PERSONALITY PROCESSES AND INDIVIDUAL

DIFFERENCES

Psychosocial Development in Adulthood: A 22^ear Sequential Study

Susan Krauss Whitbourne Michael K. Zuschlag
University of Massachusetts at Amherst Georgia Southern University

Lisa B. Elliot Alan S. Waterman

State University of New York at Geneseo Trenton State College

Data supporting the notion of adult personality stability are challenged by the presentfindings,in
which developmental change was demonstrated using the Eriksonian-stage-based Inventory of
Psychosocial Development (IPD; Constantinople, 1969). A sequential design over the ages 20-42
was used on 2 cohorts of college students and alumni originally tested in 1966 and 1976-1977 («s in
1988 = 99 and 83, respectively), and a 3rd cohort of college students in 1988-1989 (n = 292). Results
of longitudinal, cross-sectional, and sequential analyses challenged ideas about personality stabil-
ity, with evidence of increasingly favorable resolutions of the early Eriksonian psychosocial stages
up through the oldest age studied. There was evidence of a trend over the past decade toward less
favorable resolution of ego integrity versus despair. The findings were interpreted in terms of
developmental change processes during the adult years interacting with culturally based environ-
mental effects on psychosocial development.

In the past 10 years, a coherent and convincing body of data and theoretical divergence between the advocates of adult per-
has accumulated to support the position that personality in sonality stability and those who propose that there are stages in
adulthood changes little, if at all, after the age of 30 years (Con- personality development (e.g., Whitbourne, 1986). In part, the
ley, 1985; Finn, 1986; McCrae & Costa, 1990; Schaie & Parham, divergence of theory and data may be due to differing defini-
1976). Such research indicates that, for example, individuals tions of personality and consequently, differences in the vari-
who are extraverted at age 30 tend to remain extraverted 10 and ables selected for measurement. Researchers arguing for stabil-
20 years later. This view of the adult personality as stable con- ity in adulthood have typically focused on personality traits
trasts sharply with the position advocated by theorists such as that are, by definition, inherently stable dispositions. Measures
Erikson (1963), who regard adulthood as a time of continued of personality traits undergo extensive psychometric refine-
psychosocia! evolution common to most adults. Stage views of ment until they are proven to be consistent indicators of temper-
adult development (e.g., Levinson, Darrow, Klein, Levinson, & ament that are relatively impervious to errors of measurement
McK.ee. 1978) build on this notion that the adult personality from testing to testing. Thus, it is not necessarily surprising to
has the potential to undergo major transitions at predictable find a lack of intraindividual change over time on such indexes
intervals. as reported, for example, by McCrae and Costa (1990). Unan-
Various attempts have been made to resolve the empirical swered by such research is the question of whether change in
adulthood would be observed on indexes of personality func-
tioning that are intended to be sensitive to developmental pro-
cesses.
This research was supported in part by a Biomedical Research Sup-
port Grant to Susan Krauss Whitbourne from the University of Massa- Erikson's (1963) theory is an attempt to conceptualize in a
chusetts at Amherst. coherent fashion a set of theoretical effects of environmental
We wish to thank the following undergraduatestudentsat the Univer- and biological forces on personality development throughout
sity of Massachusetts: Pamela Diamond, Audrey Dumper, Edward the life span. This theory is regarded as the quintessential exam-
Adamson. Donna Fuhrmann, Karen McGee, Dina Litvack, Rebecca
Potts. Susan Read. Peter Wood, and Diana Stern. We also thank the
ple of a theory of personality "change" in adulthood, with
following Geneseo State College students: Melissa Alexander, Joanne change following a sequential arrangement in which stages un-
Fusarc, Christine Giovanniello, and Kristine Pucher. The invaluable fold in varying degrees of regularity depending on a constella-
assistance of Roger Latham and hisstaffat the University of Rochester tion of biological, psychological, and social-historical forces.
is also gratefully appreciated. Each crisis stage is theorized to build on the preceding ones and
Correspondence concerning this article should be addressed to to influence the outcome of successive ones, according to Erik-
Susan Krauss Whitbourne. Department of Psychology, Tobin Hall, son's epigenetic principle.
University of Massachusetts. Amherst, Massachusetts 01003. A point usually overlooked in descriptions of Erikson's (1963)

Journal of Personality and Social Psychology. 1992. Vol.63. No. 2. 260-271

Copyright 1992 by the American Psychological Association. Inc. 0022-3514/92/S3.00

260
 PSYCHOSOCIAL DEVELOPMENT 261

theory is that the proposed stage scheme is not specific by age, of the eight psychosocial stages described by Erikson (1963).
nor is it exclusively unidirectional. In line with the epigenetic An expanded IPD, measuring all eight stages, was constructed
principle, there is for healthy personality "a progression in 1976 and administered at that time. The scales measuring the
through time of a differentiation of parts" (Erikson, 1959/ final two stages are reported in more detail by Walaskay, Whit-
1980, p. 54) involving a succession of potentialities for signifi- bourne, and Nehrke (1983-1984).
cant interactions with people and institutions optimally occur- The current article expands on Whitbourne and Waterman's
ring within "the proper rate and the proper sequence" (Erikson, (1979) previous study in the following ways: (a) the maximum
1959/1980, p. 54). However, developmental concepts used by age has been extended beyond age 31 to age 42, to include
Erikson (1959/1980, 1963) such as vertical compensation and cross-sectional and longitudinal analysis of a period of person-
regression (vertical decompensation) suggest that an individ- ality development often considered to be highly stable; (b) lon-
ual's life circumstances and the social environment may mark- gitudinal data are present for two cohorts between the ages of
edly alter the timing and patterns of psychosocial development. 20 and 31, allowing greater internal validity through applica-
Theoretically, for any stage component, movement toward in- tion of sequential methodology to this age interval as tested
creased or decreased psychological health could occur at any over three measurement periods; and (c) contemporary statisti-
point in the life cycle as a function of history-graded or idiosyn- cal packages allow the data to be submitted to multivariate
cratic life events. analyses of variance (MANOVAs) to gain power, control for
Researchers investigating Erikson's (1963) theory and the family-wise error rate, provide a more robust repeated-mea-
more general issues regarding stability and change in adult per- sures analysis, and determine more efficiently the interplay of
sonality are encumbered by the potential confounding of age the IPD scales.
and period effects in developmental research designs, all of Three questions were explored: (a) Does intraindividual adult
which represent threats to internal validity (Schaie, 1988). Erik- personality change occur on variables theoretically expected to
son's discussion of adult personality development implies that be sensitive to developmental changes in adulthood? (b) Given
not only does personality change with age, but that similar that such changes are found, to what extent to they reflect age-
changes are often likely to occur for different people at about related developmental processes and to what extent do they
the same age. Researchers seeking to test this proposition must reflect the effects of acculturation or environmental influ-
contend with the fact that if age differences in personality are ences? and (c) Given that personality changes or differences are
found, caution must be taken before concluding that these dif- found, whatever the apparent cause, to what degree are the
ferences represent a developmental trend in personality com- results consistent with Erikson's (1963) theory of personality?
mon to most people. For cross-sectional designs, what appear Specifically, the Eriksonian stages theorized to have greatest
to be age "changes" may really be cohort differences, the prod- relevance for each cohort should be most sensitive to age and
uct of comparing individuals of different cohorts, born in dif- time effects, with intimacy issues being of most concern to
ferent decades or generations, who were exposed at birth to individuals in their 20s and generativity being more salient to
differing environmental influences on their personality devel- individuals in their 30s and early 40s. Short of such specific
opment. Furthermore, people within the same birth cohort age-stage trends, a more general finding of greater movement
continue to receive differential exposure to contemporary his- in the latter four stages compared with the earlier four stages
torical and social conditions throughout their adult lives com- also was predicted, in keeping with the previous follow-up's
pared with people of other cohorts (e.g., Elder, 1974). Thus, even results.
if personality changes are detected in a longitudinal study of Guiding the analyses presented in this study is the notion
adults of one cohort, such changes do not necessarily occur in that the interpretation of general aging effects is more plausible
other cohorts. for some patterns of results than others (Costa & McCrae,
The problems of separating age-related changes common to 1982). If personality change is entirely the product of aging
most adults from effects specific to cohorts or historical eras rather than historical or cultural changes, then the examination
can be managed by gathering data on several different samples of sequential data should reveal that general pattern, with
spanning different ages and different times of testing. Only a adults from any cohort or tested at any time of measurement
handful of studies have used these more sophisticated sequen- receiving the same personality score for any given age and
tial designs recommended by Schaie (1965) and Baltes (1968). showing a similar degree of change between ages.
Although no design can definitely lead to the conclusion that
any longitudinal changes measured will be replicated in sam-
Method
ples of different historical and cultural backgrounds, the se-
quential approach provides greater internal validity than the Design
traditional single cohort or single time of measurement designs.
In the present research, which is an extension of a previous The design of the present study is shown in Figure 1 and the accom-
panying Table 1, which describe in detail the variables used in each of
longitudinal and time-lag study of psychosocial development the multivariate designs conducted on the IPD scores. Because the
in early adulthood (Whitbourne & Waterman, 1979), a sequen- expanded IPD had not been developed for use in the first testing of
tial strategy was applied to responses to an expanded version of Cohort 1, analyses including Stages 7 and 8 could be conducted only on
the Inventory of Psychosocial Development (IPD), a question- the analyses involving 1977 and 1988 times of testing. Gender was used
naire measure of Eriksonian personality development designed as a between-subjects variable in all analyses.
by Constantinople (1969). This measure provides scores on six Given the inherent confounding of age, time, and cohort, the effects
262 WHITBOURNE. ZUSCHLAG, ELLIOT, AND WATERMAN

