Birth Order Effects on Personality and Achievement within Families Author(s): Delroy L. Paulhus, Paul D. Trapnell and David Chen Source: Psychological Science, Vol. 10, No. 6 (Nov., 1999), pp. 482-488 Published by: Sage Publications, Inc. on behalf of the Association for Psychological Science Stable URL: https://www.jstor.org/stable/40063474 Accessed: 27-03-2019 02:06 UTC JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org. Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at https://about.jstor.org/terms Sage Publications, Inc., Association for Psychological Science are collaborating with JSTOR to digitize, preserve and extend access to Psychological Science This content downloaded from 198.246.186.26 on Wed, 27 Mar 2019 02:06:32 UTC All use subject to https://about.jstor.org/terms PSYCHOLOGICAL SCIENCE Research Article BIRTH ORDER EFFECTS ON PERSONALITY AND ACHIEVEMENT WITHIN FAMILIES Delroy L. Paulhus,1 Paul D. Trapnell,2 and David Chen1 1 University of British Columbia, Vancouver, British Columbia, Canada, and2 Ohio State University at Mansfield Abstract - We investigated birth order effects on personality and (Goldberg, 1990). Using this organizational system, Sulloway's metaachievement in four studies (N = 1,022 families) including both stu- of the apparently chaotic literature exposed the predicted patanalysis dent and adult samples. Control over a wide range of variables wasIn particular, FBs were more conscientious but less agreeable and terns. effected by collecting within-family data: Participants compared their open to experience than LBs. Sulloway (in press) has followed up those siblings (and themselves) on a variety of personality and achievement analyses with new data that are consistent with his predictions. dimensions. Across four diverse data sets, first-borns were nominated Recent studies from other quarters vary from supportive (Davis, as most achieving and most conscientious. Later-borns were nominat1997; Salmon & Daly, 1998) to nonsupportive (Parker, 1998; Phillips, 1998). Jefferson, Herbst, and McCrae (1998) found mixed support: ed as most rebellious, liberal, and agreeable. The same results obtained whether or not birth order was made salient (to activate The few significant birth order differences obtained in the peer-rating stereotypes) during the personality ratings. Overall, the resultsdata sup-fell in the direction predicted by Sulloway. Self-ratings and port predictions from Sulloway's niche model of personality developspouse ratings on the same individuals, however, showed none of the ment, as well as Zajonc 's confluence model of intellectual predicted effects. achievement. All these studies used between-family designs; that is, the individ- uals being compared with respect to birth order came from different families. Among the known confounds of between-family birth order The notion that birth order has an influence on personality fell into data are social class,1 parental personality, and sibship size. Unfortu- disrepute with the publication of Ernst and Angst (1983). Althoughnately, a full range of appropriate controls is seldom available. Withinthey conceded small effects on intellectual achievement (e.g., Zajoncfamily data would provide a natural control procedure for all & Markus, 1975), Ernst and Angst disputed any link between birthbetween-family differences, including their largest contributor order and personality traits. Recently, however, a reconsideration has genetics (Dunn & Plomin, 1990). Therefore, within-family analyses been provoked by the publication of Sulloway's (1996) book, Born toshould be more powerful (Sulloway, in press), as well as more valid Rebel. In applying his new theoretical perspective, Sulloway reaf- (Rodgers, 1988). We expected they would confirm the following prefirmed the view that adult personality differs systematically across dictions: FBs would be perceived as more conscientious and achieving, birth order. According to Sulloway, the source of these differences is and LBs would be perceived as more agreeable, liberal, and rebellious. not, as traditionally argued, a differential parental treatment of chil- Our prediction of superior FB achievement was also consistent with dren of different birth orders (e.g., Hilton, 1967). Instead, Sulloway's the well-known confluence model (e.g., Zajonc & Markus, 1975). thesis was that birth order effects derive from a competition among We collected four within-family data sets by asking respondents to siblings as they fight for a family niche. compare themselves and their siblings on various personality and First-borns (FBs), having the first choice of niche, attempt to please achievement dimensions. In Study 1 , undergraduates from the Univertheir parents in traditional fashion via success in school and responsi- sity of California were asked to nominate the "rebel" and the "achiev- ble behavior. But, as other siblings arrive, FBs must deal with threats er" in their families. Study 2 applied the same methodology to a to their natural priority in the sibling status hierarchy. The resultingsample of undergraduates from the University of British Columbia adult character is conscientious and conservative. Later-borns (LBs) while evaluating an alternative hypothesis. Study 3 extended the critemust resist the higher status of FBs, while seeking alternative ways ofrion variables to include the Big Five personality traits. Finally, Study distinguishing themselves in the eyes of their parents. Accordingly, 4 replicated Study 3 in a large sample of Vancouver adults. they develop an adult character marked by an empathic interpersonal style, a striving for uniqueness, and political views that are both egalitarian and antiauthoritarian. In short, they are "born to rebel." STUDY 1: PRELIMINARY DATA FROM UNIVERSITY OF CALIFORNIA STUDENTS Sulloway's (1996) book resonated on a personal level with the general public while offering two forms of data to readers more persuaded by empirical evidence. One form was the systematic documentation of Method the social attitudes of historical figures as a function of their birth order. In fall 1996, a large, intact class (N = 164) was asked several quesThe second form of empirical support was a meta-analysis of the large number of studies on personality and birth order (see also Sulloway, tions as part of a class demonstration. Participation was voluntary. 1995, in press). To great advantage, Sulloway organized the studies First, the students were asked to write down the birth order of the boys within the influential Five-Factor Model of personality, or "Big Five" and girls in their family, including themselves (e.g., B-G-G). Then, 1. Moreover, controlling for socioeconomic status has the side effect of removing some ability variance. As a result, it is difficult to demonstrate correAddress correspondence to Del Paulhus, Department of Psychology, University of British Columbia, Vancouver, Canada V6T 1Z4; e-mail: lations between ability-related variables (e.g., achievement, conscientiousness) dpaulhus@cortex.psych.ubc.ca. and birth order. In short, between-family data entail an intractable confound. 482 Copyright© 1999 American Psychological Society This content downloaded from 198.246.186.26 on Wed, 27 Mar 2019 02:06:32 UTC All use subject to https://about.jstor.org/terms VOL. 10, NO. 6, NOVEMBER 1999 PSYCHOLOGICAL SCIENCE Delroy L. Paulhus, Paul D. Trapnell, and David Che Table 1. Proportion of families with first-borns nominated as the achiever and later-borns nominated as the (American students) First-born achievers Later-born rebels Significance Effect size Significance Effect size Sibship Observed- Difference Odds Observed- Difference Odds size n expected (O - E) Phi ratio expected (O - E) Phi ratio 2 66 .65-.50 .15* .30 3.48 .61-.50 .11* 3 38 31-33 .04 .05 1.26 .71 -.67 .04 4 29 .35-.25 .10 .13 1.89 .83-.75 .08 5-8 15 Varies .10 .13 2.29 Varies .09 Combined 148 .11** .19 2.28 .08* Note. N = 164 families; the 16 .21 .07 .10 .11 .14 one-child 2.37 1.35 1.83 2.97 2.00 families are not inc calculated by weighting the difference for each sibship size frequency. Fisher r-z transformations were applied before c ratios. Significance tests were based on binomial approxima *p < .05, one-tailed. **p < .01, one-tailed. to outcome (achiever vs. nonachiever or rebel vs. nonthey were asked to "put the acolumns square around the sibling w Calculationexample, of the LB values required a correction for number scholastically achieving,rebel). for gets the highes school." School sonably compared pants your sized were own Finally, asked with to definition that the of LBs cited in each sibship size. Consider, for example, the rate of achievwere so that young siblings c ers in sibship size 3.of The ratecollege for FBs was .368 (rounded to .37 in Table older. N siblings age or and therefore an LB was the achiever in .632 (1.00 - .368) of the put a 1), circle around the "rebel" in the families. Because there are 2 LBs in this sibship size, however, the rate told. It w of the term rebel," they were grades of achievers among LBsbe was .316 (.632/2). The values for the for both person could nominated nonachiever column of the 2 x 2 table were thetheir complements ofown .316 were asked to indicate birth same participants and .368, namely, .684 and .632. Participants were not warned in advance that the topic such the 2x2 tables, we computed our first effect-size index,was phi, order was to be addressed From in course. Nor it cov the product-moment correlation for two dichotomous variables not (see course textbook. Sulloway's (1996) book was yet a I).2 believe This index is intuitively appealing they because it represents the been inf short, there was no reason Table to that had correlation between birth order and nomination as achiever (or rebel). the recent birth order research. Averaged over all birth orders, the mean phi was .19 between birth order (favoring FBs) and nomination as the achiever. Similarly, the mean phi Results and Discussion was .14 between birth order (favoring LBs) and nomination as the rebel. Although intuitively appealing, phi coefficients are not ideal for The results are presented in Table 1, separately for each sibship combining across conditions (Fleiss, 1994). An alternative effect size but with sizes 5 to 8 pooled because of small frequencies. Thesize 16 - one that remains constant under changes in marginal sibships of size 1 (i.e., only-children) were not used. frequencies - is the odds ratio, also included in Table 1. The combined odds ratio for achievement can be interpreted as meaning that the rel- Significance tests ative proportion of achievers to nonachievers is 2.28 times higher Note from Table 1 that, for every sibship size, the observed probaamong FBs than among LBs. Similarly, the mean odds ratio for rebels (2.00) indicates that the proportion of rebels to nonrebels is twice as bility of an FB being nominated as the achiever was higher than expected by chance (i.e., the rate of FBs in that family size). Similarhigh in LBs as it is in FBs. ly, for every sibship size, the probability of an LB being nominated as the rebel was higher than expected by chance. Unfortunately, binomiFBs in the class al tests reached significance only for sibship size 2; for other sibship Recall that we had also asked participants to indicate what their sizes, the samples were too small to confirm the cell differences. When own birth order was. The rate of FBs in our sample (.44) was signifi- significance levels were combined (see Rosenthal & Rosnow, 1991, cantly p. higher than the rate expected by chance (.34), %2(1, N = 164) = 504) across the four sibship sizes, however, the hypothesis was con7.0, p < .01 . The chance rate refers to the proportion of FBs that would firmed for both FB achievers (combined z = 2.40, p < .01) and be LB expected in our sample of participants had they been randomly rebels (combined z = 1.87, p < .05). selected from the families they reported on. Effect sizes For each cell of Table 1, effect size was calculated by constructing 2. In this type of table, phi corresponds to the binomial effect-size display a 2 x 2 table of proportions (birth order by outcome). The rows of each (see Rosenthal & Rosnow, 1991, p. 281). The values may be interpreted as