An Eriksonian Measure of Personality Development in
College Students: A Reexamination of Constantinople's Data, and a Partial Replication Susan Krauss Whitbourne and Beth M. Jelsma Graduate School of Education and Human Development, University of Rochester Alan S. Waterman Trenton State College
Constantinople's finding that males showed a clearer pattern of increasing psy-
chosocial maturity during the college years than did females was challenged in a reexamination of the original data and a partial replication. It appears that Constantinople's findings regarding sex differences were attributable to the op- eration of cohort effects. The data in the original study indicated that scores for males and females varied by college class cohort in both cross-sectional and longitudinal analyses. The longitudinal data from the partial replication revealed differential patterns of change over time for men and women within two college class cohorts. The sensitivity of the Inventory of Psychosocial Development to cohort effects was discussed in the context of Erikson's theory.
Constantinople (1969) provided the first volved retesting a portion of Constantino-
research evidence of ontogenetic changes ple's sample (Whitbourne & Waterman, during the college years consistent with Er- 1979), we reexamined the original pattern ikson's (1963) epigenetic theory of person- of results. From this reexamination emerged ality development. Male and female under- the possibility that the reported sex differ- graduates were assessed on the personality ences were severely distorted by the cohort components associated with the first six effects operating within the sample. Because stages of Erikson's theory as measured by of the emphasis Constantinople placed on the Inventory of Psychosocial Development her findings for Stage 5 scores, the discussion (IPD), a paper-and-pencil instrument (see here will focus on the results for that stage. Waterman & Whitbourne, 1981, for a sum- A partial replication of her design was also mary of validity and reliability data on the undertaken to assess the reliability of the IPD). Changes were reported in the direc- findings of ontogenetic change, sex differ- tion of more mature functioning for five of ences in development, and cohort effects. the six stages, with the clearest findings oc- curring for the Stage 5 component of Iden- A Reexamination of Constantinople's tity versus Identity Diffusion. Constantino- Findings ple (1969) concluded that "males showed a clearer pattern of increasing maturity over In cross-sectional analyses of data col- the 4 years than did females" (p. 357) and lected in 1966, Constantinople (1969) found suggested that for males, "the college envi- a significant college year (freshman through ronment is more conducive to growth" (p. senior year) effect for Identity and a College 368). Constantinople also noted some sig- Year X Sex interaction for Identity Diffu- nificant differences between class cohorts. sion. Males had lower scores than females In connection with a study of psychosocial as freshmen; by senior year, these differences development during the adult years that in- were minimal. From these data Constanti- nople (1969) observed that "the males showed the expected developmental trend Requests for reprints should be sent to Susan K. Whitbourne, Graduate School of Education and Human while the females did not" (p. 361). Development, University of Rochester, Rochester, New However, when the data from the three York 14627. times of testing (1965, 1966, 1967) are ar-
369 370 S. WHITBOURNE, B. JELSMA, AND A. WATERMAN
ranged in a cross-sectional sequences design strengthens the reservations expressed con-
to highlight sex comparisons within each cerning the cross-sectional results. In view college cohort (classes of 1965 through of these differences, Constantinople's deci- 1968), a new perspective is generated on sion to combine the cohorts in graphically Constantinople's observations for the stage presenting the longitudinal data (in her Fig- of Identity versus Identity Diffusion. Males ures 1 through 4) is questionable. Further, in the class of 1965 had slightly higher Iden- only one of four analyses suggested differ- tity scale scores and somewhat lower Iden- ential patterns of development on the Iden- tity Diffusion scores than did females in the tity and Identity Diffusion scales for males senior year (the only college year for which and females, and in this analysis, sex inter- data were collected on this cohort). For the acted with college class cohort. class of 1966, males and females received almost equal Identity and Identity Diffusion A Partial Replication scores for both junior and senior years. For the class of 1967, males had more positive The results of a partial replication of the Identity but equal Identity Diffusion scores longitudinal component of Constantinople's as females for the 3 years they were tested (1969) study illustrate dramatically the lim- (i.e., sophomore, junior, and senior years). ited generality of her claims regarding