analyzed from this design are as follows: age-time of measurement for Table 1
longitudinal comparisons, age-cohort for comparisons involving Designs of Multivariate Analyses Comparing Ages and Cohorts
cross-sectional effects, and cohort-time of measurement for time-lag-
based comparisons. The two cohort-sequential main effects are age- Independent Dependent Ages
time and cohort-time. Compared with a longitudinal design, this de- Analysis variables variables compared
sign permits the relatively clear separation of developmental aging pro-
cesses from environmental influences for the age period of 20 to 31, as Longitudinal of Age-time Stages 1-6 20, 31, 42
longitudinal data are available from two cohorts tested at two time Cohort I,
1966-1988
periods. Any age-time effects observed for Cohort 1 between the ages
Longitudinal of Age-time Stages 1-8 31,42
of 31 and 42 must be regarded as potentially confounded with cohort- Cohort 1,
time of measurement influences. 1977-1988
For the purposes of clarity in presentation, the times of testing were Longitudinal of Age-time Stages 1-8 20,31
rounded off to 1966.1977, and 1988. The ages were rounded off to 20, Cohort 2,
31. and 42 years although, as is shown later, the actual mean ages of 1977-1988
those who participated in follow-ups varied somewhat around these Cross-sectional, Age-cohort Stages 1-8 20,31,42
ages. 1988
Cross-sectional, Age-cohort Stages 1-8 20,31
1 077
17//
Sample Cohort sequential Age-Time X Stages 1-6 20,31
Cohort-
The University of Rochester alumni and current students who com- Time
prised the sample were classified into three cohorts on the basis of Time lag, all Cohort-time Stages 1-6 20
when they were first tested as undergraduates. Cohort 1 included stu- cohorts
Time lag, Cohorts Cohort-time Stages 1-8 20
dents who were tested in the years 1966-68 (Constantinople's, 1969,
2 and 3
original sample), Cohort 2 was made up of students first tested in Time lag, Cohorts Cohort-time Stages 1-8 31
1976-77. and Cohort 3 consisted of students tested in 1988-89. At the 1 and 2
time of the present testing. Cohort l's mean age was 42.63, with a range
of 40-44, members of Cohort 2 had an average age of 31.88, with a
range of 29-34. and Cohort 3s average age was 20.37, with a range of
17-24.
Of the 155 respondents in Cohort 1 who were tested in 1977,30(19%) who were retested in 1988 according to attrition status is as follows. Of
could not be located. Of the remaining 125, 99 (79.2%) returned ques- the 99 members of Cohort 1 returning questionnaires in 1988,41 (41 %)
tionnaires. Cohort 2 suffered the largest attrition because of difficulty were from the Class of 1968 (sophomores at the time of original test-
obtaining current addresses (see Procedure section). Of the original ing), 27 (27%) from the Class of 1967, and 31 (31%) from the Class of
group of 298. addresses could not be found for 113 (38%). Of those 1966. The gender composition of the returning sample of Cohort 1 was
contacted. 83 (45%) completed the questionnaires. Cohort 3 consisted 62 men and 37 women. Among the 83 members of Cohort 2 tested in
of 292 current undergraduates: 45 male and 27 female freshmen, 60 1988,14(17%) were members of the Class of 1980 (freshmen in 1977),
male and 37 female sophomores. 44 male and 22 female juniors, and 30 22 (27%) from the Class of 1979, 27 (33%) from the Class of 1978, and
male and 27 female seniors. As the result of missing data because of 20 (24%) from the Class of 1977. The Cohort 2 returning sample was
incomplete questionnaires on several respondents, the numbers used composed of 43 men and 40 women. There was no difference in the
in analyses differed slightly from the numbers of respondents counted attrition pattern by college class within either of Cohorts 1 and 2. How-
as "returned" for the purpose of evaluating attrition effects. Analyses ever, there were differences in return status by gender. Among the
involving Cohort 1 in 1966 included only those respondents who were members of both cohorts, women were more likely to have been lost
followed in 1977(/;= 155) as those were the data on which the previous from the sample. This occurred because of the inability to obtain a
report (Whitbourne & Waterman, 1979) was based.' current address, x 2 (2, N = 155) = 17.67, p < .001 for Cohort 1, x 2 (2, N=
298) = 19.44, p < .001 for Cohort 2. In Cohort 2 only, men were also
The gender and original college class distribution of the two cohorts
more likely to have dropped from the sample by not responding to the
questionnaires sent to them.
No freshmen were included in the original sample of Cohort 1 re-
YEAR OF TESTING spondents whose data were available for follow-up purposes. To mini-
mize differences between Cohort 1 and the other cohorts because of
1966 1977 1988 this methodological artifact, respondents who were measured when
20 31 42 they were freshmen in college were excluded from all analyses reported
AGE
COHORT 1 in this article (14 from Cohort 2 and 73 from Cohort 3).
N- 347 N-155 N-99 Comparisons of the 1977 IPD scores of those who were followed in
1988 with those who were not (excluding freshmen) yielded neither a
20 31
COHORT 2
N-298 N-83 1
A complete time-lag analysis of cross-sectional differences within
the college years using all 300 of Constantinople's (1969) original sam-
20 ple tested in 1966 is in preparation. It should also be noted that the n of
COHORT3 155 reported here for Cohort 1 in 1966 and 1977 differs from the n of
N -292
147 used for the longitudinal analyses in Whitbourne and Waterman
(1979) as it was possible to recover previously unusable data from 8
Figure I. Design of the present study. subjects.
 PSYCHOSOCIAL DEVELOPMENT 263

