JOURNAL
OF RESEARCH IN PERSONALITY
20, 276-291 (1986)
Stability and Change in Personality: A Longitudinal Study from Early Adolescence to Young Adulthood JUDITH A. STEIN, MICHAEL University
D. NEWCOMB, AND P. M. BENTLER
of California,
Los Angeles
This study examines change and stability in personality between early adolescence and young adulthood. Subjects were 654 young adults (193 males, 461 females) who have been participating in an 8-year longitudinal study. Comparisons were made between measurements on 15 personality traits taken at 4-year intervals: Year 1, Year 5, and Year 9. Stability was assessed using correlations and change was measured through comparisons of mean levels of trait scores. Correlational results indicated continuity of personality between all intervals with the greatest stability to be found between the last two testing periods. Significant changes in mean levels were also evidenced for 10 out of the 15 personality traits. Changes, influenced by the moderator variable of sex, were in the direction of becoming more like societal expectations for adult roles. 0 1986 Academic Press, Inc.
One central issue in the study of personality has been the temporal stability and consistency of personality traits. Another issue is that of developmental change in personality due to maturational and cultural factors. There is no contradiction in describing both consistency and change within a particular group over time, since change is usually determined by comparisons of mean levels of personality traits, whereas stability is a measure of the relative ordering of traits, typically accomplished via correlation coefficients (Backteman & Magnuson, 1981; Costa, McCrae, & Arenberg, 1980; Moss & Susman, 1980). The present study uses both types of analyses to examine the relative stability of various personality traits, as well as personality changes, during an &year developmental period from early adolescence through young adulthood. There has been criticism of using a longitudinal approach to study the consistency of personality traits (Mischel, 1968; Nesselroade & Baltes, 1974), although others contend it is the method of choice for studying such processes (Block, 1971, 1977; McCaLl, 1977). Nesselroade and Baltes (1974) found that longitudinal studies were confounded by age and cohort This work was partially supported by Grants DA01070 and DA00017 from the National Institute on Drug Abuse. The assistance of Julie Speckart is gratefully acknowledged. Address correspondence and reprint requests to Judith A. Stein, Department of Psychology, University of California, Los Angeles, California 90024. 276 OO92-6566/86 $3 40 Copyright Q 1986 by Academic Press, Inc. All rights of reproduction in any form reserved.
LONGITUDINAL
STABILITY
AND
CHANGE
277
effects, although McCall (1977) considers such criticism overgeneralized and an impediment to progress in studying developmental changes. He states that longitudinal approaches and analyses must be used to study behavioral development within individuals. Block (1971) states that longitudinal studies “encompass time and the trajectory of individual lives” (p. 3) and that the longitudinal study is the only way to study certain questions about development. Many studies have found high stability of personality traits among adults (e.g., Conley, 1984b; Costa & McCrae, 1980; Costa et al., 1980; Mussen, Eichorn, Honzik, Bieber, & Meredith, 1980; Schuerger, Tait, & Tavernelli, 1982) even over a 45year period (Conley, 1984a). However, results from younger individuals indicate that adolescence is a period of both moderate stability and change. There is some evidence that stability increases with age, until young adulthood when personality traits seem to be fairly established (Moss & Susman, 1980; Nesselroade & Baltes, 1974). However, there have been relatively few longitudinal studies that have followed individuals from youth to adulthood. Three well-known studies are reported by Kagan and Moss (1%2), Block (1971) and Bachman, O’Malley, and Johnston (1978). Kagan and Moss (1962) found that traits with long-term stability tended to be those congruent with sex-typed behavior standards. Different patterns of personality consistency for males and females were interpreted as based on cultural expectations and socialization. For instance, greater stability was found for the trait of dependency for females than males. On the other hand, passivity is deemed inappropriate for males and, indeed, males showed more change on this dimension than females. Kagan and Moss state that “the individual’s desire to mold his overt behavior in concordance with the culture’s definition of sex-appropriate responses is a major determinant of the patterns of continuity and discontinuity in his development” (p. 269). Block (1971) studied a group of individuals in junior high school, high school, and then in their middle 30s. He used mean differences on personality traits to assess change and correlations between traits to assess stability. Both males and females changed in the direction of becoming more like adult stereotypes. At that same time, there was consistency of personality, particularly between junior high and high school. Acrosstime correlations between the junior high-senior high interval averaged .77 for men and .75 for women. Correlations between senior high school and adulthood were .56 and .54 for men and women, respectively. Of course, stability is inversely related to the length of time between assessments. Thus the shorter period during adolescence (4 years) should be more stable than the longer period into middle adulthood (20 years). Bachman et al. (1978) found more stability than change in an g-year study of males from the 10th grade to young adulthood. They state,
278
STEIN,
NEWCOMB,
AND
BENTLER