the 2x2 table corresponded to dichotomous birth order (FB vs. LB), and differences in rates of nominations of FBs and LBs. VOL. 10, NO. 6, NOVEMBER This content downloaded from 198.246.186.26 on Wed, 27 Mar 2019 02:06:32 UTC All use subject to https://about.jstor.org/terms 1999 483 PSYCHOLOGICAL SCIENCE Birth Order and Personality Discussion High-salience condition As in Study 1, participants (n = 217) were asked to write Overall, our results are consistent with previous evidence that FBs birth order of the boys and girls in their family (e.g., B-G are more achieving and that LBs are more rebellious. One might wonwere then asked to indicate which of the children was the scholastic der, though, whether the effects we found are limited to theachiever highlyand which was the rebel. selected student body at the University of California, Berkeley, cam- pus. Another potential limitation of this study was the susceptibility to Low-salience condition contamination by stereotypes about birth order; that is, our respon- Participants (n = 178) were asked to write down the initials of the dents may have had preconceived notions about birth order and perscholastic achiever and the rebel among the children in the family. sonality that influenced their nominations of achievers and rebels. Then they were asked to write down the birth order of each of the STUDY 2: SALIENCE STUDY WITH UNIVERSITY OF BRITISH COLUMBIA STUDENTS nominees and the total number of children. In this condition, there was no particular reason for participants to reflect on birth order and its effects while they were doing the nominations. These potential limitations were addressed in Study 2 by collecting a data set that differed in several respects from the set in Study 1 . First, Results and Discussion the data were collected at a less selective university in another country, namely, the University of British Columbia in Vancouver, Canada. of nominees for each category were again calcuThe proportions Second, the salience of birth order during the personality was family size. The two salience conditions were comlatedratings within each manipulated. Although some participants were questioned as using z tests for proportions. There were no pared exactly cell by cell in Study 1, others were asked to make their nominations by initial significant differences on either dependent variable between the high- only, and only afterward were asked to specify the birth of the conditions (all zs < 1.40). The pooled results, disandorder low-salience nominees. played in Table 2, closely resemble those from Study 1 . In particular, the effect sizes in Table 2 are remarkably similar to those in Table 1. In terms of phi coefficients, the mean effect size for FB versus LB achievers was .19 (identical to the value in Study 1). The comparable Method Data were collected in four intact classes (total N = 395). Asfigure in for LB versus FB rebels was .18 (compared with .14 in Study 1). The odds ratios were also similar across the two studies. Study 1, participants were asked to nominate the "scholastic achiever" and the "rebel in the family." They were advised that the same person This consistency suggests that the results of Study 1 were not an could be nominated for both. In two classes, birth order was made artifact of an idiosyncratic sample. The predicted pattern emerged just salient during nominations (high-salience condition); in two other as clearly at the Canadian university as at the American university. classes, the topic of birth order was not made salient during nominaBecause the sample sizes were larger than those in Study 1, however, tions (low-salience condition). To control for possible differences in of eight cells reached significance (p < .07) on a binomial test. seven morning and afternoon classes, we counterbalanced the salience conCombined across sibship size, the effects were significant for both FB ditions across time of day. As in Study 1, none of the classes achievers was (p< .01) and LB rebels (p < .01). In terms of effect sizes, the warned in advance that the topic of birth order would be covered in two the weakest associations appeared for sibship size 3 (phi coefficients of .10 and .07). course. Table 2. Proportion of families with first-borns nominated as the achiever and later-borns nominated as the rebel (Canadian students) First-born achievers Later-born rebels Significance Effect size Significance Effect size Sibship Observed- Difference Odds Observed- Difference Odds size n expected (O - E) Phi ratio expected (O-E) Phi ratio 2 165 .63-.50 .13** .26 2.91 .64-.50 .14** .27 3 115 .40-.33 .07* .10 1.56 .71-.67 .05 .07 4 64 .39-.25 .14** .19 2.52 .88-.75 .13** .17 5-8 25 Varies .12* .15 2.49 Varies .12* .14 Combined 369 .11** .19 2.31 .11** .18 3.06 1.38 2.88 5.27 2.45 Note. N = 395 families; the 26 one-child families are not inclu calculated by weighting the difference for each sibship size by frequency. Fisher r-z transformations were applied before com ratios. Significance tests were based on binomial approximatio *p < .07, one-tailed. **p < .01, one-tailed. 484 VOL. 10, NO. 6, NOVEMBER This content downloaded from 198.246.186.26 on Wed, 27 Mar 2019 02:06:32 UTC All use subject to https://about.jstor.org/terms 1999 PSYCHOLOGICAL SCIENCE Delroy L. Paulhus, Paul D. Trapnell, and David Che STUDY 3: STUDENT TAKE-HOME PACKAGE nominating one family member for each variable, students ranked all family members on each variable. The two dependent variables studied up to this point rebelliousness and intellectual achievement - capture a rather limited range of human personality. Broad trait taxonomies typically reveal Results the five-dimensional personality space known as the Big Five, or the The results are summarized in Table 3. Data for the 9 only-children Five-Factor Model (see Goldberg, 1990; Trapnell & Wiggins, 1990). Indeed, Sulloway (1995, in press) profited considerably from orga- are not included in the table. For simplicity and consistency with earnizing his findings in terms of the Big Five. He was able to show that FBs lier tables, we retained our dichotomous scoring of all dependent vari- were higher on Conscientiousness and LBs were higher on Openness toables. On agreeableness, for example, we assigned a score of 1 to the Experience and Agreeableness. Accordingly, in Study 3, we expanded individual who was nominated as highest in the family; all others were our range of questions to tap four of the Big Five traits. Neuroticism wasassigned a score of 0. The asterisks in the row showing mean differences in Table 3 indi- omitted because it has the weakest effects (Sulloway, in press). We asked participants to rank themselves and their siblings on cate that all hypotheses were supported. Moreover, the pattern is conseven variables. The terms rebellious and scholastically achieving weresistent across sibship sizes. That is, FBs were nominated as most included to correspond to the variables used in Studies 1 and 2. Factorconscientious as well as most achieving more frequently than expect1 of the Big Five (Extraversion) was represented by socially confident,ed by chance. In contrast, LBs were more frequently nominated as one of its highest loading items (Trapnell & Wiggins, 1990). Factor 2 most liberal, agreeable, and rebellious. The combined significance lev- (Agreeableness) was represented by agreeable, and Factor 3 (Consci-els were significant for all predicted outcomes. Neither of the variables entiousness) by conscientious. Factor 5 (Openness to Experience) was for which we made no predictions (creativity and extraversion),. howrepresented by rebellious, creative, and liberal (see Trapnell, 1994). ever, showed significance in either direction. Effect sizes, as indexed Based on the literature cited, we predicted that FBs would be ratedby phi coefficients, were highest for conscientious (.20) and liberal as more achieving and conscientious. We also predicted that LBs(.18). would be perceived as more liberal, rebellious, and agreeable. Creativity (despite its association with Openness) and extraversion have shown weak or mixed results in the literature (Sulloway, 1996), so we Discussion made no predictions about those two variables. Two additional potential artifacts must be considered as explanations of the observed birth order differences in Studies 1 through 3. The finding that LBs were more likely to be nominated as rebels may Method be an artifact of the age range of the raters. The youngest siblings in families of students 19 to 21 years old are likely to be teenage or The data collection (N = 203) differed from the procedure in Studyounger. In other words, the LBs are likely to be of an age for which ies 1 and 2 in that (a) rather than answering questions in class, students is commonplace. As teenagers grow out of this period, took questionnaire packages home to complete and (b) rather rebelliousness than Table 3. Proportion of families consistent with each hypothesis: Study 3 Favor first-borns Favor later-borns No prediction Scholastic Sibship size n achiever Conscientious Liberal Rebellious Agreeable Extraverted Creative 2 3 4 5 6 107 61 18 4 4 .54 .43 .39 .25 .25 .65 .61 .56 .59 .52 .51 (.04) (.15)* (.11)* (.06) (.09) (.02) (.01) .34 .77 .74 .74 .34 .69 (.09) (.01) (.10)* (.07) (.07) (.01) (.02) .39 .78 .78 .78 .33 .67 (.14) (.14) (.03) (.03) (.03) (.08) (-.08) .25 1.00 .50 .50 .00 .50 (.05) (.05) (.20) (-.30) (-.30) (-.20) (-.30) .50 1.00 1.00 .75 .25 1.00 (.08) (.33) (.17) (.17) (-.08) (.08) (.17) Mean difference from chance (.07)* (.11)* (.10)* (.06)* (.07)* (.02) (.00) Mean effect size (phi) .11 .20 .18 .10 .13 .04 .01 Note. N = 203 families; the 9 one-child families are not incl calculated so that positive values indicate differences consis entries by sample size. Some of the differences from chanc *p < .05, one-tailed. VOL. 10, NO. 6, NOVEMBER This content downloaded from 198.246.186.26 on Wed, 27 Mar 2019 02:06:32 UTC All use subject to https://about.jstor.org/terms 1999 485 PSYCHOLOGICAL SCIENCE Birth Order and Personality Table 4. Proportion of adult families consistent with each hypothesis: Study 4 Achievement Big-Five-related traits Favor first-borns Favor later-borns No prediction Sibship size n Scholastic Financial Prestige Conscientious Liberal Rebellious Agreeable Extraverted 2 3 4 5 6 55 60 48 33 44 .55 .55 .53 .60 .56 .60 .58 .55 (.05) (.05) (.03) (.10) (.06) (.10) (.08) (.05) .38 .45 .40 .50 .88 .78 .77 .33 (.05) (.12)* (.07) (.17)* (.21) (.11)* (.10) (.00) .25 .33 .29 .31 .83 .73 .83 .23 (.00) (.08) (.04) (.06) (.08) (-.02) (.08) (-.02) .36 .27 .30 .24 .92 .94 .82 .21 (.16)* (.07) (.10) (.04) (.12) (.14)* (.02) (.01) .39 .23 .32 .25 1.00 .93 .82 .11 (.22)* (.06) (.15)* (.08) (.17) (.10)* (-.02) (-.05) Mean difference from chance (.09)* (.08)* (.07)* (.10)* (.14)* (.09)* (.06)* (.00) Mean effect size (phi) .12 .11 .10 .15 .24 .13 .10 .01 Note. N = 260 families; the 20 one-child families are not in so that positive values indicate differences consistent with *p < .05, one-tailed. maturity-related 1998). would Although be an birthanalysis order differences should if any of the participant's siblings were less than 30 years of such a maturity artifact age. A total of 260 questionnaires was usable.is Ages plausible of siblings ranged w in 30 to 61 (M = 44.2, SD = 9.9). Education ranged from differe 6 years to implausible from explanation if similar adult data. 20 years, with a median of 12 years. Similarly, the finding that FBs are more scholastically achieving The package of questionnaires was similar to that used in Study 3. than LBs may follow from the age range of our raters. The LBs that One change was the inclusion of three forms of achievement (scholaswere rated, being teenagers or younger, may have had little opportu- tic, financial, prestige) instead of one. The term extraverted was used nity to exhibit intellectual achievement - at least, they may not have to represent Factor 1 (Extraversion), and the term reliable was used to had as much opportunity as the FBs. Measurements taken later in life represent Factor 3 (Conscientious). Finally, creative was dropped. It might not show the FB advantage. Study 4 was designed to over- was made clear that respondents were to rate the siblings in their fam- come