pp. 3-24). Cambridge: Cambridge University Press.
Rubin, D. C. (1989). Introduction to Part 1: The how, when, and why of studying
everyday cognition. In L.W. Poon, D.C. Rubin, & B.A. Wilson (Eds.),
Everyday cognition in adulthood and late life (pp. 3-10). New York:
Cambridge University Press.
Rubin, D. C. (1989). Issues of regularity and control: Confessions of a regularity
freak. In L.W. Poon, D.C. Rubin, & B.A. Wilson (Eds.), Everyday
cognition in adulthood and late life (pp. 84-103). New York: Cambridge
University Press.
Rubin, D.C.(1996). Introduction. In D.C. Rubin (Ed.), Remembering our past:
Studies in autobiographical memory, (pp. 1-15). Cambridge: Cambridge
University Press
Rubin, D. C. (1998). Autobiographical memory and aging: Distributions of
memories across the lifespan and their implications for survey research. In
N. Schwarz, D. Park, B. Knauper, & S. Sudman (Eds.), Cognition, aging
and self reports (pp. 163-183). U.S.A.: Taylor & Francis.
149
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Rubin, D. C. (2000). The distribution of early childhood memories. Memory, 8,
265-269.
Rubin, D. C., Rahhal, T. A., & Poon, L. W. (1998). Things learned in early
adulthood are remembered best. Memory and Cognition, 26, 3-19.
Rubin, D. C., & Schulkind, M. D. (1997a). The distribution of important and word
cued autobiographical memories in 20, 35 and 70 year old adults.
Psychology and Aging, 12, 524-535.
Rubin, D. C., & Schulkind, M. D. (1997b). The distribution of autobiographical
memories across the lifespan. Memory and Cognition, 25, 859-866.
Rubin, D.C., & Schulkind, M.D. (1997c). Properties of word cues for
autobiographical memory. Psychological Reports, 81,47-50.
Rubin, D. C., Schulkind, M. D., & Rahhal, T. A. (1999). A study of gender
differences in autobiographical memory: Broken down by age and sex.
Journal o f Adult Development, 6, 61-71.
Rubin, D. C., Wetzler, S. E., & Nebes, R. D. (1986). Autobiographical memory
across the adult life span. In D.C. Rubin (Ed.), Autobiographical memory
(pp.202-224). Cambridge: Cambridge University Press.
Ruth, J. E., & Vilkko, A.(1996). Emotions in the construction of autobiography. In
C. Magain, & S. H. McFadden (Eds.), Handbook o f emotion, adult
development and aging (pp. 167-181). San Diego: Academic Press, Inc.
Rybash, J. M. (1999). Aging and autobiographical memory: The long and bumpy
road .Journal o f Adult Development, 6, 1-10.
Sales, E. (1978). Women's adult development. In I.H. Frieze, J.E. Parsons, P.B.
Johnson, D.N. Ruble, & G.L. Zellman (Eds.), Women and sex roles: A social
psychological perspective (pp. 157-190). New York: WW Norton and
Company.
Salthouse, T. A. (1991). Theoretical perspectives on cognitive aging. Hillsdale, NJ:
Lawrence Erlbaum Associates, Publishers.
Salthouse, T. A., & Babcock, R. L. (1991). Decomposing adult age-differences in
working memory. Developmental Psychology, 27(5), 763-776.
150
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Schacter, D. L. (1996). Searching for memory: The brain, the mind, and the past.
New York: BasicBooks.
Seidlitz, L., & Diener, E. (1998). Sex differences in the recall of affective
experiences. Journal o f Personality and Social Psychology, 74(1), 262-271.
Shelton, B.A., & John, D. (1996). The division of household labor. Annual Review
o f Sociology, 22, 229-322.
Sinnott, J. D., & Shifren, K. (2001). Gender and aging: Gender differences and
gender roles. In J. E. Birren, & K. W. Schaie (Eds.), Handbook o f the
psychology o f aging (pp. 454-476). San Diego: Academic Press.
Sperbeck, D., Whitboume, S., & Hoyer, W. (1986). Age and openness to experience
in autobiographical memory. Experimental Aging Research, 12,169-172.
Tabachnick, B. G., & Fidell, L. S. (2001). Using multivariate statistics. (4th ed.).
Boston: Allyn and Bacon.
Tulving, E. (1983). Elements o f Episodic Memory. New York: Oxford University
Press.
Wang, Q., Leichtman, M. D., & White, S. H. (1998). Childhood memory and self
description in young Chinese adults: The impact of growing up an only
child. Cognition, 68,73-103.
Webster, J. D., & Cappeliez, P. (1993). Reminiscence and autobiographical
memory: Complementary contexts for cognitive aging research.
Developmental Review, 13, 54-91.
Wheeler, M. A., Stuss, D. T., & Tulving, E. (1997). Toward a theory of episodic
memory: The frontal lobes and autonoetic consciousness. Psychological
Bulletin, 121,331-354.
Williams, J. M.G. (1996). Depression and the specificity of autobiographical
memory. In D. Rubin (Ed.), Remembering our past: Studies in
autobiographical memory (pp. 244-267). Cambridge: Cambridge University
Press.
Williams, J. M.G. & Broadbent, K. (1986). Autobiographical memory in attempted
suicide patients. Journal o f Abnormal Psychology, 95,144-149.
151
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Williams, J. M.G. & Dritschel, B. H. (1988). Emotional disturbance and the
specificity of autobiographical memory. Cognition and Emotion, 2, 221-234.
Williams, J. M.G. & Dritschel, B. H. (1992). Categorical and extended
autobiographical memories. In M. A. Conway, D. C. Rubin, H. Spinnler, &
W. A. Wagenaar (Eds.), Theoretical perspectives on autobiographical
memory (pp. 391-412). Dordrecht: Kluwer Academic.
Williams, J. M.G. & Scott, J. (1988). Autobiographical memory in depression.
Psychological Medicine, 18, 689-695.
Zelinski, E. M. & Bumight, K. P. (1997). Sixteen-year longitudinal change and time
lag changes in memory and cognition in older adults. Psychology and Aging,
7, 503-513.
Zelinski, E.M. (1999). The Long Beach Longitudinal Study: A sampler of findings.
Symposium presented at the annual meeting of the Gerontological Society
on Aging, San Francisco.
Zelinski, E. M., Gilewski, M. J., & Schaie, K. W. (1993). Individual differences in
cross sectional and 3-year longitudinal memory performance across the adult
life span. Psychology and Aging, 8,176-186.
Zola-Morgan, S., Cohen, N. J., & Squire, L. R. (1983). Recall of remote episodic
memory in amnesia. Neuropsychologia, 21, 487-500.
Zollinger Giele, J. (1993). Women's role change and adaptation, 1920-1990. In K.
D. Hulbert, & D. Tickton Schuster (Eds.), Women's lives through time:
Educated American women o f the twentieth century (pp. 32-60). San
Francisco: Jossey-Bass Publishers.
152
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Evaluation of the Care Advocate Program: Bridging managed care and home community -based services