“Contrary to what might have been expected by those who view adolescence as a period of great turbulence and stress, we have found a good deal of consistency among dimensions of attitudes, aspirations, and self-concept” (p. 220). They found some change in responses to environmental circumstances and experiences, but, overall, their dominant finding was stability. In a longitudinal study of 1000 boys and girls, Backteman and Magnuson (1981) found considerable stability across a 3-year period from Age 10 to 13 for teacher ratings of various traits. Stability coefficients were 52 for boys and .48 for girls. Similarly, Olweus (1977) found high stability (.65) between the ages of 13 and 16 for ratings of aggression and peer acceptance. The trait approach to measure stability and change in personality research has been challenged by Mischel (1968). In reviewing the literature, he found little consistency in personality traits either cross situationally or longitudinally (with correlation coefficients usually no more than .30). On the other hand, Block (1977) cites a considerable amount of support for a trait approach to the study of personality, particularly in studies using self-ratings or observer ratings (as compared to artificial tests or laboratory situations, where consistency is erratic). Costa and McCrae (1980) also consider self-report methods ideal for studying personality in longitudinal research, and Block (1977) suggests that correlational results would be much higher if they were corrected for attenuation. The present study uses a self-reported trait approach to analyze change and stability in personality between early adolescence and young adulthood. Longitudinal data obtained at three time points are used to assess the relative stability and change of 15 personality traits for males and females separately. Data collection points are at 4-year intervals and began when the subjects were 13-15 years of age. It is hypothesized that greater stability will be evidenced between the second and third data points as opposed to the first and second. Implicit in this hypothesis is the belief that early adolescence is still a time of change and that personality is far from permanently established at that developmental period. Theoretical guidelines about what is being predicted in longitudinal studies of change are not as well established as those in studies of stability (Moss & Susman, 1980). It is not a hypothesis of this study that all 15 of the personality traits measured will change over time. For instance, some personality traits such as extraversion and achievement motivation appear to be fairly well established in early childhood (Emmerich, 1964). Thus, traits such as extraversion and ambition are not expected to change much during adolescence. However, it is hypothesized that traits particularly associated with self-image and self-esteem will continue to change during the teenage years. Instabilities during adolescence have been evi-
LONGITUDINAL
STABILITY
AND
CHANGE
279
dented in ratings of attractiveness, self-confidence, and conformity to authority (Kagan dz Moss, 1%2; Tuddenham, 1959). O’Malley and Bachman (1983) reported a rise in self-esteem between the ages of 13 to 23, the approximate age span of subjects in the present study. Petersen, Schulenberg, Abramowitz, Offer, and Jarcho (1984) have suggested that selfimage and self-esteem become negative during early adolescence due to the stresses of entrance into junior high school, the onset of puberty (particularly for girls), dating concerns, and concerns about peer group acceptance. Thus, there is an expectation that traits such as self-acceptance, attractiveness, law abidance, and leadership will continue to change during adolescence. Additionally, it is hypothesized that among those traits that change, such changes will be in the direction of positive societal values and better psychological adjustment. Hathaway and Monachesi (1963) found that most shifts in adolescent development tend toward better adjustment. Therefore, there is an expectation that traits such as generosity, diligence, and orderliness would show change in a positive direction. Additionally, gender will be an important moderator variable both in terms of relative levels (means) of a trait and amount of change over time. It is posited that males and females will diverge when such divergences are based upon societal expectations for men and women. For instance, Petersen et al. (1984) reported that adolescent males scored higher on achievement/leadership and had more positive self-images than females. On the other hand, females scored higher on congeniality/sociability and morals. Thus, significant sex differences are expected, with males scoring higher on traits such as ambition and leadership and females scoring higher on traits such as congeniality, religious commitment, and law abidance. METHOD Subjects Data were obtained from 654 young adults (193 males, 461 females) as part of an g-year longitudinal study of adolescent development and drug use (Huba & BentIer, 1982; Newcomb & Bentler, 1986). Only those subjects who responded in Year 1, Year 5, and Year 9 were included. Each subject was paid $12.50 to complete the follow-up questionnaire as a young adult and was informed at each wave that his/her responses were protected by a grant of confidentiality from the U.S. Department of Justice. Table 1 presents the sample description for all subjects including sex, age, ethnicity, current life activities, income for the past year, and current living arrangements. When these participant characteristics were compared to national surveys of young adults (e.g., Bachman et al., 1984; Miller et al., 1983) and other samples of young adults (e.g., Donovan, Jessor, & Jessor, 1983; Kandel, 1984), very similar patterns were noted. This group of young adults does not appear markedly different from young adults in general regarding life activity or living arrangements. The main difference is that the current sample has a greater percentage of women than men.