these possible artifacts in the student samples of Studies 1 ilies of origin, not their own children. through 3. Results and Discussion STUDY 4: VANCOUVER ADULT TAKE-HOME PACKAGE The results are summarized in Table 4. They appear remarkably similar to the data obtained with college students. All three forms of achievement showed the predicted pattern, and the findings for the Questionnaires were administered to a large sample of adults who other variables replicated earlier studies. The combined significance were older than 40 years of age. These adults were asked to provide levels for the seven predictions were all significant. Note from Table 4 personality rankings of their own families of origin. Because of the that the largest mean effect sizes were .15 for conscientious (favoring age restriction, the possible artifacts attributable to student samples FBs) and .24 for liberal (favoring LBs). should have been eliminated. For one thing, all the rated individuals were well beyond the "rebellious" teenage years and were, therefore, more comparable. And by age 30, an individual's intellectual achieve- GENERAL DISCUSSION ment (as well as other forms of achievement) should be evident. The Asboost a whole, replication of our findings in this older generation would our the studies reported here confirm the birth order confidence in the robustness of these birth order effects. ferences predicted by the family-niche model of personality dev ment (Sulloway, 1996), as well as the confluence model of intelle development (e.g., Zajonc & Markus, 1975).3 Given the mixed sup Method from recent between-family studies, our success likely derives f A sample of 309 adults was solicited by asking university students to take home a questionnaire package to be completed by their parents 3. Of course, our data are mute with respect to the dynamics of these m or other adults over 40. A returned questionnaire was excluded from els, and are consistent with other theories (Rodgers, 1988). 486 VOL. 10, NO. 6, NOVEMBER This content downloaded from 198.246.186.26 on Wed, 27 Mar 2019 02:06:32 UTC All use subject to https://about.jstor.org/terms 1999 PSYCHOLOGICAL SCIENCE Delroy L. Paulhus, Paul D. Trapnell, and David Che our use of the powerful within-family methodology. ticipants This additional develop from false attributions that, nevertheless, have a power follows from the built-in control over a variety of between-fampermanent, substantial impact on self-conception akin to a self- ily differences, namely, social class, family size, and, especially, fulfilling prophecy. The argument further requires that, throughout their lives, siblings systematically ignore bona fide evidence of their genetics. brothers' and sisters' traits in favor of erroneous impressions fostered by maturity-related or age-related roles and stereotypes within their families. To us, it seems far more reasonable to believe that such Big Five Personality Traits The results for the personality traits largely followed the birth order pattern emerging from Sulloway's (1995) meta-analysis. The weakest effects were for Extraversion, one of the two weakest factors in Sulloway's summary. Clear differences were found, however, in Consci- entiousness, Agreeableness, and two of our indicators of Factor 5 (Openness to Experience), namely, liberalism and rebellion. Another indicator of Openness, creativity, did not show significant differences. This null finding is consistent with Sulloway's (1996) historical data, as well as data he has collected on contemporary samples (Sulloway, in press). Because intelligence is confounded with openness in person perception (Trapnell, 1994), attributions of creativity combine one trait favoring FBs with another trait that favors LBs. This combination has a null relation with birth order. stereotypes exist because they are true - that is, birth order does influence personality development - rather than to believe that the stereotypes, the self-perceptions, and the peer perceptions are all faulty. The achievement results seem least assailable for two reasons. First, the facts about which adult sibling achieved the most or which child sibling received the best grades should be relatively concrete and objective. We can see no self-serving motivation for our participants to have named the FB as the achiever. Only a fraction were nominating themselves, so allegations of self-serving responding cannot be sustained (Paulhus, 1991). Second, these nominations are backed up by a concrete indicator in Study 1: The proportion of FBs (44%) was significantly higher than chance (34%). Although this comparison is vulnerable to the usual confounds of between-family designs (Rodgers, 1988), this concrete indicator converges with the within-family indi- Intellectual Achievement cators to provide mutual support. The age restriction in our first three studies raised the possibility of Our finding that FBs are perceived as more intellectually achieving two artifacts: The first was that LB siblings of college students were than LBs is consistent with previous work using concrete indicators less achieving because they had less opportunity to display achieve- (e.g., Paulhus & Shaffer, 1981; Zajonc & Markus, 1975). The results ment behavior. The second was that these LBs were perceived as held whether intellectual achievement was operationalized as school rebellious because many were experiencing the (notoriously rebel- grades or ratings on intellectual achievement. This effect was not lious) teenage years. This limitation was overcome in Study 4 by samrestricted to intellectual achievement, but extended to financial and pling respondents age 40 and up. Given that the lowest reported prestige achievement. Future research showing a similar pattern for sibling age was 30, even the youngest LBs in this study had lived sufunconventional forms of achievement would support Zajonc 's claims ficiently long to display intellectual achievement. Moreover, these for a general intellect advantage in FBs. Finding that this pattern LBs were well past the inherently rebellious teenage years. reversed (or at least diminished) for unconventional forms of achievement would support Sulloway's theory that LB achievement typically has a radical flavor. Effect Sizes Needless to say, perceptions of intellectual achievement are not Effect sizes for the FB-versus-LB comparisons varied in a coherequivalent to concrete indicators (Davis, 1997; Paulhus, Lysy, & Yik, ent fashion across the dimensions we assessed. Averaged over all four 1998). The former incorporate perceptions of conscientiousness, studies, the effect size for liberalness was strongest (phi = .21). The which may inflate the association with FBs (Sulloway, in press). comparable figure for intellectual achievement was .16. These figures are likely to be underestimates because we used only single-item indicators for all our variables. For an index combining all five predictions Controlling Artifacts in Study 3, the effect size was .24 - a substantial figure on a par with reported by Sulloway (in press). We dealt with several threats to internal validity in Studiesthat 2 through 4. Study 2 revealed no differences between an administration Some critics might argue that our within-family design is too powcondition in which birth order was made salient and a conditionerful in in the sense that it detects birth order differences that are trivial in daily life (Ernst & Angst, 1983; Harris, 1998; Jefferson et al., 1998). which birth order was not mentioned until after the sibling nominaThe between-family studies, which found weak results at best, are said tions had been made. Even without the direct activation of stereotypes, to be more representative of life beyond the family of origin. But even the predicted effects still obtained. those critics concede the likelihood that birth order has an impact (a) Perhaps the stereotypes run deeper than that: They may have during the developmental years and (b) during continuing interactions already had a permanent impact on the way our subjects perceived with one's family of origin. These effects alone are reason to take our their brothers, their sisters, and themselves. If so, this impact must birth order effects seriously. Certainly, we can rule out the claim that endure, because our adult sample showed the same pattern and size birth order effects are "parent-specific" (Ernst & Angst, 1983), of birth order effects as much younger samples, despite (presumably) they may be "family-context-specific." living apart from siblings for many years. Yet the very stability although of these perceptions across the life span undermines the accusation thatAnother reason for taking our results seriously is the emerging consensus (e.g., Bouchard, 1997; Dunn & Plomin, 1990; Jang, McCrae, they are artifactual and makes a stereotype perspective difficult to Angleitner, Riemann, & Livesley, 1998; Rowe, 1997) that (a) distinguish from standard conceptions of personality. The stereotype between-family differences in personality and intellect are dominated argument implies that the birth order differences reported by our par- VOL. 10, NO. 6, NOVEMBER This content downloaded from 198.246.186.26 on Wed, 27 Mar 2019 02:06:32 UTC All use subject to https://about.jstor.org/terms 1999 487 PSYCHOLOGICAL SCIENCE Birth Order and Personality K.L., McCrae, R.R., Angleitner, A., Riemann, R., & Livesley, W.J. (1998). HeriI by genetic variance and (b) the environmental varianceJang, is largely withtability of facet-level traits in a cross-cultural twin sample: Support for a hierarchi- in family. It follows that social scientists interested in intervention cal model of personality. Journal of Personality and Social Psychology, 74, 1556-1565. must turn their attention to processes that operate to differentiate the Jefferson, T, Herbst, J.H., & McCrae, R.R. (1998). Associations between birth order and children within the same family. personality traits: Evidence from self-reports and observer ratings. Journal of 32, 498-509. Parker, W.D. (1998). Birth order effects in the academically talented. Gifted Child Quarmodest relative to the Note, finally, that any single within-family source, such as birth Research in Personality, order or peer effects (Harris, 1998), may seem terly, 42, 29-36. between-family genetic variance. Yet even modest effect sizes can Paulhus, D.L. (1991). Measurement and control of response bias. In J.P Robinson, P.R. translate into dramatic social consequences (RosenthalShaver, & Rosnow, & L.S. Wrightsman (Eds.), Measures of personality and social psychologi- cal attitudes (pp. 17-59). San Diego: Academic Press. 1991; Sulloway, in press). And it is precisely the within-family effects Paulhus, D.L., Lysy, D.C., & Yik, M.S.M. (1998). Self-reports of intelligence: Are they useful as proxy IQ tests? Journal of Personality, 66, 525-554. Paulhus, D.L., & Shaffer, D.R. (1981). Sex differences in the impact of number of older and younger siblings on scholastic aptitude. Social Psychology Quarterly, 44, that are most amenable to the benevolent tools of psychology. Acknowledgments - We thank Jeff McCrae, Dan Ozer, Dave Shaffer, 363-368. and Frank Sulloway for comments on an earlier draft. Thanks toPhillips, Jennifer Beer,February). Birth order and personality- Not again? Paper presented S.M. (1998, at the annual meeting Paul Wehr, and Kim Barchard for assistance with the analyses. This of the Eastern Psychological Association, Boston. Rodgers, J.L. (1988). research was supported by a grant to the senior author from the Social Sci- Birth order, SAT, and confluence: Spurious correlations and no causality. American Psychologist, 43, Aid-All. ences and Humanities Research Council of Canada. Rosenthal, R., & Rosnow, R.L. (1991). Essentials of behavioral research: Methods and data analysis (2nd ed.). New York: McGraw-Hill. Rowe, D.C. (1997). Genetics, temperament, and personality. In R. Hogan, J.A. Johnson, & REFERENCES S.R. Briggs (Eds.), Handbook of personality psychology (pp. 367-386). San Diego: Academic Press. Bouchard, T.J. (1997). The genetics of personality. In K. Blum, E.P. Noble, R.S. Sparkes, Salmon, C.A., & Daly, M. (1998). The impact of sex and birth order on familial sentiment: T.H.J. Chen, & J.G. Cull (Eds.), Handbook of psychiatric genetics Middleborns (pp. 273-296). are different. Evolution and Human Behavior, 19, 299-312. Boca Raton, FL: CRC Press. Sulloway, F.J. (1995). Birth order and evolutionary psychology: A meta-analytic overview. Davis, J.N. (1997). Birth order, sibship size, and status in modern Canada. Human Nature, Psychological Inquiry, 6, 75-80. 