multivariate main effect of dropout nor interactions between attrition they completed the questionnaires. After turning in their question-
status and the other independent variables of gender and cohort. The naires, Cohort 3 students were given the gift certificate and a short
smallest Wilks's lambda was for the cohort by dropout interaction with debriefing form explaining the goals of the study in which they had just
3.2% of the variance accounted for(0'= .968, p= .164). None of the participated. They were also informed that they might be contacted by
univariate tests of this interaction or any other interaction were signifi- the researchers at a future point after their graduation from college.
cant. The only significant univariate test with attrition status as an
independent variable was for the Industry scale, F(l, 370) = 6.53. p =
.011; the multivariate main effect of attrition status was i'= .975, p = Measures
.32. Respondents who dropped out of the study in 1988 tended to have
lower Industry scores in 1977 (M= 9.57) than respondents who partici- The IPD is a questionnaire measure based on Erikson's (1963)
pated in the 1988 testing (M = 12.78). theory, developed by Constantinople (1969) and extensively validated
As reported in Whitbourne and Waterman (1979), students at the in subsequent research (summarized in Waterman & Whitbourne.
University of Rochester tend to be raised in middle-class and upper- 1981). It has a total of 60 items and yields scores that indicate the extent
middle-class families. At the time of the present follow-up, and on the to which each of the first six Eriksonian psychosocial crises have been
basis of social class of the head of household, almost all members of successfully resolved. Items testing resolution of the last two stages
Cohort 1 were in Social Classes I (46.4%)or II (49.5%) on Hollingshead's were added from scales described in Walaskay et al. (1983-1984). The
(1957) two-factor index. Comparing the social class index of the fol- IPD includes five items representing the positive resolution and five
lowed members of Cohort 1 between 1977 and 1988 revealed that more items representing the negative resolution of each stage. Each item is
were in Social Class II in 1988 and fewer were in Social Classes I, III, rated on a 7-point scale ranging from 1 (definitely most uncharacteristic
and IV than would be expected based on 1977 data. x2(3, N = 292) = of you) to 7 (definitely most characteristic of you). Scores used in the
11.62. p < .001 (based on Social Classes I-IV). analyses presented in this article were based on a composite, or differ-
In 1988, the social class of head of household for the members of ence score, obtained by subtracting the summed score on the negative
Cohort 2 at follow-up was comparable to that of Cohort l's when they items from the summed score on the positive items. Thus, the range for
were 31 years of age, x2(3, N = 207) = 4.77, p > .05, with nearly half each scale is from -30 to 30. The total number of items was 80.
(48.5%) in Social Class I, another 27.3% in Social Class II, 18.2% in Respondents in Cohorts 1 and 2 were also asked to complete a bio-
Social Class III, and the remainder (6.1%) in Social Class IV The par- graphical data questionnaire, identical to the one administered to
ents of students in Cohort 3 were nearly evenly distributed among alumni in the previous follow-up. This questionnaire covered present
Social Classes I—III (I = 29.8%, II = 38.3%, and III = 23.85%). This and past educational, occupational, and family history. Cohort 3 re-
distribution represented a significant difference from the distribution spondents completed a demographic questionnaire identical to the one
of the parents of Cohort 2 members when they were in college. x2(3, administered to Cohort 2 when they were in college. This included
;V= 440) = 23.9, p < .001, as more parents of Cohort 3 respondents information on the students status in college (year and major), family
were in Social Class II compared with parents of Cohort 2 in 1977 background, and 10-year goals.
(21.2%), and fewer parents of Cohort 3 members were in Social Class I
than was true for Cohort 2's parents (45.95%).
Results
Procedure To facilitate the presentation of results, scores from the sepa-
In the early spring of 1988, we contacted the University of Rochester rate IPD scales will be referred to in terms of the psychosocial
alumni office for updated information on the members of the sample crisis stage they represent: Stage 1: trust versus mistrust; Stage
who had been tested in 1976-77. At that time, it was discovered that 2: autonomy versus shame and doubt; Stage 3: initiative versus
the university's computer records of student identification numbers guilt; Stage 4: industry versus inferiority; Stage 5: identity ver-
had been totally revised and as a result, graduates in the late 1970s sus identity diffusion; Stage 6: intimacy versus isolation; Stage
could not be identified by the code numbers available to the present 7: generativity versus stagnation: and Stage 8: ego integrity ver-
group of researchers. Because more complete name and address infor- sus despair.
mation was available to the study authors on Cohort 1, fewer respon-
The means and standard deviations of all eight IPD stage
dents from this group were lost at the time of this follow-up.
scores pooled for all respondents (except freshmen) available at
Using the available names and addresses provided by the alumni
each testing occasion are shown in Table 2. Before conducting
office, questionnaires with an explanatory cover letter were sent to
members of Cohorts 1 and 2 in October 1988. A consent form was also MANOVAs, we calculated correlations among the IPD scales.
included with this material, to be returned in a sealed envelope along The eight scales of the IPD were all significantly positively
with the questionnaire packet. In December 1988, follow-up letters correlated with each other for all ages. Average interscale corre-
and questionnaire packets were sent to those who had not returned the lation coefficients depended on heterogeneity of the sample,
materials up to that point. ranging from .468 for age 20 (Cohorts 1-3) to .569 for age 42
In the spring semester of 1988, an initial attempt was made to recruit (Cohort 1 only). Only three of the correlation coefficients calcu-
current undergraduates from the University of Rochester. However, lated were below .30: Stage 4 with Stage 2 at age 20 (r= A 38),
because the senior author was no longer a faculty member at the insti- Stage 7 with Stage 2 at age 20 (/ = .196). and Stage 6 with Stage 2
tution, it was difficult to obtain access to a sufficient number of under- at age 42 (r= .274). Four correlations were over. 70. all for age 42
graduates to meet the design requirements of the study for 300 stu-
and all with Stage 8: Stage 1 (r = .763), Stage 5 (r = .708), and
dents. It was decided to recruit respondents by offering a tangible in-
centive for their participation. Students were offered a $5 gift Stage 7 (r = .746). Such interscale correlations are expectable on
certificate to a local restaurant in return for completing the question- the basis of the theory of epigenesis. given that the successful
naires. This reward was considered comparable to the food provided to resolution of the crises of early stages is said to provide the
the Cohort 2 students, who, in 1977, had been offered snacks while foundation for the successful resolution of later crises. Simi-
 264 WHITBOURNE. ZUSCHLAG, ELLIOT, AND WATERMAN