280
STEIN,
NEWCOMB,
AND
BENTLER
TABLE 1 DESCRIPTIONOF SAMPLE Variable Total Sex Male Female Age 21 22 23 24 Ethnicity Black Hispanic White Asian Current life activity Military Junior college University Part-time job Full-time job None Income last year None Less than $5000 Between $5000 and $15,000 Over $15,000 Current living arrangements Alone With parent(s) Spouse Spouse and child(ren) Cohabitation Dormitory With roommates Other
Number of subjects
Percentage of sample
654
100
193 461
30 70
253 214 170 17
39 33 26 2
97 64 432 61
15 10 66 9
17 79 139 89 305 25
3 12 21 14 47 4
62 215 300 76
9 33 46 12
27 311 68 43 58 37 81 29
4 48 10 I 9 6 12 4
The study originally began with 1634 adolescents who were in the seventh, eighth, and ninth grades located at 11 Los Angeles County schools. The schools were selected from a larger sample initially contacted through their district offices and then solicited for their voluntary participation in a longitudinal study of adolescent substance use. The individuals were students at cooperating schools whose parents had signed informed consent forms.
Attrition
Effects
In the first year of the study (1976) 64% of the sample were females and 36% were males. This indicates that the differential representation by sex in the young adult sample
LONGITUDINAL
STABILITY
AND
CHANGE
281
(1984) of 30% males and 70% females was also evident in the original sample and was not solely a result of differential attrition. A series of analyses were run to determine whether the attrition in sample size from 1976 to 1984 (junior high school to young adulthood) was due to any systematic influence. Those who were able to be located and provided completed questionnaires in 1984 were compared with those who were not assessed in 1984 in terms of data obtained in 1976. These groups were contrasted in terms of frequency of use for 13 different drug substances and also on mean levels of 25 different personality traits from the 1976 data set. Using the Bonferroni procedure to adjust for multiple simultaneous comparisons, not one of these 38 variables was able to significantly differentiate the new sample from those lost at the .OS level of significance. The average (absolute) point biserial correlation for these 38 tests was .04, whereas the average squared correlation was .002. The largest difference accounted for less than 1% of the variance between groups and was not significant when using the Bonferroni method to correct for capitalizing on chance. These analyses indicate that very little of the attrition between 1976 and 1984 was due to self-selection based on drug use or personality traits. To tease out any remaining differences, a stepwise multiple regression analysis was run using the thirty-eight 1976 drug use and personality variables as the predictor pool and retention in 1984 as the criterion variable. Using this procedure, 9 variables were chosen to differentiate the groups. Although significant, this equation, created by selecting all of the best predictors, was only able to account for less than 5% of the variance between groups. Those who continued in the study reported more beer use, less cigarette use, less attractiveness, more generosity, more intelligence, more vulnerability, less liberalism, less orderliness, and more trustful qualities in 1976 than those who did not continue in the study. These extensive analyses indicate that the loss of subjects between 1976 and 1984 was only slightly due to systematic self-selection or other influences based on personality, drug use, or sex.
Measures Personality traits were assessed using a self-rating test modified for this research program, but based on the Bentler Psychological Inventory (BPI) (Bentler & Newcomb, 1978; Huba & Bentler, 1982). Although the BP1 was developed with multivariate methods, the items have a high degree of face validity. In addition, half of the items for each trait are reverse scored to minimize response bias or acquiescence. In Year 4 of the study, the average coefficient (Y for internal consistency for the 15 personality scales was .72 (Huba & Bentler, 1982). The BP1 has proved useful in studies of marital success and failure (Bentler & Newcomb, 1978), criminal behavior (Huba & Bentler, 1983) and adolescent substance use (Huba & Bentler, 1982). The traits that were assessed included ambition, attractiveness, congeniality, deliberateness, diligence, extraversion, generosity, invulnerability, law abidance, leadership, liberalism, objectivity, orderliness, religious commitment, and self-acceptance. Four items were used to assess each trait and each item was rated on a 5-point bipolar scale. Thus, each scale had a range of 4 to 20. The same personality test was given at each administration of the questionnaire to the subjects over the 8 years.
Analysis Stability of personality traits was assessed by obtaining correlations between Year 1 and Year 5, Year 5 and Year 9, and Year 1 and Year 9 for all the personality variables. In addition, Z scores of the differences between the dependent correlations for each trait at different time periods were assessed (Steiger, 1980). To provide a more accurate and higher estimate of stability, the correlations were corrected for attenuation using the formula of Heise (1969) which separates reliability and stability in test-retest correlations. The method of Heise requires three measurement points and adjusts correlation coefficients for mea-
282
STEIN,
NEWCOMB,
AND
BENTLER
surement error. It provides an estimate of true scores over time (O’Malley & Bachman, 1983), assuming equal reliability of the instrument at each testing occasion. Reliabilities for each measure were also calculated using the method of Heise (1969). This method derives period-free reliability estimates and is comparable to other period-free measures such as coefficient cr or split-half procedures (Conley, 1984b). Change in personality was measured with a series of two-way repeated measures ANOVAs using sex and time as the independent factors and personality scores as the dependent variables.
