Exceptional Children. Vol. 55, No.3. pp. 266-272. © 1988 The Council for Exceptional Children.

Relationship Between Personality Adjustment and High Intelligence: Terman versus Hollingworth INGRID N. GROSSBERG DEWEY G. CORNELL

ABSTRACT: Work by Terman helped refute the negative stereotype that highly inlel/igelll children were maladjusted, but Hollingworth contended that children in the very highest lQ groups were prone to maladjustment problems. The present study examines whether the rela/ionshp between intelligence and personality adjustment is negative or positive within the gifted range. Subjects are 83 7- to l l-vear-old children. with IQs between 120 ami 168, who were ellrolled in either (I public or private school gijied program. Results with both parent- and child-report instruments (Personality Inventory Fir Children, Coopersmith Self-Esteem lnventorv, Revised Children's Mcnifest A/uiety Scale) support the view that intelligence is positively related to adjustment.

o

Despite over 50 years of published research, the relationship between personality adjustment and high intelligence continues to be a topic of controversy (Janos & Robinson, 1985). Early views (e.g. Lornbroso, 1891) held that high intelligence was associated with insanity or a propensity for adjustment problems, This negative stereotype was largely refuted by Lewis Terman's longittidinal studies (Terman et aI., 1925-1959). Terman and colleagues demonstrated convincingly that highly intelligent children, defined by Stanford-Binet IQs greater than 140, tended to be better adjusted than the norm on a wide range of adjustment variables. Research with other samples has supported this view (Kelly & Colangelo, J 984; Lehman & Erdwins, J 981; Reynolds & Bradley, J983). Concern over the adjustment of gifted-level children, however, has not abated, Even at the time of Terman's landmark work, another respected

INGRID N. GROSSBERG is Instructor, School Social Work, Wayne State University, Detroit, Michigan. DEWEY G. CORNELL is Assistant Professor, School of Education, University ()f Virginia, Charlouesville.

266

authority in the field, Leta Hollingworth (1942), contended that highly intelligent children were prone to develop social and emotional adjustment problems. Similar concerns have been repeated by many others (Austin & Draper, 1981; Janos & Robinson, 1985; Lajoie & Shore, 1981; Powell & Haden, 1984; Roedell, J984; Schauer, 1976). Thus, it appears that two distinct, seemingly antithetical, views exist regarding the adjustment of highly intelligent children. Research in the Terman tradition generally indicates that high intelligence is associated with healthy adjustment. What might be termed the Hollingworth perspective regards high intelligence as associated with adjustment problems (Hollingworth, 1942). (Note: The point of this article is not to set the personal views of Terman and Hollingworth at opposite ends of a continuum, creating a strawman argument. Rather, the influential contributions and status of both authorities make their names convenient guideposts for identifying-and legitimizing-two contrasting perspectives on the relationship between adjustment and IQ.) Taken to its logical conclusion, one view leads to the hypothesis that there is a positive correlation between intelligence and adjustment. In contrast, the November 1988

other view suggests that the correlation would be

negative. One possible explanation is that the relationship between IQ and adjustment is curvilinear, changing from positive to negative at some point within the gifted range. However, determination of the point at which IQ becomes a liability rather than an asset to healthy adjustment is difficult. In the space of a few pages, Hollingworth (1942, pp. 264-265) mentioned possible cutoff scores of 170, 160, and 150. It should be noted that Hollingworth's scores were ratio IQs based primarily on the 1916 Stanford-Binet. Ratio and deviation IQs are not equivalent. For example, according to the manual for the third revision of the Stanford-Binet (Terman & Merrill, 1973), a ratio IQ of 180 for an l l-year-old would correspond to a deviation IQ of 171, and a ratio IQ of 150 would be comparable to a deviation IQ of 144 (using Pinneau norms). A number of studies have claimed poor adjustment in very high IQ children, but these studies were based on uncertain adjustment criteria and typically did not include appropriate comparison groups (Kincaid, 1969; Selig, 1951; Zorbaugh, Boardman, & Sheldon, 1951). Studies that contrasted higher and lower IQ groups directly suggest a different view. Gallagher and Crowder (1957) found that 35 children with 150+ IQs were generally well adjusted, despite a few exceptions. Gallagher (1958) found that 150+ IQ (Stanford-Binet) children were among the most popular children in their respective classes and found no differences between subgroups of 150-164-IQ and 165-205-IQ children. Lewis (1943), however, did report greater maladjustment and underachievement in a higher (145 +) lQ group than in a lower (125-144) IQ group. Freeman (1979) contrasted a "High IQ" group (IQs of 141-170, mean 155) with a "Moderate IQ" group UQs of97-140, mean 120) on both parent- and child-report measures of adjustment. Results were generally favorable in both groups, and there was little indication that the High IQ group was less well adjusted. Janos (1983) compared 32 "highly gifted" children (IQs above 164) and 49 "moderately gifted" (IQs of 120-140) on several standard adjustment measures. On Achenbach and Edclbrock's Child Behavior Checklist, there were not significant differences between groups. Interestingly, even within the highly gifted group, higher IQs were associated with better adjustment on the Behavior Problems subscale (r = -.518). There were no differences between groups on the Connor's Teacher Rating scale, the Vineland Adaptive Behavior scales, Exceptional Children