Table 2 Table 4
Composite Expanded Inventory of Psychosocial Development Results of Multivariate Analyses Comparing Ages and Cohorts
Scale Scores bv Cohort and Year Tested
Wilks's
Cohort 1 Cohort 2 Cohort 3 Analysis n lambda P

Scale- 1966 1977 1988 1977 1988 1988 Longitudinal of Cohort 1,

1966-1988 98 .526 <.0005
Stage 1 Longitudinal of Cohort 1,
M 10.05 10.94 11.49 10.09 11.23 9.88 1977-1988 98 .483 <.0005
SD 8.36 8.30 8.13 7.38 7.04 8.20 Longitudinal of Cohort 2,
Stage 2 1977-1988 69 .518 <.0005
\f 8.28 8.16 8.95 7.54 7.77 7.20 Cross-sectional, 1988 383 .831 <.0005
SD 5.18 5.85 5.01 5.80 4.70 5.38 Cross-sectional, 1977 378 .900 <.0005
Stage 3 Cohort sequential
M 10.83 11.02 12.64 11.00 12.33 10.81 Effect of age-time 224 .672 <.0005
SI) 6.99 7.10 7.04 7.49 7.64 7.90 Effect of cohort-time 224 .936 .025
Stage 4 Effect of Age-Time X
.1/ 6.54 13.58 16.52 9.19 14.32 9.86 Cohort-Time 224 .932 .019
SD 8.23 7.78 7.01 8.81 7.41 9.12 Time lag, all cohorts, age 20 595 .955 .008
Stage 5 Time lag, Cohorts 2 and 3, age 20 439 .929 <.0005
.1/ 7.56 9.71 10.39 7.51 9.91 7.52 Time lag. Cohorts 1 and 2, age 31 224 .742 <.0005
SD 6.87 6.22 6.59 7.10 6.68 7.11
Stage 6
\1 11.17 13.12 13.33 11.98 14.25 11.77
SD 6.92 7.02 7.06 7.83 7.95 8.24
Stage 7 hort-Year X Age-Time interaction of the cohort-sequential anal-
.1/ — 8.90 9.58 7.30 8.77 7.00 ysis, the only significant interaction in any MANOVA. With the
SD — 5.68 5.91 5.71 5.53 5.77 exception of the time-lag analysis for 31 -year-olds, the strongest
Stage 8
M — 7.74 3.85 5.64 1.99 2.72 associations were between age and the IPD, whether the MAN-
SD — 7.83 8.12 7.46 7.47 8.45 OVA was conducted longitudinally, cross-sectionally, or sequen-
X
tially. The strongest effect was for the longitudinal MANOVA
155 155 99 223 69 219
of Cohort 1 from 1977 to 1988 (age 31-42), with 52% of the
Sole. The samples on which these data were calculated were those variance accounted for by the age-time variable.
used i n the cross-sectional analyses. Dashes indicate no data were avail- Main effects of gender were observed in the cross-sectional
able. MANOVA of 1988 data for all three cohorts, U = .920, p <
.0005; in both time-lag MANOVAs for college-aged respon-
dents, U = .961, p = .0009 for Stages 1-6 for all three cohorts,
larly. unsuccessful resolutions of early stage crises reduce the U = .906, p < .0005 for Stages 1-8 for Cohorts 2 and 3 at college
likelihood of future successful resolutions. age; and in the time-lag MANOVA for age 30 respondents, U =
Correlations across test intervals were also conducted to de- .926, p = .035. However, all these multivariate effects were
termine the extent of intraindividual change over the ages of weak, with less than 10% of the variance being accounted for in
20-42 and 31 -42 for Cohort 1 and the ages 20-31 for Cohorts 1 any case. All gender differences favored women.
and 2 combined. These correlations are presented in Table 3.

MANOVAs Univariate Analyses of Age and Cohort

As shown in Table 4, all MANOVAs comparing cohorts, The univariate inferential statistics for each multivariate ef-
years, or ages on the IPD were significant, including the Co- fect for each of the IPD scales are shown in Table 5. The means
for each IPD scale based on the longitudinal and sequential
comparisons are plotted in Figures 2-9. The results are summa-
Table 3 rized in the following order: (a) age-time effects based on the
Correlations Across Ages Tested for Cohorts I and 2 longitudinal analyses of Cohort 1 between the ages of 20 and
Cohort 1 Cohort 1 Cohort 1 Cohort 2 40, Cohort 1 between the ages of 30 and 40, and Cohort 2
Stage (20-42 years) (31-42 years) (20-31 years) (20-31 years) between the ages of 20 and 31, and the age-time effect in the
cohort-sequential analysis comparing Cohorts 1 and 2 between
Stage 1 .53 .66 .62 .57 the ages of 20 and 31; (b) age-cohort effects as determined by
Stage 2 .32 .34 .52 .26
Stage 3 .52 .44 .49
the two cross-sectional analyses conducted in 1977 and 1988; (c)
.40
Stage 4 .36 .48 .42 .54 cohort-time effects based on the two time-lag analyses on age
Stage 5 .49 .48 .46 .55 20 scores (Cohorts 1,2, and 3) and age 31 scores (Cohorts 2 and
Stage 6 .52 .66 .48 .67 3) and the cohort-time effect of the cohort-sequential analysis;
Stage 7 — .55 — .54 and (d) Age-Time X Cohort-Time interaction effects, based on
Stage 8 — .64 — .36
the cohort-sequential analysis. Only significant effects are re-
Xotc. Dashes indicate no data were available. ported in this summary.
 PSYCHOSOCIAL DEVELOPMENT 265

Table 5
Univariate Fs and Significance of Cohort and Age Comparisons
Stage

Analysis 1 2 3 4 5 6 7 8

Longitudinal of
Cohort 1, 1966-
1988
F(2, 192) .98 2.32 4.42 60.57 5.67 8.77 — —
p .38 .10 .01 <.001 .004 <.001 — —
Longitudinal of
Cohort 1. 1977-
1988
F(l,96) .47 4.15 5.20 11.40 1.24 1.53 .76 35.75
P .50 .04 .03 .001 .27 .22 .39 <.001
Longitudinal of
Cohort 2, 1977-
1988
F(\,61) 4.92 .18 .32 13.85 7.05 5.07 1.86 8.92
p .03 .68 .58 <.001 .01 .03 .18 .01
Cross-sectional, 1988
F(2. 377) 1.59 2.61 1.20 19.10 5.20 2.07 6.12 1.32
P .21 .08 .30 <.001 .01 .13 .01 .27
Cross-sectional. 1977
F( 1,375) 1.24 1.06 .01 26.37 10.35 2.54 7.89 7.19
P .27 .30 .92 <.001 .01 .11 .01 .01
Cohort-sequential
effect of age-
time
F( 1.220) 6.98 .28 .46 73.96 18.71 13.94 — —
P .01 .60 .50 <.001 <.001 <.001 — —
Effect of cohort-time
F( 1,220) .02 .16 1.78 8.06 .15 2.00 — —
p .90 .69 .18 .01 .70 .16 — —
Effect of Age-Time
X Cohort-Time
F( 1.220) .82 .07 .11 10.25 .02 .04 — —
P .37 .79 .74 .01 .90 .85 — —
Time lag, all cohorts
F[2, 589) .01 1.43 .08 9.52 .18 .88 — —
P .99 .24 .93 <.001 .84 .42 — —
Time lag. Cohorts 2
and 3
F( 1,435) .01 .05 11 3.59 .51 .44 .04 10.39
P .94 .82 .64 .06 .48 .51 .84 .01
Time lag. Cohorts 1
and 2
F( 1,220) .08 .23 1.66 .55 .08 1.36 .01 26.25
P .77 .64 .20 .46 .78 .24 .91 <.001

Note. Dashes indicate no data were available.