RESULTS Stability Columns l-3 of Table 2 present the correlations and stability coefficients between the personality traits over the two 4-year intervals for all subjects and also separately by sex. Stability coefficients, in parentheses, were calculated using the formula of Heise (1969), which corrects test-retest TABLE OBSERVED
CORRELATIONS
AND STABILITY OVERALL
2
COEFFICIENTS
FOR THREE
~-YEAR
TIME
PERIODS
AND BY SEX
Correlations
Ambition Males Females Attractiveness Males Females Congeniality Males Females Deliberateness Males Females Diligence Males Females Extraversion Males Females Generosity Males Females Invulnerability Males Females Law abidance Males Females
Year l-5
Year 5-9
Year l-9
.31 .22 .30 .37 .42 .35 .30 .28 .30 .27 .21 .26 .36 .33 .38 .43 .45 .43 .42 .35 .44 .32 .24 .34 .41 .40 .40
.56 .56 .53 .60 .59 .60 .53 .58 .50 .44 .45 .43 .46 .42 .49 .63 .70 .60 .44 .48 .41 .53 .53 .51 .59 .63 .56
.24 .ll .24 .33 .33 .35 .33 .32 .34 .22 .18 .23 .31 .29 .31 .41 .53 .36 .29 .28 .28 .25 .26 .23 .29 .30 .27
(.43) (.20) (.45) (55) (.56) (.58) (.62) (.55) (.68) (.50) (.40) (.53) (.67) (.69) (.63) (.65) (.76) (.60) (.66) (.58) (.68) (.47) (.49) (.45) (.49) (.48) (.48)
(.77) (.50) (.80) (.89) (.78) (1.00) (1.10) (1.14) (1.13) (.81) (.67) (.88) (.86) (.88) (.82) (.95) (1.18) (.84) (.69) (.80) (.64) (.78) (1.08) (68) (.71) (.75) (.68)
(.33) (.lO) (.36) (.49) (44) (.58) (.68) (.63) (.77) (.41) (.27) (.47) (.58) (.61) (.52) (.62) (.89) (.50) (.45) (.47) (44) (.37) (.53) (.31) (.35) (.36) (.33)
Z diff -6.10*** -4.12*** -4.12*** -6.18*** - 2.46** -5.71*** - 5.83*** -4.10*** -4.31*** -3.85*** - 2.08* -3.27*** -2.45** -1.15
-2.13* -5.88*** -4.so*** - 4.07*** -0.49 - 1.75* 0.72 -4.93*** -3.77*** -3.32*** -4.76*** - 3.29*** - 3.34***
LONGITUDINAL TABLE
STABILITY
AND CHANGE
283
?-Continued
Correlations
Leadership Males Females Liberalism Males Females Objectivity Males Females Orderliness Males Females Religious commitment Males Females Self-acceptance Males Females
Year l-5
Year 5-9
Year l-9
.39 .34 .38 .34 .30 .35 .46 .36 .46 .37 .32 .39 .55 .61 .51 .29 .26 .30
.60 .54 .62 .44 .46 .43 57 .53 .55 56 .52 .58 .71 .74 .69 .40 .31 .42
.33 .24 .33 .27 .25 .28 .40 .29 .40 .36 .25 .40 .51 .56 .47 .27 .28 .25
(.55) (44) (53) (.61) (.54) (.65) (.70) (.55) (.73) (.64) (.48) (.69) (.72) (.76) (.68) (.68) (30) (.60)
(.85) (.71) (.87) (.79) (.83) (.80) (.87) (.81) (.87) (.97) (.78) (1.03) (.93) (.92) (.92) (.93) (1.08) (.83)
(.47) (.31) (.46) (.49) (.45) (.52) (.61) (44) (.63) (.63) (.38) (.71) (.67) (.70) (.63) (.63) (.97) (SO)
Z diff -5.90*** -2.66*** -5.35*** - 2.35** - 1.95* - 1.52 -3.16** - 2.20* -2.22* -5.20*** -2.53** -4.35*** -6.05*** -2.83** -5.30*** -2.33** -0.58 -2.32**
Note. Stability coefficients, in parentheses, calculated with formula of Heise (1969). Coefficients greater than unity are caused by larger correlations between Years l-9 than between Years l-5. Z scores of differences between Year I-5 correlations and Year 5-9 correlations calculated with formula of Steiger (1980). * p 6 .05. ** p s .Ol. *** p s .OOl.