8, 205-230. Sulloway, F.J. (1996). Born to rebel: Birth order, family dynamics, and creative lives. New Dunn, J., & Plomin, R. (1990). Separate lives: Why siblings are so different. New York: York: Pantheon. Basic Books. Sulloway, F.J. (in press). Birth order, sibling competition, and human behavior. In P.S. Ernst, C, & Angst, J. (1983). Birth order: Its influence on personality. New York: Davies & H.R. Holcomb (Eds.), The evolution of minds: Psychological and philoSpringer-Verlag. sophical perspectives. Boston: Kluwer Academic. Fleiss, J.L. (1994). Measures of effect size for categorical data. In H. Cooper & L.V. Trapnell, P.D. (1994). Openness versus intellect: A lexical left turn. European Journal of Hedges (Eds.), Handbook of research synthesis (pp. 245-260). New York: Russell Personality, 8, 273-290. Sage Foundation. Trapnell, P.D., & Wiggins, J.S. (1990). Extension of the Interpersonal Adjective Scales to Goldberg, L.R. (1990). An alternative "description of personality": The Big-Five factor include the Big Five dimensions of personality. Journal of Personality and Social structure. Journal of Personality and Social Psychology, 59, 1216-1229. Psychology, 59, 781-790. Harris, J.R. (1998). The nurture assumption: Why children turn out the way they do. New Zajonc, R.B., & Markus, G.B. (1975). Birth order and intellectual development. Psychological Review, 82, 74-88. York: Free Press. Hilton, I. (1967). Differences in the behavior of mothers toward first-born and later-born children. Journal of Personality and Social Psychology, 7, 282-290. 188 VOL. 10, NO. (Received 7/9/98; Accepted 1 1/24/98) 6, NOVEMBER This content downloaded from 198.246.186.26 on Wed, 27 Mar 2019 02:06:32 UTC All use subject to https://about.jstor.org/terms 1999 Attachment & Human Development, 2013 Vol. 15, No. 2, 219–239, http://dx.doi.org/10.1080/14616734.2013.754985 Interlinkages between attachment and the Five-Factor Model of personality in middle childhood and young adulthood: a longitudinal approach Mari Franssona*, Pehr Granqvistb, Gunilla Bohlina and Berit Hagekulla a Department of Psychology, Uppsala University, Uppsala, Sweden; bDepartment of Psychology, Stockholm University, Stockholm, Sweden (Received 11 January 2012; final version received 18 August 2012) In this paper, we examine concurrent and prospective links between attachment and the Five-Factor Model (FFM) of personality from middle childhood to young adulthood (n ¼ 66). At age 8.5 years, attachment was measured with the Separation Anxiety Test and at 21 years with the Adult Attachment Interview, whereas the personality dimensions were assessed with questionnaires at both time points. The results showed that attachment and personality dimensions are meaningfully related, concurrently and longitudinally. Attachment security in middle childhood was positively related to extraversion and openness, both concurrently and prospectively. Unresolved/disorganized (U/d) attachment was negatively related to conscientiousness and positively related to openness in young adulthood. U/d attachment showed a unique contribution to openness above the observed temporal stability of openness. As attachment security was also associated with openness, the duality of this factor is discussed together with other theoretical considerations regarding attachment theory in relation to the FFM. Keywords: attachment; Separation Anxiety Test; Adult Attachment Interview; personality; Big Five Introduction The aim of this study was to further our understanding of the interlinkages between attachment and personality development, more specifically the Five-Factor Model of personality, from middle childhood to young adulthood. This aim strikes at the core of some of the fundamental questions raised in developmental and personality psychology: How does a person become who she is? And what are the constituents of her personality? Personality and the Five-Factor Model An important thread of research in personality psychology has been devoted to understanding and tapping the dimensions underlying individual differences in personality ‘‘traits’’. There has been considerable controversy regarding the number *Corresponding author. Email: Mari.Fransson@psyk.uu.se © 2013 Taylor & Francis 220 M. Fransson et al. (e.g., Block, 1995; Eysenck, 1991), definition (e.g., Zuckerman, Kuhlman, Joireman, Teta, & Kraft, 1993), and stability (e.g., Mischel, 1969; Pervin, 1994) of such traits as well as whether the trait concept is to be understood as explanatory (e.g., motivational; McCrae & Costa, 2008) or as merely descriptive (e.g., consistency in functioning; Hogan & DeSoto, 1977; Pervin, 1994). Setting these controversies aside1, the most established model of the structure of personality is presently the Five-Factor Model (FFM, also named the ‘‘Big Five’’; Digman, 1990; John, Naumann, & Soto, 2008; McCrae & Costa, 2008). The five dimensions of this model, gained through factor analyses of trait-descriptive terms across a variety of studies, are extraversion, agreeableness, conscientiousness, neuroticism, and openness (John et al., 2008). These factors have also been found in many different cultures, though deviations in the number and meaning of the factors sometimes occur (McCrae & Costa, 1997a; McCrae et al., 2005). Extraversion reflects surgency, energetic and positive emotions, and the tendency to actively seek, instead of avoiding, the company of others. Agreeableness refers to the tendency to be compassionate, empathetic, and cooperative towards others rather than suspicious and antagonistic. Conscientiousness captures socially prescribed impulse control, the tendency to act in task- and goal-directed ways, and to be able to delay gratification. Neuroticism reflects the presence and effects of negative affect such as anger, anxiety, and sadness, as opposed to emotional stability. Openness to experiences refers to complexity, depth, and quality of a person’s mental and experiential life, reflecting appreciation for creativity, curiosity, and a variety of experience (John et al., 2008; Shiner & Caspi, 2003). Openness has been the most difficult factor to consistently conceptualize across studies and instruments: labels and their associated contents have varied between, for example, culture, intellect, and openness to experiences (McCrae & Costa, 1997b). The development of the five factors from childhood to adulthood Factor analyses of parent and teacher ratings of children’s personality have supported the FFM also in childhood (Shiner & Caspi, 2003), although the evidence for openness is weaker (Lamb, Chuang, Wessels, Broberg, & Hwang, 2002). Additionally, the predictive validity of the FFM in childhood has been supported by empirical links to constructs such as cognitive performance and adjustment to school (John, Caspi, Robins, Moffitt, & Stouthamer-Loeber, 1994; Lamb et al., 2002). Furthermore, there is at least moderate continuity in personality dimensions across childhood, adolescence, and adulthood (Roberts & DelVecchio, 2000). Although personality trait research has been successful in describing the broad structure of individual differences in personality, less is known about the developmental trajectories to these five factors of personality from childhood to adulthood (John et al., 2008). There is, however, growing evidence for a temperamental core to personality (Caspi, Roberts, & Shiner, 2005), but how temperament is elaborated into personality dimensions is rather unclear. Empirical studies have supported a substantial genetic contribution to personality (Bouchard & Loehlin, 2001), which is in line with the idea of a temperamental origin of personality. Some researchers even argue that personality traits are largely independent of environmental influences (McCrae & Costa, 2008), whereas others stress the importance of both genes and environment in personality development (Lewis, 2001; Rothbart, Ahadi, & Evans, 2000; Shiner & Caspi, 2003). Attachment & Human Development 221 Attachment theory and personality development Attachment theory provides a complementary framework for understanding personality development. Although most research in attachment has been devoted to understanding socio-emotional aspects of development, Ainsworth and Bowlby (1991) portrayed attachment theory as a theory of personality development. Attachment theory, in contrast to most theories of temperament and personality, emphasizes the importance of caregiving (an aspect of nurture) in development, and more specifically in the development of individual differences in attachment organization (i.e., secure, avoidant, ambivalent, and disorganized/disoriented attachment; Ainsworth, Blehar, Waters, & Wall, 1978; Main & Solomon, 1990). Because attachment organization has been found to foreshadow a broad repertoire of developmental outcomes (Sroufe, Egeland, Carlson, & Collins, 2005), attachment theory may be a suitable theoretical model also for understanding environmental contributions in the development of personality (Fraley & Shaver, 2008). According to Bowlby (1969/1982), the attachment system evolved because it potentiated physical proximity between offspring, who are helpless on their own early in life, and their protective caregivers in times of threat and potential danger. Through repeated sequences of interaction with the caregiver, an emotional tie (i.e., an attachment) is formed between infants and their caregivers. The quality of this tie varies depending on aspects of caregiving. For example, sensitive responsiveness in caregivers foreshadows secure attachment, while insensitive and frightening caregiving as well as maltreatment are related to insecure (i.e., avoidant and ambivalent/resistant) and disorganized/disoriented attachment (e.g., Ainsworth et al., 1978; Carlson, Cichetti, Barnett, & Braunwald, 1989; De Wolff & van IJzendoorn, 1997; Madigan et al., 2006; Main & Hesse, 1990). A set of cognitive-affective representations (i.e., internal working models; IWMs) of self and others, are formed in the context of the infant’s attachment-related experiences. These IWMs presumably become more generalized over time, and come to guide the child’s expectations and behavioral inclinations in future relationships. Securely, unlike insecurely, attached children gain a sense of being worthy of care, and come to expect others as competent and reliable providers of care (Bowlby, 1973). People tend to carry forward relational behaviors learned within their family experiences into their interactions with the broader social world, in turn reconfirming their mental models of the self, others, and relationships across the life span (e.g., Weinfield, Sroufe, Egeland, & Carlson, 2008). Thus, continuity of attachment patterning over time is generally expectable (Hesse, 2008). However, IWMs are also susceptible to revision based on, for example, marked changes in contextual conditions affecting the caregiver’s sensitivity (Bowlby, 1973). Regardless of continuity or discontinuity in attachment patterning, the attachment system continues to be of great importance across the life course (Bowlby, 1969/1982). In early research on individual differences in attachment organization, the main focus was on the organized forms of attachment (secure, avoidant, and ambivalent/ resistant attachment; Ainsworth et al., 1978). This research led to an extensive literature on the developmental pathways to and from individual differences in attachment security. Since the introduction of disorganized attachment (Main & Solomon, 1990), the importance of attachment disorganization for future externalizing problems and psychopathology has been emphasized (e.g., Fearon, Bakermans-Kranenburg, van IJzendoorn, Lapsley, & Roisman, 2010; van IJzendoorn, Schuengel, & Bakermans-Kranenburg, 1999). 