sex Females in the class of 1968 had lower Iden- differences in personality development dur- tity Diffusion scores than males during ing the college years. From the original freshman year and slightly more favorable Whitbourne and Waterman (1979) sample Identity Diffusion scores in the two subse- of 147 University of Rochester freshmen quent years. Since this class was the only (class of 1980) and sophomores (class of college cohort to provide a freshman year 1979) tested in 1977, 62 participated in a sample, it was not possible to establish the 1979 retest (27 as juniors and 39 as seniors). generality of the Identity Diffusion scale sex Attrition effects over the 2-year period were difference beyond that class. However, Con- comparable to those observed by Constan- stantinople's interpretation of apparent sex tinople. differences in development does not include Analyses of positive IPD scale scores by the necessary qualifications arising from the college class cohort and time of testing1 re- college cohort and college year confound in vealed differential changes over time by col- the data. Variations by cohort in sex differ- lege class cohort, sex, and IPD scale. Neg- ences across sophomore, junior, and senior ative IPD scales varied over time of testing years do appear to exist and should be cited by college class cohort. Combining the re- as examples of cohort and/or time-of-testing sults of positive and negative scale score effects. analyses, longitudinal gains were shown for In longitudinal analyses of 2-year and 3- men in the 1980 college class cohort and for year follow-ups of the individuals tested in women in the 1979 college class cohort. 1965, Constantinople reported significant These results were viewed as demonstrating time-of-testing effects for Identity in both developmental changes within each cohort analyses and for Identity Diffusion in the 3- that appear to be due to the operation of year follow-up. Effects for college class co- ontogenetic effects. However, the varying hort were found for both Stage 5 scales in nature of sex differences in these changes by the 2-year follow-up. Constantinople (1969) college class cohort prohibits any inferences concluded that "the longitudinal data gave concerning developmental patterns for men unqualified support to the cross-sectional and women. findings only with reference to changes in Discussion Identity and Identity Diffusion" (p. 364), but she did not make it clear whether she Constantinople's (1969) conclusion re- intended to include sex differences in this garding the existence of sex differences in summary statement. With respect to these longitudinal data, ' A complete set of tables may be obtained from the the finding of college class cohort effects first author. PERSONALITY DEVELOPMENT IN COLLEGE STUDENTS 371
patterns of ontogenetic development does such as a special freshmen orientation pro-
not appear supportable either from her own gram, changes in student advisement poli- data or from the partial replication. Rather, cies, or changes in the physical plant af- the effects observed can be traced to differ- fecting student housing, socialization, and ences among the cohorts studied. study opportunities. At the present time, it The apparent sensitivity of the IPD to co- is not possible to determine which particular hort effects as demonstrated here should not factors were responsible for the observed necessarily be interpreted as a weakness of cohort differences in the present study. Nev- this particular instrument. Erikson's theory ertheless, their existence is a clear indication incorporates the role of social, historical, and of the need to take long-term situational cultural factors into the development of the variables into account in interpreting the ego at each major crisis stage. It follows that IPD scores of college students. to be a valid index of the individual's de- velopment at a particular point in time, the References IPD must reliably and validly distinguish Constantinople, A. An Eriksonian measure of person- among respondents who differ on sensitivity ality development in college students. Developmental to external influences as well as on inner Erikson,Psychology, 1969, 1, 357-372. E. H. Childhood and society (2nd ed.) New psychological dimensions. The overall ten- York: Norton, 1963. dency of one college class to score higher Waterman, A. S., & Whitbourne, S. K. The Inventory than another may reflect differences among of Psychosocial Development: A review and evalua- the groups due to personal characteristics of tion. JSAS Catalog of Selected Documents in Psy- chology, 1981, //, 5. (Ms. No. 2179) the members, such as having been recruited Whitbourne, S. K., & Waterman, A. S. Psychosocial on a particular basis for one particular year. development during the adult years: Age and cohort In contrast, there may be a specific influence comparisons. Developmental Psychology, 1979, 15, that affects members of a given class in a 373-378. unique fashion during their college years, Received August 13, 1981 •