correlations for attenuation. As expected, the stability coefficients were substantially higher than the raw test-retest correlations. (Coefficients greater than unity were caused by larger correlations between Year 1 and 9 than between Year 1 and 5.) In all cases, the correlations between Year 5 and Year 9 were higher than those between Year 1 and Year 5, indicating greater stability between late adolescence and young adulthood than between early and late adolescence. Z scores of the differences between the raw correlations for Year 1-5 and Year 5-9 were calculated using the method of Steiger (1980). Results of these analyses are given in column 4 of Table 2. Fiftyeight percent of the comparisons were significant beyond the .OOl level, 18% reached significance at the .Ol level, and 13% were at the .05 level. Only 5 out of the 45 comparisons between the correlations (11%) did not reach significance. Reliabilities for each trait score for all subjects and by sex are presented in Table 3. These coefficients are based on test-retest data, but are free
284
STEIN,
NEWCOMB, TABLE
AND
BENTLER
3
ESTIMATEDRELIABILITY OF TRAIT SCORESOVERALL AND Trait Ambition Attractiveness Congeniality Deliberateness Diligence Extraversion Generosity Invulnerability Law abidance Leadership Liberalism Objectivity Orderliness Religious commitment Self-acceptance Note. Reliability
BY
SEX
Total
Male
Female
.72 .67 .48 .54 .53 .66 .64 .68 .83 .71 .55 .66 .58 .77 .43
1.12
.66 .60 .44 .49 .60 .72 .64 .75 .83 .71 .54 .63 .57 .75 .50
.75 .51 .68 .48 .59 .60 .49 .84 .77 .55 .66 .67 .81 .29
coefficients calculated with formula of Heise (1969).
of temporal change effects (Heise, 1969). An average reliability of .64 was obtained for the two 4-year intervals. In 10 out of the 15 traits, reliabilities for males were higher than those of the females. One anomalous score (1.12 on ambition for males) was due to the very low correlation between Year 1 and Year 9 on this trait. Mean Comparisons Columns l-3 of Table 4 present the mean scores for males and females on each personality trait for the three time periods. A two-way analysis of variance with repeated measures was performed for each personality trait. Sex and time were the independent variables, trait scores were the dependent variables. F ratios and significance levels of sex differences, time effects, and the sex x time interactions are presented in columns 4-6 of Table 4. Figures l-2 show the changes over time that have occurred by sex for each personality trait. Sex diflerences. Collapsed over time, 10 of the personality traits showed a significant main effect for sex. Males reported more ambition, deliberateness, invulnerability, leadership, objectivity, and self-acceptance than females. On the other hand, females scored higher on measures of generosity, law abidance, orderliness, and religious commitment than males. Although it was hypothesized that females would score higher on congeniality, this was not found. Changes over time. As a main effect of time, scores increased significantly for attractiveness, congeniality, diligence, generosity, invulnerability, law
TABLE MEAN
PERSONALITY
SCORESFOR
MALES
4
AND FEMALES
AT THREE
Trait scale means
Ambition Males Females Attractiveness Males Females Congeniality Males Females Deliberateness Males Females Diligence Males Females Extraversion Mates Females Generosity Males Females Invulnerability Males Females Law abidance Males Females Leadership Males Females Liberalism Males Females Objectivity Males Females Orderliness Males Females Religious commitment Mates Females Self-acceptance Males Females
F statistics
Year 1
Year 5
Year 9
15.58 13.92
15.75 14.22
15.57 14.05
12.16 10.71
13.06 13.04
13.89 14.07
13.43 13.68
14.93 15.31
15.46 15.68
13.32 12.69
13.28 12.65
13.51 12.79
13.74 13.40
14.16 14.39
15.37 15.08
12.90 13.11
13.07 13.04
12.64 12.84
11.95 12.72
13.34 14.34
13.42 14.22
13.25 12.33
14.33 12.94
14.23 12.68
12.20 12.99
12.16 13.55
13.48 14.49
13.81 12.25
14.72 13.93
14.99 13.97
9.38 10.10
9.86 10.00
9.46 9.51
14.58 12.91
14.84 13.20
14.81 13.24
13.70 13.79
13.35 14.43
14.24 14.82
14.81 15.47
14.87 15.86
14.77 15.98
16.06 14.81
16.06 15.77
16.36 15.70
Sex
Time
Interaction
52.24***
1.06
0.10
3.50
137.39***
2.93
139.33***
0.27
0.72
0.06
0.34
62.61***
2.55
0.28
2.84
0.42
14.67***
54.45***
0.37
28.23***
11.75***
1.74
18.79***
36.10***
1.60
33.90***
70.47***
4.25*
3.01
6.17**
4.26*
2.41
0.04
10.77***
56.00***
* p s .05. ** p s .Ol. *** p s ,001. 285
TIME PERIODS
16.26***
5.51*
11.40***
4.28*
11.91***
1.51
1.91
13.29***
7.86***
4.55**
286
STEIN,
NEWCOMB,
AND
BENTLER
Attractiveness
Leadership
Liberalism
-
Males
-
Males
-
Males
----
Females
----
Females
----
Females
16.0
.