222 M. Fransson et al. Attachment theory and the Five-Factor Model Although described as a theory of personality development by its founders (Ainsworth & Bowlby, 1991), few empirical investigations have been undertaken to study the specific interrelations between attachment and the FFM. This is presumably because attachment theory portrays the mind as inherently relational, rather than as made up by general traits. Nevertheless, early experiences with attachment figures may serve as a foundation for the acquisition of a broad range of future abilities, such as social skills, emotion regulation capabilities, and exploratory behaviors (e.g., Sroufe et al., 2005; Weinfield et al., 2008), that are presumably linked to personality development. Also, links between attachment and the specific dimensions of the FFM can be theoretically substantiated, especially with regard to attachment security. Presumably through its association with a positive view of the self as a worthy and capable agent and of others as responsive to the self, attachment security is linked to aspects of increased sociability, such as a readiness to establish new relationships, which are in turn core constituents of extraversion (e.g., Londerville & Main, 1981; Main & Weston, 1981; Schneider, Atkinson, & Tardiff, 2001). Similarly, along with affirmative caregiver behaviors, a positive view of self and others facilitates aspects of social skills, such as cooperation and reciprocity, which are core constituents of agreeableness (e.g., Bohlin, Hagekull, & Rydell, 2000; Schneider et al., 2001; Sroufe et al., 2005). Also, although conscientiousness partly refers to higher-order cognitive (rather than emotional, social, or relational) capacities, secure attachment has been linked to better delay of gratification, executive capacity, and flexibility of attention (Belsky, Garduque, & Hrncir, 1984; Jacobsen, Huss, Fendrich, Kruesi, & Ziegenhain, 1997; Main, 2000), suggesting it might be related to higher conscientiousness as well. This is presumably because a secure attachment relationship liberates mental resources for efficient information processing rather than being occupied with defensive strategies (cf. Bowlby, 1973; Main, 2000). Moreover, conceivably due to a sensitive attachment figure’s reliable responsiveness and competent assistance during states of distress, secure attachment is associated with efficient emotion regulation skills (e.g., Cassidy, 1994; Waters et al., 2010), which are relatively absent for individuals high in neuroticism. Finally, as secure attachment is characterized by a freedom to explore (initially using the caregiver as a secure base), security is, not surprisingly, linked to increased exploration (Belsky et al., 1984; Hazen & Durrett, 1982; Matas, Arend, & Sroufe, 1978; van den Boom, 1994), which may later be expressed in high openness to experience. However, openness to experience has also been related to different forms of alterations in consciousness, such as dissociation, absorption, and hypnotizability (Glisky, Tataryn, Tobias, Kihlstrom, & McConkey, 1991; Ruiz, Pincus, & Ray, 1999). Disorganized attachment, including unresolved/disorganized loss and abuse, has, in turn, been linked to dissociation and absorption (Carlson, 1998; Granqvist, Fransson, & Hagekull, 2009; Hesse & van IJzendoorn, 1999), suggesting that disorganized attachment and openness might also be associated. In one of few extant studies of the associations between attachment and the FFM, attachment security in infancy was found to be positively related to extraversion and openness, and negatively related to neuroticism in middle childhood (Hagekull & Bohlin, 2003). The only study of which we are aware that has investigated the relation between a representational measure of adult attachment (the Adult Attachment Interview [AAI]; Main, Goldwyn, & Hesse, 2003) and the FFM in adulthood found that attachment security was positively but modestly Attachment & Human Development 223 associated only with conscientiousness and extraversion (Roisman et al., 2007). None of these studies included information on attachment-related disorganization. Although social psychological measures of romantic attachment (or adult attachment styles) tend to be only modestly related to developmental attachment measures (e.g., Roisman et al., 2007), results from the larger social psychological literature on romantic attachment and the FFM are conceptually relevant. The most consistent result in that literature is that attachment security is negatively related with neuroticism, and positively related to extraversion and conscientiousness (Noftle & Shaver, 2006). Missing pieces and the present study The purpose of the present study was to examine concurrent and longitudinal relations between attachment and the FFM of personality in middle childhood and young adulthood. An additional aim was to explore whether attachment, in middle childhood as well as in young adulthood, made a unique contribution to personality in young adulthood over and above any observed temporal stability in personality. To our knowledge, no prior study has investigated the longitudinal relation between childhood attachment and adult personality (in terms of the FFM), nor between unresolved/disorganized attachment and personality. The present study addressed these gaps in the literature, although unresolved/disorganized status was assessed only in adulthood. Based on joint considerations of theoretical assumptions and results from previous studies, we expected concurrent and longitudinal associations between attachment security and high extraversion, high agreeableness, high conscientiousness, low neuroticism, and high openness in middle childhood as well as in young adulthood. In addition, we expected a concurrent link between unresolved/ disorganized attachment status and high openness in young adulthood. The scarcity of relevant literature prevented us from making directed predictions regarding relations between unresolved/disorganized status and the four remaining personality factors. For the same reason, we did not have expectations regarding the unique contribution of attachment, beyond stability in personality, to specific adult personality factors in young adulthood. Although our expectations regarding associations between attachment and personality covered all the dimensions of the FFM, we did not expect strong links between attachment and personality. To the contrary, we expected generally modest relations on the grounds that attachment organization stems mainly from interactions with the attachment figure, whereas personality factors are substantially constitutional in origin. Method Participants and procedure The Uppsala Longitudinal Study (ULS) The sample for which data will be presented in this article is based on 66 (34 female) participants (54% continuation rate) of the ULS who took part in relevant data collections in middle childhood and early adulthood. A restriction of the sample was undertaken thus that participants who did not partake at either assessment point 224 M. Fransson et al. were excluded from the analyses so as to obtain comparable statistical power and results across analyses. One participant who fulfilled this criterion was excluded due to a serious brain injury, which occurred between the two occasions. The ULS originally involved 123 children from middle-class families followed from infant age 6 weeks to child age 9 years (for a more detailed description, see Hagekull & Bohlin, 2003). The main focus of the ULS was to investigate if and how attachment, temperament, and family context interact in children’s socio-emotional development. Educational levels of the parents were fairly high as can be expected in a university area like Uppsala County. The contacts were intensive during the first four years (13 data waves). Thereafter, three data collections were undertaken between ages 8 and 9 years. At the age of 21, a follow-up study was completed. Reasons for attrition over the years have been illness/death in the family, moving out of the county, travels abroad, objections to interview/questionnaire content, and time shortage. The present study concerns attachment and personality at ages 8.5 and 21 years. Age 8.5 years When the children were in the second grade of elementary school, 91 participants visited our department laboratory together with their mothers. As a part of the procedure, the so called ‘‘Seattle version’’ of the Separation Anxiety Test (SAT; Slough & Greenberg, 1990; Slough, Goyette, & Greenberg, 1988) was performed with the children. At the same occasion, the mothers filled out a Five-Factor Model questionnaire to assess the children’s personality (Lanthier, 1993). The children’s schoolteachers filled out the same questionnaire some months later when a research assistant visited the children in their schools. At that time, the majority of the teachers had known the children for almost two school years. Age 21 years Of the original sample, 112 participants, whose addresses could be found through the Swedish National Tax Board, were invited to participate in a follow-up study. Ninety-nine persons responded, and 85 agreed to take part. Main attrition causes were lack of time and living abroad. Some of the participants who agreed to take part in this assessment had not participated in the 8.5 year data collection. At the recruitment as well as at the scheduled laboratory visit, confidentiality of participation was explained. Two weeks before the visit to the department, questionnaires containing among other instruments the Big Five Mini-Markers (Saucier, 1994) were sent to the participants. They were asked to fill it out at least one week before their visit, and bring their marked questionnaire to the laboratory visit. Eighty-five participants filled out the questionnaires. The first task during the full day laboratory visit consisted in the individual administration of the Adult Attachment Interview (AAI; Main et al., 2003). One participant denied to be recorded on tape during the AAI, and therefore was not interviewed, as this would have made coding impossible. The participants received a compensation of 500 Swedish crowns (US$79) for participating. For independent validation purposes, permission was obtained during the lab visit to contact one to three peers in order to send them some of the same Attachment & Human Development 225 questionnaires the participants had filled out with regard to how the peers perceived the participants. They were instructed to select peers who knew them well, but were not their partner. Provided that the participant had agreed, peers were contacted by phone a few weeks after the laboratory visit. They were informed about confidentiality of participation. If the peers were willing to participate, a questionnaire was sent to them by mail. They were asked to fill out the questionnaire and return it by mail. When more than one peer took part, the mean value of their scores was used. The peers received a cinema gift certificate worth 90 Swedish crowns (US$14). Ninety friends returned the questionnaires, distributed on 54 participants. The present sample The child mean age for the 66 participants comprised in the present study was 8.6 years (SD ¼ 2.9 months) when they visited the laboratory for the SAT and the FFM measure, and 9.1 years (SD ¼ 2.9 months) when the school teachers filled out the FFM measure. At the follow up, mean age of the participants was 21.3 years (SD ¼ 5.4 months). Regarding occupation, 52% of the sample were students (university 36%, high school level 8%, professional program 8%), and 51% were in the work force (fulltime 21%, part-time 30%; students often had part-time jobs), and 12% were in search of work. Forty-six participants had peer ratings on the Big Five MiniMarkers. According to t tests, the 66 participants in the present sample did not differ from the original sample on mother’s or father’s educational level (ps 4 .10). In addition, there were no differences on the attachment or personality variables at 8.5 years between the participants who dropped out from the age 21 years assessment and those who remained in the study (all ps 4 .10). Two-tailed significance tests were used in these and all remaining analyses reported in this article. Instruments The attachment variables were obtained via semi-structured interviews (i.e., the SAT and the AAI), whereas information regarding personality was obtained via questionnaires. To rule out shared method variance and self-report biases as alternative explanations to any observed relations between attachment and personality, we used independent coders’ representational assessments (not selfratings) of attachment and included independent raters’ (teachers and peers) ratings of personality. In the longitudinal analyses, we controlled for temporal stability in personality. Measurement characteristics, descriptive statistics, and reliabilities are shown in Table 1. Middle childhood instruments The Seattle version of the Separation Anxiety Test (SAT; Slough & Greenberg, 1990; Slough et al., 1988) was used to measure attachment at age 8.5 years. This is a modification of the test originally presented by Hansburg (1972) and adapted by Klagsburn and Bowlby (1976) and Kaplan (1987). This version consists of six pictures (photos) showing separations between children and their parents. Three of 226 M. Fransson et al. Table 1. Measurement characteristics and descriptive statistics for the attachment and personality scales. Variable Scale range No. of Items M SD Reliability coefficienta Attachment dimensions SAT security AAI Coherence of transcript AAI Unresolved/disorganized 1–12 1–9 1–9 1 1 1 10.02 4.46 3.33 2.30 1.84 1.84 .84 .87 .75 Big Five variables Age 8.5 years Extraversion Agreeableness Conscientiousness Neuroticism Openness 1–5 1–5 1–5 1–5 1–5 9 13 10 6 4 3.77 3.68 3.71 2.26 4.05 0.60 0.49 0.77 0.80 0.58 .86 .86 .92 .89 .80 Age 21 years Extraversion Agreeableness Conscientiousness Neuroticism Openness 1–9 1–9 1–9 1–9 1–9 8 8 8 8 8 6.48 7.25 6.01 4.25 6.18 1.45 0.74 1.13 0.96 1.03 .90 .68 .75 .62 .71 Note: aReliability for SAT was computed with Cohen’s kappa, and the AAI dimensions with Pearson correlations; reliability for the personality variables was computed with coefficient alphas. the situations are considered severe (e.g., parents leaving for a two-week vacation) and the rest are considered mild (e.g., mother tucks child in bed for the night and then leaves the room). The child is presented with the picture together with an oral vignette explaining the situation. The interviewer then asks what the depicted child may feel, why he/she may feel that way, and what the child might do. In the Seattle version, the participating child is first asked about the feelings of the pictured child and then how the participating child him- or herself would feel in the same situation. In the present study only answers referring to the pictured child are reported, as analyses of answers to questions about the child’s own feelings did not change conclusions. Verbal answers were transcribed and scored by a person who had not previously been involved in the project (for inter-rater agreement, see Table 1). In order to minimize the number of variables, all scales used in the Seattle scoring procedure (attachment, self-reliance, and avoidance) were not used in the present study. Instead, a total score reflecting the overall security of attachment representations was calculated and used in subsequent analyses. Within the SAT system, security is reflected in an ability to verbalize feelings of vulnerability in relation to separation and at the same time imagine resources to constructively cope with separations (Kaplan, 1987). In the coding procedure used in this study, described in Verschueren and Marcoen (1996; Verschueren, 1996), the 21 Seattle response categories are organized into three groups: (1) the four most insecure response categories (e.g., avoidant or bizarre), which receive a weight of 0, (2) the six fairly insecure categories (e.g., atypical attachment or atypical self-reliant), which receive a weight of 1, and (3) the remaining 11 categories considered secure, which receive a weight of 2. Thereafter, the responses of each child are scored in accordance with these guidelines Attachment & Human Development 227 and added across the six pictures, resulting in a scale ranging from 0 to 12. Validity of the SAT has been indicated by associations to measures of the quality of the representation of self (Verschueren, 1996). The Five Factor Model instrument used in middle childhood was a Swedish adaptation of a questionnaire developed from Lanthier’s (1993) factor analysis of an instrument for self- and other ratings of 10–15-year-olds. The questionnaire contains 42 personal descriptors, relating to the five dimensions: extraversion/surgency (9 items), agreeableness (13 items), conscientiousness (10 items), neuroticism (6 items), and openness to experiences (4 items). In the present measure for children, openness is focused on creativity and curiosity rather than on culture and intelligence. All items were rated on scales ranging from 1 (not at all characteristic of my child) to 5 (very characteristic). Mother–teacher agreement across 5 months was: extraversion r ¼ .51, agreeableness r ¼ .40, conscientiousness r ¼ .38, neuroticism r ¼ .36, and openness r ¼ .28, all ps 5 .05. To increase reliability and capture the child’s personality as shown in two important contexts, home and school, individual scale scores were obtained by averaging mother and teacher item scores, yielding the internal consistency estimates reported in Table 1. Young adulthood instruments Adult attachment was measured with the Adult Attachment Interview (AAI; Main et al., 2003; translated into Swedish by Broberg, Ivarsson, & Hinde, 1996), a semistructured interview containing 20 questions with specific guidelines for follow-up probes. The participants are asked to describe and evaluate their childhood through attachment-specific memories. The interviews normally vary in length between 45 and 90 minutes and are coded from verbatim transcriptions. The most important questions ask participants to select adjectives to describe their childhood relationships with parents, which they are subsequently asked to support by recalling specific episodes; to describe what happened as a child when they were emotionally upset, ill, and in pain; to recall feelings associated with physical separation from parents; to elaborate on experiences of rejection and fear; and to speculate on the effects of childhood experiences on current personality. Another set of important questions concerns loss through death and experiences of abuse. The AAIs were coded according to Main and colleagues’ (2003) scoring and classification system. The transcripts were coded on three types of 9-point scales: (a) probable experiences, (b) organized states of mind, and (c) Unresolved/disorganized (U/d) attachment. Individuals were then classified into one of five categories (i.e., autonomous, dismissing, preoccupied, unresolved/disorganized, and cannot classify) based on their state of mind and U/d scores. However, to retain full scale variance and maximize statistical power in view of this study’s expected modest associations, these categories were not used in subsequent analyses. Scores on the specific organized states of mind scales are reflected in the overall coherence of discourse. Coherence refers to the extent to which participants are collaborative, and the transcript provides a credible, internally consistent, freeflowing picture of the participant’s experiences and feelings regarding attachment. High coherence may be present regardless of whether participants’ experiences with parents had been primarily positive or negative. Coherence is directly assessed on two scales: coherence of transcript and coherence of mind. Coherence of transcript refers to the speaker’s narrative clarity and internal consistency, and is the 228 M. Fransson et al. conceptual counterpart of security in the AAI system (Main et al., 2003). Low scores on coherence of transcript are present for individuals who are scored high on any of the specific ‘‘negative’’ state of mind scales (i.e., idealization, insistence on lack of memory, derogation of attachment, involving anger, and/or passivity of thought processes in relation to attachment). Besides these aspects of coherence, coherence of mind scores are based on the speaker’s apparent beliefs/belief systems. Unusual beliefs, especially surrounding trauma or loss, as shown in unresolved/disorganized (U/d) loss/abuse (below), can be stated coherently but indicate lapses in the monitoring of reasoning. Thus, coherence of mind scores incorporate signs of U/d attachment, while coherence of transcript is more independent of U/d considerations (Main et al., 2003). As U/d scores were assigned in their own right, we report findings based on coherence of transcript (rather than coherence of mind) in this article. Unresolved/disorganized loss–abuse is scored on a separate scale tapping speech specifically surrounding loss–abuse experiences. A high score refers to discourse characterized by one or more of three subtypes of unresolved speech: (1) striking lapses in the monitoring of reasoning (e.g., as implied in statements implying that a dead person remains alive in the physical sense or via considerable spatial–temporal confusion surrounding the loss event), (2) striking lapses in the monitoring of discourse (e.g., visual–sensory images related to the trauma intrude discourse), and (3) extreme–lingering behavioral reactions to the traumatic event (e.g., the interviewee is suicidal and this is believed by the person to be due to the loss of a close loved one). The highest score assigned to any given loss or abuse incident is used as the overall U/d loss–abuse score. To minimize the number of statistical analyses, only the continuous coherence of transcript and U/d scale scores were used in subsequent analyses. The interviews were conducted by the first author and coded by two certified coders: the second author and Professor Anders G. Broberg a certified AAI trainer. The coders were blind to all other data, except sex and age. The reliability and validity of the AAI are well-established (Hesse, 2008). For reliability of the coherence of transcript and U/d ratings, see Table 1. Disagreements were resolved through discussion. As for the FFM in adulthood, participants completed the Big Five MiniMarkers (Saucier, 1994), which is a shortened version of Goldberg’s unipolar BigFive Markers (Goldberg, 1992). This version consists of 40 adjective markers, evenly distributed over the five scales, instead of 100 as in the original version. These were rated on a 9-point scale ranging from ‘‘Extremely unlike me’’ to ‘‘Extremely like me’’ in the self-rated instrument, and ‘‘Extremely unlike my friend’’ to ‘‘Extremely like my friend’’ for the peer-ratings. The five dimensions are labeled: extraversion, agreeableness, conscientiousness, emotional stability, and intellect or openness and are described above in the Big Five paragraph. For the sake of consistency with the child scales, the terms neuroticism and openness are henceforth used also for the adult scales. The internal consistency has been found acceptable (Saucier, 1994), also in a Swedish population (Ghaderi & Scott, 2000). The peer-ratings were exclusively used to estimate interjudge agreement. The concordances between self and peer ratings were: extraversion r ¼ .59, conscientiousness r ¼ .37, neuroticism r ¼ .39, openness r ¼ .43, all ps 5 .05, and agreeableness r ¼ .28, p 5 .10. These levels of interjudge agreement conform to those found in other FFM studies (Connelly & Ones, 2010) and provide further support for the psychometric properties of the instrument. Attachment & Human Development 229 Results Descriptive statistics, preliminary analyses, and stability of attachment and personality Descriptive results are shown in Table 1. Sex differences in personality and attachment were analyzed. On the Big Five questionnaire at age 8.5 years, one significant sex difference was found; girls were rated as more agreeable, t(64) ¼ 2.40, p 5 .05. Two marginally significant differences were also found; girls were rated as more conscientious, and less neurotic, t(64) ¼ 2.00, p ¼ .050; t(64) ¼ 1.92, p 5 .10. At age 21, women scored higher on agreeableness, t(64) ¼ 2.97, p 5 .01. These sex differences were controlled in the relevant analyses. There were no significant or trend-significant sex differences on the remaining