PDF

Heart disease among middle -aged and elderly persons in the United States: Trends and a multistate model

PDF

Filial expectations and social exchange patterns among older Taiwanese parents and their adult children

PDF

Education and intelligence test scores: Predictors of dementia?

PDF

In and out of marriage: A study of the multiply divorced

PDF

Course completion and older adult students: The effects of self-efficacy, self -regulation, and motivation

PDF

A longitudinal examination of gender and ethnicity across two models of community treatment for individuals with schizophrenia and related disorders

PDF

I. Oxidation of diazo phosphonates and diazo bisphosphonates. II. NMR and crystallographic studies of bioactive semicarbazides and Schiff bases of hydroxyguanidine

PDF

Effects of migration and modernization on perceptions of role significance and status among elderly Samoans in the Los Angeles area

PDF

Intergenerational social support and the psychological well-being of older parents in China

PDF

Determinants of life satisfaction among the high-functioning elderly: Implications of effects of psychological factors and activity participation

PDF

Coevolution of technology and organization: A case study of BrokerHouse

PDF

Cross -cultural examination of mental health measures in dementia caregivers: Assessment of the Center for Epidemiologic Studies -Depression Scale and the Zarit Burden Inventory

PDF

At what age are gait characteristics mature? Evaluation of gait kinematics, kinetics, and intersegmental dynamics in 7 year-old children

PDF

Effects of cultural values and behaviors in sociocultural stress and coping among Korean-American and Korean caregivers

PDF

An analysis of environmental conflicts involving transnational corporations, states and environmentalists in Latin America

PDF

Acquisition, consolidation and storage of an associative memory in the cerebellum

PDF

An adaptive temperament -based information filtering method for user -customized selection and presentation of online communication

PDF

Cost -efficient design of main cohort and calibration studies where one or more exposure variables are measured with error

PDF

Insurance mechanisms, forest clearance, and the effect of government policies in rural economies

Asset Metadata

Creator Housen, Patricia (author)

Core Title Gender and the bump: An investigation of the reminiscence effect in the Long Beach Longitudinal Study

School Graduate School

Degree Doctor of Philosophy

Degree Program Gerontology

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

Tag Gerontology,OAI-PMH Harvest

Language English

Contributor Digitized by ProQuest (provenance)

Advisor Zelinski, Elizabeth M. (committee chair), Crimmins, Eileen M. (committee member), Kempler, Daniel (committee member), Liebig, Phoebe (committee member), Wilber, Kathleen (committee member)

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c16-373500

Unique identifier UC11334926

Identifier 3103903.pdf (filename),usctheses-c16-373500 (legacy record id)

Legacy Identifier 3103903.pdf

Dmrecord 373500

Document Type Dissertation

Rights Housen, Patricia

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