Stage 1 Age-Time Effects this increase was preceded by a smaller, nonsignificant de-
crease, a pattern also seen in Cohort 2's scores.
The significant effects on this IPD scale involved the longitu-
dinal analysis of Cohort 2 between the ages of 20 and 31 and a
significant effect of age-time in the cohort-sequential analysis.
As can be seen from Figure 2, scores on Stage 1 increased Stage 3 Age-Time Effects
slightly between these two age-times. Cohort 1 respondents showed a significant increase on Stage
3 scores between the ages of 20 and 42 as revealed in the 1966 to
Stage 2 Age-Time Effects 1988 longitudinal analysis. A comparison of the cells involved
in this analysis show that scores at age 20 (M = 10.60) were not
Scores on Stage 2 increased significantly between the ages of significantly different from age 31 (M = 10.91), /(97) = .42, but
31 and 42 for Cohort 1, as indicated by the significant longitu- scores at age 42 (M = 12.64) were significantly higher than the
dinal effect for this measure. Inspection of Figure 3 reveals that scores of either age 20. ;(97) = 2.99, p = .004, or 31, t(97) = 2.32.
266 WHITBOURNE, ZUSCHLAG, ELLIOT. AND WATERMAN

Cohorts Cohorts
— - Cohorti — Cohorti
21 21n
1 — • Cohort2 — Cohort2

1 • Cohort3 19 • Cohorts

"1 17

"I
13
15-

1
11
1
91
7]
54
3
1
1966 1977 1986 1986 1977 1988
Time of Testing TkneolTerilng

Figure 2. Means of Inventory of Psychosocial Development Stage 1 Figure 4. Means of Inventory of Psychosocial Development Stage 3
by year tested and cohort. Age at time of testing indicated on figure. by year tested and cohort. Age at time of testing indicated on figure.

/) •= .023. These findings are consistent with the significant Age-cohort effects. In the cross-sectional analysis of the
longitudinal effect shown for Cohort 1 between 1977 and 1988. 1977 data. Cohort l's (age 31) mean Stage 4 score was signifi-
cantly higher than Cohort 2's (age 20). Planned comparisons for
Stage 4 the cross-sectional analysis for the 1988 data revealed that the
difference between Cohorts I and 2 (Cohort 1 M = 16.52, Co-
Age-lime effects. All analyses with age-time as independent hort 2 M = 14.32) was not significant, F(l, 377) = 3.10, p = .079.
variables produced significant effects on Stage 4 scores. As re- The scores of Cohort 1 and 2 combined were significantly
vealed by planned comparisons in the longitudinal analysis of higher than Cohort 3's (M = 9.86), F(l, 377) = 43.36, p < 0001.
Cohort 1, Stage 4 scores increased significantly from 1966 (age Cohort-lime effects. A significant cohort-time effect was ob-
20. M = 7.53) to 1977 (age 31. M= 13.96, /[97] = 7.53. p < .001), served in the time-lag analyses of all cohorts at college age and
and again from 1977 (age 31) to 1988 (age 42. M = 16.52, /[97] = in the cohort-sequential analysis. It appears that these effects
3.30. /> = .001). The cohort-sequential analysis also produced a are accounted for by the relative depression of Cohort l's Stage 4
significant age-time effect with the means in the same direc- scores when they were of college age. In contrasts of the cohorts
tion for the years between ages 20 and 31. in the time-lag analysis of all three cohorts in college, Cohort 2

Cohorts Cohorts
— Cohorti — Cohorti
21- 21
-— Cohort2 — Cohort2
19- 19
• Cohorts -— Cohorts
17- 17
15- 15
13- 13
11 - 11
9- 20 20 9
^ - *
31
7- 31 "20 7-
5- 5
3 3
1 - 1
1966 1977 1988 1966 1977 1988
Time ot Testing Time of Testing

l-'ifiiirc 3. Means of Inventory of Psychosocial Development Stage 2 Figure 5. Means of Inventory of Psychosocial Development Stage 4
by year tested and cohort. Age at time of testing indicated on figure. by year tested and cohort. Age at time of testing indicated on figure.
 PSYCHOSOCIAL DEVELOPMENT 267
Cohort* Cohorts

— Cohorti - — Cohorti
21 — Cohort 2 — Cohort 2

19 —•- Cohort3
19 • Cohorts

17- 17-

15 15

13- 13

11 11

9 9

7 7

S 5

3 3

1 1
1966 1977 1988 1966 1977 1988
Time of Testing Tlmo of Testing

Figure 6. Means of Inventory of Psychosocial Development Stage 5 Figure 8. Means of Inventory of Psychosocial Development Stage 7
by year tested and cohort. Age at time of testing indicated on figure. by year tested and cohort. Age at time of testing indicated on figure.

(M = 9.19) was not significantly different from Cohort 3 (M = 4 scores of Cohort 1 at college age (M= 6.54) were significantly
9.86), F(l, 589) = .66, p = .42, but Cohort l's scores (M = 6.54) lower than the Stage 4 scores of Cohort 2 at college age (M =
were significantly lower than those of Cohorts 2 and 3, F(l, 11.12), t(222)= 3.96. p<. 001.
589)= 13.85, p < .001.
Age-Time X Cohort-Time effects. The cohort-sequential Stage 5
analysis revealed a significant Age-Time X Cohort-Time inter-
action for Stage 4 in addition to the significant main effects of Age-lime effects. There was a significant age-time effect
age-time and cohort-time. Contrasts of the cells in this design observed in Stage 5 scores in the longitudinal analysis of vari-
revealed that the Stage 4 scores of both Cohorts 1 and 2 in- ance (ANOVA) of Cohort 1 between the years of 1966 and 1988,
creased significantly from college age to 31 years old, /(154) = an effect accounted for by the significant increase between the
10.14, p < .001, and ?(68) = 3.72, p < .001, respectively. In these agesof20and31 (Mfor20yearsold = 8.03,Mfor31 yearsold =
contrasts, the Stage 4 scores for Cohort 1 at age 31 (M = 13.58) 9.67), 7(98) = 2.23, p = .028. The increase from 1977 (31 years
were not significantly different from the Stage 4 scores of Co- old, M= 9.67) to 1988 (42 years old, M= 10.39) for Cohort 1 is
hort 2 at age 31 (M = 14.32), 7(222) = .67, p = .507, but the Stage not significant in either the longitudinal ANOVA of Cohort 1