personality dimensions or on the attachment dimensions at either time point. The stability of the Big Five factors was studied with Pearson correlations. The results showed that extraversion and openness were significantly stable from 8 to 21 years (r ¼ .47, p 5 .001; and r ¼ .35, p 5 .01, respectively) whereas the stability of neuroticism was marginally significant (r ¼ .24, p 5 .10) over the time period. Agreeableness and conscientiousness did not show significant stability (rs ¼ .16 and .17, n.s.). Concerning continuity and predictability of attachment, SAT security was not related to AAI coherence of transcript nor to AAI unresolved/disorganized attachment (rs ¼ .08 and .20, n.s.). Tests of hypotheses and research questions Concurrent relations between attachment and personality To determine if attachment and personality were related at 8.5 years, SAT security was correlated with each of the Big Five dimensions at 8.5 years. The results (see Table 2) showed that children who were rated high on attachment security were also rated high on extraversion and openness. The same analyses were performed with the AAI and Big Five variables at 21 years (see Table 2). High values on unresolved/disorganized loss/abuse were significantly associated with low scores on conscientiousness and high scores on openness. Coherence of transcript was close to significantly related to extraversion; those who were coded as more coherent rated themselves as more extraverted. Prospective relations between attachment at 8.5 years and personality at 21 years To establish if attachment at 8.5 years was related to the Big Five dimensions at age 21 years, the two sets of measures were correlated. The results (see Table 2) revealed that children with a high degree of attachment security were significantly more extraverted and close to significantly more agreeable and open in adulthood. The unique contribution of attachment to extraversion and openness at 21 years The overall results indicate that the Big Five dimensions most consistently related to attachment were Extraversion and Openness. These two Big Five dimensions were eligible for tests of unique contributions from attachment, as they proved both to be stable between the two age periods, and to be related to attachment variables at either time-point. To determine if SAT security at age 8.5 years, AAI coherence of 230 Table 2. M. Fransson et al. Concurrent and prospective correlations between attachment and personality. SAT 8.5 years AAI 21 years Security Coherence of transcript Unresolved/disorganized Big Five 8.5 years Extraversion Agreeableness Conscientiousness Neuroticism Openness .42** 7.02 7.09 7.02 .25* – – – – – – – – – – Big Five 21 years Extraversion Agreeableness Conscientiousness Neuroticism Openness .39** .22þ 7.09 .06 .22þ – .23þ 7.01 7.14 .05 .10 – .02 7.09 7.28* .19 .34** Note: Ns ¼ 66 except for correlations with AAI Unresolved/disorganized (N ¼ 64), as two participants had no experience of loss or abuse, which made U/d scoring inapplicable. þ p 5 .10; *p 5 .05; **p 5 .01; two-tailed. transcript, and/or AAI unresolved/disorganized at age 21 years contributed uniquely to these personality dimensions at age 21, when stability of the relevant personality dimension was controlled, hierarchical regressions were performed. As we were also interested in the individual contribution of the attachment variables, representing different age periods and instruments, to adult extraversion and openness, middle childhood and adult attachment were entered in separate steps. For each regression analysis, one of the eligible dimensions of the Big Five at 8.5 years (i.e., Openness or Extraversion) was entered in the first step, SAT security was entered in the second step, and AAI Coherence of transcript and AAI Unresolved/ disorganized were entered as a block in the third step. The results from the regression for extraversion (see Table 3) revealed that SAT security showed a marginally significant (p ¼ .06) contribution to Extraversion at age 21, when controlling for Extraversion at 8.5 years. The results from the regression for Openness (see Table 4) showed that AAI Unresolved/disorganized had a unique contribution to Openness at age 21 when the stability of openness was controlled. Discussion This study showed that attachment and the FFM of personality are concurrently interlinked in middle childhood and young adulthood. The results also revealed longitudinal associations between attachment in middle childhood and personality in young adulthood. Regarding the relative role of early personality and attachment in the development of adult personality, unresolved/disorganized attachment in adulthood made a unique contribution above the effect of stability in personality between the two age periods. Although neither attachment security nor unresolved/ disorganized status was related to all five factors, and the significant relations obtained were of modest-moderate strength, the results suggest that attachment and personality are meaningfully interrelated (cf. Hagekull & Bohlin, 2003; Roisman et al., 2007). 231 Attachment & Human Development Table 3. 21 years Hierarchical regression results for attachment variables predicting extraversion at Extraversion at 21 years DR 2 Step 1: Age 8.5 years Extraversion .22*** Step 2: Age 8.5 years SAT security .05þ Step 3: Age 21 years AAI Coherence of transcript AAI Unresolved/disorganized .03 b .46*** .24þ .15 7.10 Note: N ¼ 64. þ p 5 .10; *p 5 .05; **p 5 .01; ***p 5 .001; two-tailed. Table 4. years. Hierarchical regression results for attachment variables predicting openness at 21 Openness at 21 years 2 DR Step 1: Age 8.5 years Openness .11** Step 2: Age 8.5 years SAT security .02 Step 3: Age 21 years AAI Coherence of transcript AAI Unresolved/disorganized .09* b .33** .14 .00 .30* Note: N ¼ 64. þ p 5 .10; *p 5 .05; **p 5 .01. Extraversion and Openness were the personality factors most consistently related to attachment. Securely attached children in middle childhood were concurrently rated as more extraverted by their parents and schoolteachers. As young adults, these individuals also rated themselves as relatively high in extraversion compared to those who were judged less secure at age 8.5 years. These relations are in accordance with the assumption that securely attached individuals are more confident in taking place in the social world than insecurely attached children. Stepping outside the framework of the FFM, similar relations have been found in studies investigating concepts closely related to extraversion. In these studies, securely (compared to insecurely) attached children were more socially active and positive toward peers and non-peers alike (e.g., Bohlin et al., 2000; Schneider et al., 2001). The possibility of 232 M. Fransson et al. predicting young adult extraversion from middle childhood attachment security was further supported as the prospective link between the two phenomena remained marginally significant also when controlling for stability of extraversion. Regarding the relative contribution of attachment across time and instruments, attachment security in middle childhood, unlike attachment variables in adulthood, was found to be predictive of extraversion in young adulthood, though only marginally so when the stability in personality was taken into account. Nevertheless, to the best of our knowledge, a prospective link between childhood attachment and the FFM in adulthood has not been previously established. Regarding openness, and viewed across developmental periods, the links between attachment and openness observed in this study are intriguing, as attachment security in middle childhood was concurrently positively related to openness and unresolved/disorganized attachment in adulthood was similarly concurrently positively related to openness. Although both findings were predicted based on theoretical considerations and previous research, they may at first glance seem conflicting, and thus should be interpreted with caution until replicated. Notably, the two classifications are often but by no means always (e.g., Steele, Steele, & Croft, 2008) related to different outcomes, with security generally linked to adaptive outcomes and disorganization (including unresolved status) to maladaptive ones (Hesse, 2008). How can we, then, understand that unresolved/disorganized status in adulthood and attachment security in middle childhood were both positively related to openness? A first possibility concerns equifinality (i.e., different pathways leading to the same outcome) where (in this case) childhood security and adult unresolved/disorganized status would both set a path leading to high openness to experience. A second possibility is that unresolved/disorganized status and security are related to different facets of openness. Notably, openness is the one FFM factor surrounded by most controversies regarding definition (McCrae & Costa, 1997b). Glisky and colleagues (1991) argued that the openness factor in the FFM instrument NEO Personality Inventory (Costa & McCrae, 1985) could be split into two facets: absorption and social liberalism/curiosity, because the absorption facet, but not social liberalism/curiosity, was related to Tellegen’s Absorption Scale (Tellegen & Atkinson, 1974) and hypnotizability. Based on previous findings linking unresolved/disorganized attachment to absorption (e.g., Granqvist et al., 2009; Hesse & van IJzendoorn, 1999), we thus speculate that unresolved/ disorganized attachment is related to aspects of openness associated with absorption. We also suggest that attachment security might be associated with the social liberalism/curiosity facet, reflecting freedom to explore social and other matters in an open, non-defensive manner (cf. Steele et al., 2008). Put differently, openness can be insistent vigilance and a need to be open in order to control events and people in one’s world (i.e., as in the case of unresolved/disorganized attachment) or calm reflective curiosity (i.e., as in the case of security). Although the FFM instrument employed in the present study does not permit facet-level analyses, post-hoc correlations between the attachment dimensions and items on the openness factor were to some degree in agreement with these speculations, suggesting that this is not a case of equifinality. Briefly, unresolved/disorganized adult attachment was significantly linked to the items Deep and Complicated, and marginally linked to Philosophical, whereas secure childhood attachment was close to significantly linked only to Imaginative. Attachment & Human Development 233 As noted, unresolved/disorganized attachment was concurrently positively related to openness in young adulthood also when the stability in openness from middle childhood had been controlled. This unique contribution suggests that unresolved/disorganized attachment in early adulthood is more important for openness in young adulthood than is attachment security in middle childhood. This could naturally be a result of their temporal proximity, but may also suggest a particular influence of unresolved/disorganized attachment on aspects of openness. Unresolved/disorganized attachment was also negatively related to conscientiousness in young adulthood, whereas attachment security was not related to conscientiousness. Unresolved/disorganized attachment is conceptually characterized by an incapability to congruently regulate frightening cognitive-affective material, related to loss or trauma (Hesse & Main, 2006). If this is generalized to the organization of more purely cognitive material is unclear. Liotti (1992) has suggested that due to the exposure of traumatizing experiences with parents, children with disorganized attachment are more vulnerable for dissociation. Dissociative tendencies are also apparent in unresolved/disorganized adults, for example by lapses in the monitoring of reasoning in the AAI, and frightening/frightened parental behaviors toward offspring (see Hesse & Main, 2006). Hesse and Main (2006) have suggested that memories of the overwhelming experiences associated with unresolved/disorganized status might intrude on attentional processes, leading to decreased functioning of the working memory. As working memory is