14.0
.8
.8
.6
.6
.4
13.4
15.2
.2 13.0
.a
.8
.
.6
.6
.4
12.4
14.2
.2
14.0
12.0
.8
.8
.6
i :
.6 11.4
13.0 12.8
: . 1
5 Year
9
12.2
.4 .2
:
9.0
:
.8
::
.6
:
.4
10.4
10.0.4 .6 .8 .2
:
.6
.6
::
::
13.2
.2
/’
.-_--.
/
.4
11.0 .8
11.0
15.0
.4
:
.2
. 1
8.0 5 Ye-X
9
1
5
9
YEXW
FIG. 1. Mean scores of males and females over three time periods on attractiveness, leadership, and liberalism.
abidance, leadership, orderliness and self-acceptance. There were significantly lower scores over time on liberalism. Interactions. There were five significant sex x time interactions for the traits of attractiveness, leadership, liberalism, orderliness, and selfacceptance. These interactions represent interesting cultural and maturational effects. These changes can be visualized most effectively by referring to the appropriate figure. Initially, females (13-15 years of age) were low on assessments of their attractiveness (Fig. 1) compared to the males, perhaps reflecting a lack of confidence in and possible dissatisfaction with their appearance. This result supports the suggestion of Petersen et al. (1984), that girls in early adolescence develop more negative self-images. Eight years later, these women reported slightly greater satisfaction with their appearances than the men, although both groups scored significantly higher on this trait over time. For leadership (Fig. l), there were significant main effects and a significant interaction. Females were initially much lower than males on this trait, and then increased greatly into late adolescence, with no increase beyond
LONGITUDINAL
STABILITY
AND CHANGE
287
Orderliness
Self-Acceptance
-
Hales
-
Males
----
Females
----
Females
16.8 .6 16.4 .
.2 15.0 .8
,’
.6
.
.’
.2
/
.I
.6
I
I’
.-
.8
11’
.4
.4 .
14.0 .8
16.0
.2 15.0
.’
.8
:
/’
I’
---.
I
.6 .4
.\
./
.4 .6
.2
.2
13.0
14.0
.8
.8
.6
.6
.4
.4
12.2
13.2 1
5
1
9
Year
5
9
Year
FIG. 2. Mean scores of males and females over three time periods on orderliness and self-acceptance.
this level into young adulthood. Meanwhile, the males evidenced consistent increases at each time period. Liberalism (Fig. 1) for females declined steadily over time, whereas these scores for males spiked up during late adolescence and then declined to match those of females in young adulthood. The scores for orderliness (Fig. 2) resemble a mirror image of the liberalism scores. Females reported steady increases on this trait over the 8 years, while males in late adolescence showed a sharp drop in orderliness and then an increase into young adulthood. Self-acceptance scores exhibited varying patterns for the males and females (Fig. 2). The males showed a trend of greater self-acceptance over time, whereas the females initially started off quite low, increased more rapidly during late adolescence and then showed a slight drop into young adulthood. DISCUSSION
The present study had two main goals: to evaluate evidence for increasing stability in personality over time, and to examine changes in personality during the critical period from early adolescence through young adulthood.
288
STEIN,
NEWCOMB,
AND
BENTLER
Stability was assessed through comparisons of correlations between scores at different time periods. Change was studied by looking at differences in mean scores over the three testing occasions. The correlations between the same traits across time were all quite significant, indicating a high degree of continuity in personality over this g-year period. Indeed, the average correlation between Years 1 and 9, after correction for attenuation, was .52, a substantial indication of consistency of personality traits during the so-called turbulent years of adolescence. As Costa and McCrae (1980) point out, high correlations over a period of years are evidence of stability in a trait rather than mere evidence of the reliability of the measure. The overwhelming finding of stability supports the basic results of Bachman et al. (1978) in their study of teenage males. However, as stable as the traits appeared over the 8 years, there were considerable differences in degree of stability for the two 4-year time intervals. As hypothesized, there was greater stability between late adolescence and early adulthood than between early and later adolescence. In other words, there was a greater amount of flux and variability in personality during adolescence than at the young adult period, when personality appears to have become more firmly established. There were significant mean changes over time on many personality traits. However, several traits did not change over time and apparently were well-established before early adolescence. As predicted, ambition and extraversion did not change over time. Other traits that had no main effects for time were deliberateness, objectivity, and religious commitment. Among the traits that did not change, only extraversion did not have a strong sex difference. Males scored higher on ambition, deliberateness, and objectivity. Females scored higher on religious commitment. These divergences between males and females may be caused by early cultural influences. Traits related to self-image and acceptance of adult roles did change over time. Significant changes in ratings of attractiveness, self-acceptance, leadership, and law abidance occurred as predicted. In addition, significant change over time was obtained for congeniality, diligence, generosity, invulnerability, liberalism, and orderliness. The direction of these changes indicates modeling of generally positive adult roles and expectations over time. The decrease in liberalism appears to support the notion of increasing conservatism with age. However, greater change did not always occur during the earlier time interval as hypothesized originally. Both males and females did exhibit more change during the earlier interval on measures of congeniality, generosity, invulnerability, and leadership. However, there were divergences based on the moderator variable of sex as predicted. Only females showed greater change in attractiveness during the first time period,