Cohorts Cohorts

— Cohorti - - Cohorti
21 21
- — Cohort2 — Cohort2
19 • Cohorts 19 - — Cohort3

17 17-
15 15-
13- 13-

11 11

9 9

7 7-

5 5
3 3
1 1
1966 1977 1988 1966 1977 1988
Time of Testing Time of Testing

Figure 7. Means of Inventory of Psychosocial Development Stage 6 Figure 9. Means of Inventory of Psychosocial Development Stage 8
by year tested and cohort. Age at time of testing indicated on figure. by year tested and cohort. Age at time of testing indicated on figure.
268 WHITBOURNE, ZUSCHLAG, ELLIOT, AND WATERMAN

from 1977 to 1988, or in contrasts of the 1966 to 1988 longitu- observed in both the time-lag analysis of college-age respon-
dinal ANOVA, /(98) = 1.06, p = .29. The longitudinal analysisof dents and the time-lag analysis of 31 -year-old respondents, with
Cohort 2 between the ages of 20 and 31 also produced a signifi- earlier cohorts on both occasions having higher scores than
cant effect of age-time. These findings from Cohorts 1 and 2 later cohorts.
analyzed independently were replicated in the cohort-sequen-
tial analysis, which yielded a significant age-time effect be-
tween the ages of 20 and 31 years. Summary of Significant Effects Across Stages
Age-cohort effects. The higher scores of 31-year-olds com- Age-time effects. Varying patterns of significant age-time
pared with 20-year-olds observed in the longitudinal analyses effects were observed in the sequential longitudinal compo-
also appeared in the significant contrasts of Cohort 2 (31 years nents of the present research. Significant age-time effects were
old. M =9.91) with Cohort 3 (20 years old, M= 7.52), F(l, observed for the longitudinal analysis of Stage 1 scores for Co-
377) = 6.14, p<. 001, in the cross-sectional analysis of the 1988 hort 2 between the ages of 20 and 31, with a slight increase in
data. There was no significant difference in Stage 5 scores be- scores between these measurement points. Scores on Stages 2
tween Cohort 2 in 1988 (31 years old, M =9.91) and Cohort 1 in and 3 showed a significant increase between ages 31 and 42 for
1988 (42 years old, M= 10.39), F(l, 377) = .20, p= .66. Cohort 1. Stage 4 scores, which increased between the ages of 20
and 31 for Cohort 1, also increased up to the age of 42 for this
Stage 6 Age-Time Effects cohort. An increase in Stage 5 scores for both Cohorts 1 and 2
occurred between the ages of 20 and 31, as did an increase in
There was a significant age-time effect in the longitudinal Stage 6 scores for Cohort 1 only. No age-time effects were ob-
analysis of Cohort I's Stage 6 scores between 1966 and 1988 served on Stage 7 scores. There was a drop in Stage 8 scores for
because of the significant difference between the scores of Co- Cohorts 1 and 2, corresponding to decreases between the ages
hort 1 in 1966 (age 20, M=\\ .00) compared with their scores in of 20-31 and 31-42 years.
1977 (age 31, M = 12.80), ?(98) = 2.56, p = .012. This difference Age-cohort effects. Age-cohort effects, reflecting cross-sec-
was replicated in the significant main effect of age-time in the tional differences, were found for Stage 4 scores in the 1977 and
cohort-sequential analysis, which indicated that Stage 6 scores 1988 data, with college students at both times receiving the
increased for both Cohorts 1 and 2 as they aged from 20 to 31 lowest scores. Cross-sectional differences were also found in
years. Cohort Is Stage 6 scores in 1988 (at age 42, M = 13.33) 1988 between 20- and 31-year-olds in Stage 5 scores, with the
were not significantly different from 1977 scores (age 31, M = younger age-cohort group receiving lower scores. College stu-
12.80). ?(98)=.91./?=.37. dents also received lower scores on Stage 7, both in 1977 and
1988. A similar age-cohort effect was observed on Stage 8 in
Stage 7 1977 alone.
Cohort-time and Age-Time X Cohort-Time effects. Evidence
Age-time effects. Unfortunately, no longitudinal data for for social-historical effects was obtained from the cohort-time
ages 20-42 of one cohort are available for this scale, as it was element of the present sequential design for Stages 4 and 8. On
introduced in the 1977 testing when Cohort 1 was 31 years old. Stage 4 scores, Cohort 1 received much lower scores in college
The longitudinal analyses available fail to reveal significant in 1966 than did younger cohorts at college age in later years.
change in Stage 7 over time, although for both Cohort 1 and Higher scores on Stage 8 were obtained in 1977 than in 1988
Cohort 2, average Stage 7 scores did tend to increase between both for respondents in college and at age 31 years.
the ages of 20 and 31 as shown in Figure 8.
Age-cohort effects. As indicated in the 1977 cross-sectional
analysis, the average Stage 7 score of Cohort 1 (age 31 M= 8.90) Discussion
was significantly higher than the average of Cohort 2 (age 20
Overall Analysis
M - 7.30). In 1988, Cohort I's scores at age 42 (M =9.58) were
not significantly different from Cohort 2's scores at age 31 (age The analyses reported in this study involving scores on re-
31 M = 8.77). Cohort 3's scores (M = 7.00) were significantly peated testings with the expanded IPD provided an unequivo-
lower than the combined scores of Cohorts 1 and 2, F(l, 377) = cal affirmative response to the first question posed by the pres-
13.41, / x . 0 0 0 1 . ent study. Consistent patterns of personality change were evi-
dent on a measure theoretically expected to be sensitive to
developmental changes in adulthood. The results of this study
Stage 8 therefore join those of other research (Haan, Millsap, &
Age-time effects. From 1977 to 1988, Stage 8 averages for Hartka, 1986; Helson & Moane, 1987; Stevens & Truss, 1985) in
both Cohorts 1 and 2 show sharp and significant declines, as a growing body of evidence indicating the existence of adult
indicated by the longitudinal analyses. personality changes on a variety of theoretically derived vari-
Age-cohort effects. In the cross-sectional comparison of ables even during the supposedly placid decade of the 30s. The
1977 data. Cohort 1 (at age 31) had a significantly higher Stage 8 extent of intraindividual change observed in this study ex-
average than Cohort 2 (at age 20). This difference was not main- ceeded patterns of stability reported in longitudinal research
tained in the 1988 cross-sectional analysis. involving the use of trait-based personality measures on adults
Cohort-time effects. Significant cohort-time effects were over the age of 30 years (McCrae & Costa, 1990). As reported in
 PSYCHOSOCIAL DEVELOPMENT 269