involved in organizing and planning of a broad range of behaviors and actions (Baddely, 2007), this might substantiate why unresolved/disorganized attachment was found to be negatively related to conscientiousness. We expected secure attachment to be related to a higher degree of agreeableness, but the only association between the two in the present study was a marginally significant bivariate link between attachment security in childhood and agreeableness in young adulthood. As attachment security is related to higher quality of children’s friendships in terms of responsiveness, companionship, and conflict (Kerns, Klepac, & Cole, 2006; Lieberman, Doyle, & Markiewicz, 1999), it may be surprising that they were not related to a greater extent. Also, in the adult attachment style literature, attachment security has been associated with agreeableness (Noftle & Shaver, 2006). However, attachment was not related to agreeableness in Roisman and colleagues’ study (2007) nor in our previous study (Hagekull & Bohlin, 2003). One reason for this lack of a clear association in the present study may be a restriction in range in our measures of agreeableness (see Table 1). In contradiction to our expectations, attachment security was not related to neuroticism. This lack of an association is in accordance with the study of Roisman and colleagues (2007), but not with our previous study (Hagekull & Bohlin, 2003), nor with the adult attachment style literature (Noftle & Shaver, 2006). One reason why this association could not be established in the present study could be that the attachment security variables we used combined preoccupied and dismissing attachment on the insecurity pole. Dismissing attachment has been associated with suspiciously low self-reports of psychopathology and feelings of vulnerability (Kobak & Sceery, 1988; Pianta, Egeland, & Adam, 1996), and preoccupied attachment with high levels of distress (Pianta et al., 1996). This is in accordance with Roisman and colleagues’ study (2007), where deactivation (cf. dismissing 234 M. Fransson et al. attachment) was associated with lower levels of neuroticism and hyperactivation (cf. preoccupied attachment) with higher levels. Methodological considerations and future studies This study had notable methodological strengths. It is the first study of which we are aware that has studied the relations between attachment and the Five-Factor Model of personality with a prospective, longitudinal design, which also allowed controlling for cross-temporal stability in personality. Additionally, we know of no previous study that has investigated the relation between unresolved/disorganized adult attachment and the FFM. Secondly, although this was not the focus of our study, we emphasize the stability and interjudge agreement of the FFM observed here. Three factors (i.e., extraversion, openness, and neuroticism) were significantly, or close to significantly, stable over the 13-year time span. None of these correlations were more than moderate in magnitude, but keeping in mind that there were different raters and instruments at the two age periods, these relations contribute to the literature showing that personality dimensions tend to be stable across major developmental periods (e.g., Roberts & DelVecchio, 2000). In particular, the stability of openness may be surprising, as this factor has been difficult to obtain in previous child studies (Lamb et al., 2002). Regarding interjudge agreement, although peer ratings could not be obtained for all participants in young adulthood (46 out of 66 participants had peer ratings), most FFM factors displayed significant interjudge (i.e., self vs. peer) agreement, showing that the self-ratings of personality did not just reflect self-report biases or impression management. This study also had some limitations. First, the Seattle version of the SAT (Slough & Greenberg, 1990; Slough et al., 1988) does not provide a measure of disorganized attachment. Therefore, we could not investigate links between disorganization in childhood and the FFM at either time point. Second, the attachment variables were not related over time, which challenges the notion of continuity in attachment. One reason for this discontinuity could be that the attachment variables used in our study reflected different aspects of attachment at the two time points, preventing the relation from being optimally investigated: the middle childhood measure incorporates both signs of attachment security and disorganization, whereas these dimensions were separated into two attachment variables in young adulthood. Relatedly, although there is some empirical support for attachment continuity from infancy (using the strange situation) as well as from the kindergarten period (using the 6th-year reunion procedure [Main & Cassidy, 1988] or Kaplan’s [1987] SAT ) to young adulthood (e.g., Fraley, 2002; Main, Hesse, & Kaplan 2005), we are aware of no study that has documented a link to adulthood based on attachment assessments in middle childhood, nor of any study that has documented such a link based on the Seattle version of the SAT. Finally, lack of continuity may of course also reflect genuine discontinuity, some of which might well be lawfully related to attachment at either time-point (e.g., through intervening lifeevents). Future studies aiming to test developmental interlinkages between attachment and the FFM should include a measure of disorganized attachment in childhood, as it might add interesting information regarding trajectories to future personality. To avoid the risk of obscuring a true relation between attachment security and the five Attachment & Human Development 235 factors of personality through combining dismissing and preoccupied attachment on the insecurity pole, the AAI categories or the AAI Q-Set (Kobak, 1993) may be used in future studies. As attachment can be considered a one-shot measure of previous interactions, measures that more directly assess parental behaviors, as sensitivity and discipline, should also be included to more fully capture relational processes behind personality development. Finally, in order to investigate the associations between attachment and the facets of the FFM, we suggest the use of NEO-PI-R (Costa & McCrae, 1992) in future studies. Concluding remarks The present study was an attempt to unify contemporary research on personality, as guided by the FFM, and the developmental literature on attachment. Overall, the results suggest that attachment considerations may partially contribute to the understanding of personality development. In particular, attachment security in middle childhood and unresolved/disorganized status in adulthood were associated with personality, especially with extraversion and openness. Attachment security in young adulthood was, on the other hand, less predictive of personality than might be expected. One discussion often raised in the attachment literature is how widely or narrowly to conceive of the attachment construct. We argue that the relation between attachment and personality can be understood in terms of relating to others, exploring the surroundings, and regulating emotions and affects. These concepts are all attachment-related, but can also be considered as inherent aspects of the FFM. Shiner and Caspi (2003) have suggested that personality researchers should focus more on developmental processes behind personality development than on the constructs (e.g., ‘‘traits’’) per se. Based in part on the results presented here, we conjecture that personality in adulthood is partially influenced by nurture, as represented by attachment. However, it is beyond the reach of this study to disentangle more precisely how genes and environments jointly influence personality development. We encourage other researchers to fill this important knowledge gap. Note 1. It is beyond the scope of this paper to resolve these trait-related controversies. In our understanding, ‘‘traits’’ are merely convenient descriptive summaries of individual (and thus individual difference) characteristics. To avoid confusion in view of the more ambitious explanatory connotations that may be attributed to the trait construct, we prefer to use the purely descriptive terms personality ‘‘factors’’ and ‘‘dimensions’’. Acknowledgements We are grateful to Prof. Anders G. Broberg for assistance with AAI coding. The research reported in this article was supported by grants to Professors Bohlin and Hagekull from The Swedish Research Council and The Bank of Sweden Tercentenary Foundation. References Ainsworth, M.D.S., & Bowlby, J. (1991). An ethological approach to personality development. American Psychologist, 46, 331–341. Ainsworth, M.D.S., Blehar, M.C., Waters, E., & Wall, S. (1978). Patterns of attachment: A psychological study of the Strange Situation. Hillsdale, NJ: Erlbaum. 236 M. Fransson et al. Baddely, A.D. (2007). Working memory, thought, and action. Oxford: Oxford University Press. Belsky, J., Garduque, L., & Hrncir, E. (1984). Assessing performance, competence, and executive capacity in infant play: Relations to home environment and security of attachment. Developmental Psychology, 20, 406–417. Block, J. (1995). A contrarian view of the five-factor approach to personality description. Psychological Bulletin, 17, 187–215. Bohlin, G., Hagekull, B., & Rydell, A.-M. (2000). Attachment and social functioning: A longitudinal study from infancy to middle childhood. Social Development, 9, 24–39. Bowlby, J. (1969/1982). Attachment and loss: Vol. 1. Attachment. New York, NY: Basic Books. Bowlby, J. (1973). Attachment and loss: Vol. 2. Separation: Anxiety and anger. New York, NY: Basic Books. Bouchard, T.J., & Loehlin, J.C. (2001). Genes, evolution, and personality. Behavior Genetics, 31, 243–273. Broberg, A., Ivarsson, T., & Hinde, M. (1996). Anknytningsintervjun: En metod för att studera vuxnas mentala representationer av sina barndomserfarenheter [The attachment interview: A method to study adults’ mental representations of their childhood experiences]. Göteborgs Universitet: Rapport från Psykologiska Institutionen. Carlson, E.A. (1998). A prospective longitudinal study of attachment disorganization/ disorientation. Child Development, 69, 1107–1128. Carlson, V., Cichetti, D., Barnett, D., & Braunwald, K. (1989). Disorganized/disoriented attachment in maltreated infants. Developmental Psychology, 25, 525–531. Caspi, A., Roberts, B.W., & Shiner, R.L. (2005). Personality development: Stability and change. Annual Review of Psychology, 56, 453–484. Cassidy, J. (1994). Emotion regulation: Influences of attachment relationships. Monographs of the Society for Research in Child Development, 59(2/3), 228–249. Connelly, B.S., & Ones, D.S. (2010). An other perspective on personality: Meta-analytic integration of observers’ accuracy and predictive validity. Psychological Bulletin, 136, 1092–1122. Costa, P.T., Jr., & McCrae, R.R. (1985). The NEO Personality Inventory: Manual. Odessa, FL: Psychological Assessment Resources. Costa, P.T., Jr., & McCrae, R.R. (1992). NEO PI-R. Professional Manual. Odessa, FL: Psychological Assessment Resources. De Wolff, M.S., & van IJzendoorn, M.H. (1997). Sensitivity and attachment: A meta-analysis on parental antecedents of infant attachment. Child Development, 68, 571–591. Digman, J.M. (1990). Personality structure: Emergence of the Five-Factor Model. Annual Review of Psychology, 41, 417–440. Eysenck, H.J. (1991). Dimensions of personality: 16, 5, or 3? Criteria for a taxonomic paradigm. Personality and Individual Differences, 12, 773–790. Fearon, R.P., Bakermans-Kranenburg, M.J., van IJzendoorn, M.H., Lapsley, A.-M., & Roisman, G.I. (2010). The significance of insecure attachment and disorganization in the development of children’s externalizing behavior: A meta-analytic study. Child Development, 81, 435–456. Fraley, R.C. (2002). Attachment stability from infancy to adulthood: Meta-analysis and dynamic modeling of developmental mechanisms. Personality and Social Psychology Review 6, 123–151. Fraley, R.C., & Shaver, P.R. (2008). Attachment theory and its place in contemporary personality theory and research. In O.P. John, R.W. Robins, & L.A. Pervin (Eds.), Handbook of personality: Th...
Purchase answer to see full attachment