LONGITUDINAL
STABILITY
AND
CHANGE
289
whereas the males showed equal change during each life span. More change was apparent during the second interval (compared to the first) on law abidance for both sexes, but especially for males. The results support the stereotype of the rebellious teenage male and indicate a relatively later acceptance of extrinsic societal rules. The remaining traits showed mixed results. Females showed greater early changes and males later changes in diligence, orderliness, and self-acceptance, whereas this pattern was reversed for liberalism. The females dropped slightly in self-acceptance during the last interval, while the males continued to increase. The magnitude of the drop was quite small, but the fact remains that the males continued to increase on this measure. Bachman et al. (1978) also found a consistent increase in self-esteem over time for males. This result may lend support to Block’s (1971) conclusion that males exhibited more self-control and selfconfidence over time, while females increased in culturally conventional femininity. The ‘females did increase dramatically in assessments of their attractiveness. A decrease in self-acceptance may be an unfortunate concomitant of acceptance of the conventional feminine role. The raw correlations for self-acceptance were generally lower than those for the other measures, which bears out Conley’s (1984b) contention that self-opinion is less stable than personality. He suggested a hierarchical sequence of decreasing stability for intelligence, personality, and selfopinion, and estimated annual stabilities of .99, .98 and .94, respectively, for adults. This hierarchical relationship may also be valid for younger individuals, although not with annual stabilities of the magnitude reported for adults. One possible weakness of the present study is the use of a single cohort, which limits generalization to other cohorts (Nesselroade & Baltes, 1974). There is often a confounding of age, cohort, and year of assessment in measuring change. However, the results of the present study support findings of consistency obtained from other cohorts. Nesselroade and Baltes reported a median correlation of .59 for their stability coefficients. The attrition rate in the present study may have influenced the results. However, the detailed analyses described previously indicate few substantial differences among subjects who have left the study and those continuing. In addition, as Moss and Susman (1980) point out, if the most extreme subjects drop out of a study, “homogeneity of the sample could increase, thus making it more difficult to demonstrate stability” (p. 542). Since significant stability was shown between all time periods for the present sample, the essential heterogeneity of the present sample does appear to be intact. Two other limitations of the present study include (1) the use of a personality questionnaire not used widely in other studies and (2) possible retesting effects. The average reliability of the personality traits for the
STEIN,
290
NEWCOMB,
AND
BENTLER
4-year intervals (.64) compares favorably with average reliabilities reported by Conley (1984b) for several standard personality questionnaires. He reported average reliabilities of .61 and 66 for selected personality measures on the 16PF and MMPI, respectively. In addition, the present personality inventory has been used effectively in a variety of important research (e.g., Bentler & Newcomb, 1978; Huba & Bentler, 1982). In regard to testing effects, the 4-year intervals between the occasions should decrease the probability of retest or learning effects. In summary, the results of this study support the concept of simultaneous stability and change during the period of development from adolescence through young adulthood. Relative stability of personality traits is evidenced by the large stability coefficients obtained between the two time intervals, particularly the later interval. Change, usually in the direction of positive societal values, also is manifested during this critical period of development. The results of this study indicate that the personality is not firmly set in late childhood, but rather continues to develop and mature during the adolescent period. In addition, significant differences, either genetic or culturally based, indicate early acceptance of roles and attitudes defined by sex. REFERENCES Bachman, J. G., O’Malley, P. M., &Johnston, J. (1978). Adolescence to adulthood: Change and stability in the lives of young men. Ann Arbor, MI: Institute for Sociological Research. Bachman, J. G., O’Malley, P. M., & Johnston, L. D. (1984). Drug use among young adults: The impacts of role status and social environment. Journal ofPersonality and Social
Psychology,
47, 629-645.