Table 3, although correlations as high as .67 were observed in was significant) for Cohort 1 between the ages of 31 and 42.
the present study, the majority of values ranged in the .3-5 Such a pattern may reflect a movement toward a greater sense
range for the 31-42 year age interval. of personal control during the 30s, as individuals are able to
The second question addressed in the present research con- move to positions of greater power and authority than they held
cerns the degree to which adult personality changes reflect de- in their 20s. Such movement may have been augmented by the
velopmental processes that are independent of historical and economic growth occurring in the 1980s. Regarding initiative
cultural influences. Answers to this question can be offered versus guilt, the gains shown by Cohort 1 between the ages of 31
only for the age period of 20-31, and although cultural factors and 42 may reflect increased sexual awareness and a more re-
cannot be ruled out, the replication of changes within two co- laxed attitude toward sexual expression in the context of long-
horts provides strong evidence for the effects of development. term intimate relationships, feelings that had not yet developed
Whether this will hold true for the patterns of increases shown by the age of 31 (several items on this scale tap sexual openness
in the 11-year period following age 31 is a question that only and awareness). The quality of initiative also reflects the ability
further replication can address. to "play" without hindrance or inhibition and the members of
Within the 20-31 age span, there were clear age-related devel- Cohort 1 may rediscover this quality through their interactions
opmental trends for scores on the stages of trust versus mistrust with growing children.
(Stage 1), identity versus identity diffusion (Stage 5), and inti- All of the analyses described thus far reflect a pattern of
macy versus isolation (Stage 6). Both Cohorts 1 and 2 received increasing psychosocial resolution with age. A striking excep-
increasingly favorable scores on these stage scales, even though tion is the pattern shown on Stage 8 scores, ego integrity versus
they were tested over nonoverlapping time periods. The trends despair. Both cohorts tested over the 1977-88 period showed a
observed between the ages of 31 and 42 years toward increasing precipitous decline on these scores. One interpretation of this
scores on autonomy versus shame and doubt (Stage 2), increas- decline incorporates the findings regarding the industry versus
ing scores on initiative versus guilt (Stage 3), and decreasing inferiority scale (Stage 4), the other stage to reflect large effects
scores on ego integrity versus despair (Stage 8) need further of aging and also, to a lesser extent, cohort. Items on the ego
replication. integrity versus despair scale relate to the constructs of whole-
The findings for industry versus inferiority (Stage 4), al- ness, honesty, and meaning in life and to having a sense of
though suggestive of aging effects for the full age range of the connection with humanity and the welfare of others (Walaskay
study, also appear to reflect complex interactions with environ- et al., 1983-1984). The general decline over time for both Co-
mental influences, as was the case in the previous study (Whit- horts 1 and 2, coupled with the low scores of Cohort 3, may be
bourne & Waterman, 1979). In particular, Cohort l's very low symptomatic of the rise of materialism in the 1980s, a material-
Stage 4 scores in 1966 stand as an anomaly from the general ism that has led to reduced social welfare programs and an
patterns shown for college-aged subjects and may reflect spe- emphasis on yuppies and their corresponding notoriously
cific environmental influences operating on this cohort at the empty life-style focused on wealth and possessions. Cohort 1 in
time of its testing in 1966. At that time, the student body of the particular is experiencing a dramatic increase in the sense of
University of Rochester may have reflected the shifting trends industry along with its associated focus on work and material
in the 1960s toward disenchantment with the work ethic of the success and thus may be suffering in terms of its resolution of
1950s and into the period of protest that characterized the late issues revolving around ego integrity. Such a process is reminis-
1960s and early 1970s. Once out of college, Cohort 1 experi- cent of the "midlife crisis," but closer inspection reveals this not
enced the need to achieve in the work world, a pressure that to be a viable explanation. Looking at the data from all three
may have stimulated them to "catch up" by age 31 to reach the cohorts, the low Stage 8 scores may be seen as reflecting a more
same level in their Stage 4 scores as did Cohort 2 by that age. general societywide crisis of morality and purpose affecting
One important consideration that must be examined in the adults of all ages. Indeed, these data show a striking correspon-
present data concerns the results of the attrition analysis indi- dence to the results of nationwide surveys of college freshmen
cating slightly lower industry versus inferiority scores for those between the years of 1970 and 1988 showing a drop of from 76%
members of the sample who dropped out of the study between to 51% in the life objective of "developing a meaningful philo-
1977 and 1988. However, the impact of this finding on the sophy of life" (U.S. House of Representatives, 1989). It is likely
overall interpretations made from the present study seems to be that the consistent drop in ego integrity scores for the cohorts
minimal. There was no significant multivariate effect of attri- tested in the present study reflects a similar erosion of philo-
tion, and the effect for industry versus inferiority was the only sophical values.
univariate effect found out of many tests for a main effect of
attrition or interaction involving attrition. Therefore, this effect
Erikson's Theory as Applied to Adulthood:
of attrition on this scale may very well be a Type I error.
Support and Clarification
Any explanations that can be offered for the pattern of find-
ings for scale scores of autonomy versus shame and doubt, initia- The present investigation represents the first large-scale
tive versus guilt, and generativity versus stagnation would in- study of men and women tested over the 20 years of early to
volve considerable speculation, given that none of the effects in middle adulthood with a quantitative measure based on Erik-
these analyses were replicated across cohorts tested at the same son's (1963) theory. The sequential design of the study made it
age. Inspection of Figure 3 suggests a slight decrease for both further possible to test Erikson's proposition that psychosocial
Cohorts 1 and 2 during their 20s, followed by an increase (which development proceeds in an orderly sequence of stages, in-
270 WHITBOURNE, ZUSCHLAG, ELLIOT, AND WATERMAN

fluenced by a combination of inner psychological processes logical, or social trajectories. The issues identified with any
and external social and cultural forces. The findings that prior (and possibly future) stage of development may take prior-
emerged from the analyses reported here provided considerable ity over the stage concerns nominally identified with the indi-
support for Erikson's theory, as well as clarification regarding vidual's current developmental stage. If this alternative is oper-
the timing of the psychosocial crises throughout the early to ating, then proportionately greater changes could occur for
middle adult years. scales tapping prior or future stage components than those
Support for Erikson's (1963) psychosocial theory emerged changes found for the current stage scale. Thus, as observed in
from several sources. First, as in the previous study based on the present study, the most dramatic changes observed were
the 1976 follow-up, it was clear that changes over time in the gains in industry versus inferiority between the ages of 20 and
stage scores reflected relatively pure "aging" effects (i.e., inner 42, whereas there was a dramatic age/time decrease for integrity
psychological changes) as well as the effects of exposure to a versus despair across cohorts. Both changes are out of the ex-
particular historical, social, and cultural reality of the external pected time frame, and both seem most plausibly interpreted in
environment. Thus, in addition to replicated effects of age- terms of the impact of cultural and historical events.
time across cohorts, several stage scores reflected the influence Although it is interesting to develop a speculative account for
of age-cohort and cohort-time, both factors regarded as sensi- the relative strengths of changes occurring with respect to
tive to environmental influences. Second, Erikson's theory was various stage components, it should be recognized that the avail-
supported in terms of the proposition that favorable resolution able instrumentation does not allow for a high degree of preci-
of one psychosocial crisis stage is dependent on successful reso- sion in this regard. The various stage scales of the IPD, although
lution of previous stages and, in turn, influences the resolution possessing a common possible range, are, like personality
of subsequent stages. The intercorrelations among the Erikson- scales in general, ordinal in nature. Mean change scores of dif-
ian stage scores were all high and consistently positive in direc- fering magnitudes for various IPD scales do not necessarily
tion. A third basis for support of Erikson's theory concerned warrant the conclusion that different developmental gains have
the timing of the two psychosocial stages studied over the age been achieved.
ranges during which they are theorized to attain ascendancy, The interpretation of a developmental trajectory that de-
namely intimacy versus isolation and generativity versus stag- viates from the much-publicized timetable of regular progress
nation. With regard to generativity, further testing will indicate from Stages 1-8 further emphasizes the point so clearly made
whether this trend represents a cohort effect or a developmental by Erikson (1963); namely, that the psychosocial stages not be
change that will continue through this stage's proposed time of regarded as prescriptions for development. Added by our find-
further evolution into the decade of the 50s. ings is the possibility of movement through the psychosocial
Clarification of Erikson's (1963) principle of movement issues in a way that departs in a specified manner from the
through his proposed developmental matrix of age period by diagonal progress through the epigenetic matrix.
psychosocial stage also emerged from the present findings. In
addition to evidence for continued development on stages theo- References
rized to be of maximum ascendancy, there appeared to be
Baltes, P. B. (1968). Longitudinal and cross-sectional sequences in the
movement on the scales tapping the early stages of develop- study of age and generation effects. Human Development, 11, 145—
ment well into the years of middle adulthood, and particularly 171.
between the period of 20-31 years when data from two cohorts Conley, J. J. (1985). Longitudinal stability of personality traits: A mul-
were available. One possible explanation of this finding is that titrait-multimethod-multioccasion analysis. Journal of Personality
as the individual begins to confront the issues of a new stage, a and Social Psychology, 49, 1266-1282.
process of reorganization begins that stimulates the individual Constantinople, A. (1969). An Eriksonian measure of personality de-
to address past and future psychosocial concerns. If this expla- velopment in college students. Developmental Psychology, 1, 357-
nation were correct, age-related changes ought to be found for 372.
every scale at every age. However, given that there is a central Costa, P., & McCrae, R. (1982). An approach to the attribution of
stage-specific issue for each stage, the greatest developmental aging, period, and cohort effects. Psychological Bulletin, 92, 238-
250.
movement should be found for the component then undergoing Elder, G. H. (1974). Children ofthe Great Depression: Social change and
its time of special ascendancy. The present findings do not sup- life experiences. Chicago: University of Chicago Press.
port this explanation, however. Although evidence was found Erikson, E. H. (1963). Childhood and society (2nd ed.). New York: Nor-
for developmental progressions during the stages of intimacy ton.
versus isolation and generativity versus stagnation, the patterns Erikson, E. H. (1980). Identity and the life cycle. New York: Norton.
of change for these stages were not as strong as for stages that (Original work published 1959)
were "off-diagonal" in their proposed sequence. Thus, a further Finn, S. E. (1986). Stability of personality self-ratings over 30 years:
explanation was sought. Evidence for an age/cohort interaction. Journal of Personality and
Social Psychology, 50, 813-818.
The most likely alternative explanation is the proposition
Haan, N., Millsap, R., & Hartka, E. (1986). As time goes by: Change
that the sequencing of Erikson's (1963) stages is not unidirec- and stability in personality over 50 years. Psychology and Aging, 1,
tional and that there is not an epigenetic unfolding of develop- 220-232.
mental issues. Rather, all psychosocial issues can reach ascen- Helson, R., & Moane, G. (1987). Personality change in women from
dancy at any particular time in the individual's life, depending college to midlife. Journal of Personality and Social Psychology, 53,
on unique factors specific to that individual's biological, psycho- 176-186.
 PSYCHOSOCIAL DEVELOPMENT 271

Hollingshead, A. B. (1957). Two factor index of social position. Unpub- US. House of Representatives. (1989). U.S. Children and their families:
lished manuscript, Yale University, New Haven, CT. Current conditions and recent trends: 1989. Washington, DC: US.
Levinson, D. J., Darrow, C. N., Klein, E. B., Levinson, M. H., & McKee, Government Printing Office.
B. (1978). The seasons of a man's life. New York: Alfred A. Knopf. Walaskay, M., Whitbourne. S. K.. & Nehrke. M. F. (1983-1984). Con-
McCrae, R. R., & Costa, P. T, Jr. (1990). Personality in adulthood. New struction and validation of an ego-integrity status interview. Interna-
York: Guilford. tional Journal of Aging and Human Development, IS, 61-72.
Schaie, K. W (1965). A general model for the study of developmental Waterman. A. S.. & Whitbourne, S. K. (1981). The inventory of psycho-
change. Psychological Bulletin, 64, 92-107. social development. JSAS: Catalog ofSelected Documents in Psychol-
Schaie, K. W (1988). Internal validity threats in studies of adult cogni- ogy 11, 5. (Ms. No. 2179)
tive development. In M. L. Howe & C. J. Brainerd (Eds.), Cognitive Whitbourne, S. K. (1986). Adult development (2nd ed.). New York:
development in adulthood'(pp. 241-272). New York: Springer-Verlag. Praeger.
Schaie, K. W, & Parham, 1. A. (1976). Stability of adult personality Whitbourne, S. K.. & Waterman. A. S. (1979). Psychosocial develop-
traits: Fact or fable? Journal of Personality and Social Psychology, 34, ment during the adult years: Age and cohort comparisons. Develop-
146-158. mental Psychology, 15, 373-378.
Stevens, D. P., & Truss, C. V (1985). Stability and change in adult per-
sonality over 12 and 20 years. Developmental Psychology, 21, 568- Received October 23. 1991
584. Accepted February 15, 1992 •

AMERICAN PSYCHOLOGICAL ASSOCIATION

SUBSCRIPTION CLAIMS INFORMATION Today's Date:

We provide this form to assist members, institutions, and nonmember individuals with any subscription problems. With the
appropriate information we can begin aresolution. If you use the services of an agent, please do NOT duplicate claims through
them and directly to us. PLEASE PRINT CLEARLY AND IN INK IF POSSIBLE.

PRINT FULL NAME OR KEY NAME OF INSTITUTION MEMBER OR CUSTOMER NUMBER (MAYBE FOUND ON AN Y PAST ISSUE LABEL)

DATE YOUR ORDER WAS MAILED (OR PHONED):

ADDRESS
P.O. NUMBER:

PREPAID CHECK CHARGE

CITY STATE/COUNTRY CHECK/CARD CLEARED DATE:_

(If possible, send a copy, front and back, of your cancelled check to help us In our
YOUR NAME AND PHONE NUMBER research of your claim.)
ISSUES: MISSING DAMAGED

TITLE VOLUME OR YEAR NUMBER OR MONTH

Thank you. Once a claim is received and resolved, delivery of replacement issues routinely takes 4-6 weeks.

DATE RECEIVED: DATE OF ACTION:

ACTION TAKEN: INV. NO. & DATE:
STAFF NAME: LABEL NO. & DATE:

SEND THIS FORM TO: APA Subscription Claims, 750 First Street, N.E., Washington, DC 20002

PLEASE D O N O T R E M O V E . A PHOTOCOPY M A Y BE USED.