Backteman, G., & Magnusson, D. (1981). Longitudinal stability of personality characteristics. Journal of Personality, 49, 148-160. Bentler, P. M., & Newcomb, M. D. (1978). Longitudinal study of marital success and failure. Journal of Consulting nnd Clinical Psychology, 46, 1053-1070. Block, J. (1971). Lives through time. Berkeley, CA: Bancroft Books. Block, J. (1977). Advancing the psychology of personality: Paradigmatic shift or improving the quality of research? In D. Magnusson & N. S. Endler (Eds.), Personality at the crossroads: Current issues in interactional psychology (pp. 37-63). Hillsdale, NJ: Erlbaum. Conley, J. J. (1984a). Longitudinal consistency of adult personality: Self-reported psychological characteristics across 45 years. Journal of Personalify and Social Psychology, 47, 1325-1333. Conley, J. J. (1984b). The hierarchy of consistency: A review and model of longitudinal findings on adult individual differences in intelligence, personality and self-opinion. Personality
and
Individual
Differences,
5, 11-25.
Costa, P. T., Jr., & McCrae, R. R. (1980). Still stable after all these years: Personality as a key to some issues in adulthood and old age. In B. P. Baltes & 0. G. Brim, Jr. (Eds.), Life-span Development andBehavior Vol. 3, pp. 65-102). New York: Academic Press. Costa, P. T., Jr., McCrae, R. R., & Arenberg, D. (1980). Enduring dispositions in adult males. Journal of Personality and Social Psychology, 38, 793-800.
LONGITUDINAL
STABILITY
AND
CHANGE
291
Donovan, J. E., Jessor, R., & Jessor, L. (1983). Problem drinking in adolescence and young adulthood. Journal of Studies on Alcohol, 44, 109-137. Emmerich, W. (1964). Continuity and stability in early social development. ChildDevelopment, 35, 31 l-332. Hathaway, S. R., & Monachesi, E. D. (1963). Adolescent personality and behavior: MMPZ patterns of normal, delinquent, dropout, and other outcomes. Minneapolis: The Univ. of Minnesota Press. Heise, D. R. (1969). Separating reliability and stability in test-retest correlation. American Sociological Review, 34, 93-101. Huba, G. J., & Bentler, P. M. (1982). A developmental theory of drug use: Derivation and assessment of a causal modeling approach. In B. P. Baltes & 0. G. Brim, Jr. (Eds.), Life-span Development andBehavior (Vol. 4, pp. 147-203). New York: Academic Press. Huba, G. J., & Bentler, P. M. (1983). Causal models of the development of law abidance and its relationship to psychosocial factors and drug use. In W. S. Laufer & J. M. Day (Eds.), Personality theory, moral development, and criminal behavior (pp. 165215). Lexington, MA: Lexington Books. Kagan, J., & Moss, H. A. (1962). Birth to maturity. New York: Wiley. Kandel, D. B. (1984). Marijuana users in young adulthood. Archives of General Psychiatry, 41, 200-209.
McCall, R. B. (1977). Challenges to a science of developmental velopment,
psychology. Child
De-
48, 333-344.
Miller, J. D., Cisin, I. H., Gardner-Keaton, H., Wirtz, P. W., Abelson, H. I., & Fishbume, P. M. (1983). National survey on drug abuse: Main Jindings 1982. National Institute on Drug Abuse. Washington, DC. Mischel, W. (1968). Personality and assessment. New York: Wiley. Moss, H. A., & Susman, E. J. (1980). Longitudinal study of personality development. In 0. G. Brim, Jr., & J. Kagen (Eds.), Constancy and change in human development (pp. 530-595). Cambridge, MA: Harvard Univ. Press. Mussen, P., Eichom, D. H., Honzik, M. P., Bieber, S. L., & Meredith, W. M. (1980). Continuity and change in women’s characteristics over four decades. International Journal
of Behavioral
Development,
3, 333-347.
Nesselroade, J. R., & Baltes, P. B. (1974). Adolescent personality development and historical change: 1970-1972. Monographs of the Society for Research in Child Development, 39, (Whole No. 154), l-79. Newcomb, M. D., & Bentler, P. M. (1986). Drug use, educational aspirations, and workforce involvement: The transition from adolescence to young adulthood. American Journal of Community Psychology, 14, 303-321. Olweus, D. (1977). Aggression and peer acceptance in pre-adolescent boys. Two shortterm longitudinal studies of ratings. Child Development, 48, 1301-1313. O’Malley, P. M., & Bachman, J. G. (1983). Self-esteem: Change and stability between ages 13 and 23. Developmental Psychology, 19, 257-268. Petersen, A. C., Schulenberg, J. E., Abramowitz, R. H., Offer, D., & Jarcho, H. D. (1984). A self-image questionnaire for young adolescents (SIQYA): Reliability and validity studies. Journal of Youth and Adolescence, 13, 93-l 11. Schuerger, J. M., Tait, E., & Tavernelli, M. (1982). Temporal stability of personality by questionnaire. Journal of Personality and Social Psychology, 43, 176-182. Steiger, J. H. (1980). Tests for comparing elements of a correlation matrix. Psychological Bulletin, 87, 245-251. Tuddenham, R. D. (1959). The consistency of personality ratings over two decades. Genetic Psychology
Monographs,
60, 3-29.