or the Piers Harris Children's Self-Concept scale. However, Janos did conclude that a "significant minority (20-25%)" (p. 96) of highly gifted children suffered adjustment problems. Feldman (1984) reviewed follow-up data on the 26 subjects in Terman's project who scored above 180 IQ and 26 subjects with lower, but still gifted-level, IQs (mean 150), There were few differences between the two groups; and he concluded that exceptionally high IQs did not truly distinguish these individuals from other, more moderately high IQ subjects. In summary, there are persistent reports of adjustment problems among the highest IQ groups, but the evidence is contradictory and largely unpersuasive. Some very high IQ children with adjustment problems may be expected by chance without the existence of a specific relationship between IQ and adjustment. One problem is that the use of widely varying cutoff scores makes comparison across studies difficult. An alternative approach would be to investigate the direction and magnitude of correlations between IQ and adjustment within the gifted range, eliminating the need for an arbitrary cutoff. A second problem is that most studies consider the relationship between IQ and adjustment in isolation and do not control for possible confounding variables that could obscure the relationship. Two kinds of confounding variables are considered in the present study: respondent bias or defensiveness in reporting adjustment problems, and overall family adjustment. In previous work Cornell and Grossberg (1986) found that controlling for respondent bias through validity scales associated with the adjustment instrument altered the pattern of results in a meaningful way. Cornell and Grossberg (1987) also found that family adjustment, as assessed by the family Cohesion scale of the Family Environment Scale (Moos & Moos, 1981), was consistently related to both parent and child measures of child adjustment.

METHOD

Subjects Parents of 7- to I l-year-old children enrolled in either a public or private school gifted program were contacted by letter inviting them to participate in the study. Of 153 families contacted through the private school, 51 (33%) agreed to participate, while 32 of 93 (34%) families contacted through the public school agreed to participate in the study. The final sample consisted of 83 children.

267

All 83 children received individual IQ testing with either the Stanford-Binet (65 children) or the Wechsler Intelligence Scale for Children-Revised (WISC-R; 18 children). Mean IQ for the entire sample was 139.95 (SD, I l. 3). There was no significant difference between children tested with the Stanford-Binet and the WISC-R. The mean IQ for public school children (143.3) was significantly higher than that for the private school children (137.7), t = 2.29, p ( .05. This mean IQ difference appears to be attributable to a difference in selection criteria for the gifted programs by the two schools. The policy for the public school was not to include children with IQs below 130, while the private school admitted children with IQs as low as 120, provided that there was evidence of exceptional talent or ability in some area (e.g., creative writing). In preliminary analyses, there were no differences between public and private school children in any of the 12 adjustment measures. Simple correlations were calculated between IQ and each adjustment measure separately for public and private school children. When correlations were compared by r to z transformation, there were no significant differences between groups. For these reasons, public and private school children were combined in the data analyses to follow. The families were predominantly white, middleclass families residing in suburban Southeast Michigan. On the 7-point Hollingshead scale for educational attainment (Hollingshead, 1975), mothers obtained a mean score of 5.77 and fathers a mean score of 6.16. Additional demographic data and methodological information are reported elsewhere (Grossberg, 1985).

Instruments and Procedures The Revised Personality Inventory for Children, Short Form (PIC) (WIRT, Lachar, Klinedinst, & Seat, 1977) is a 280-item true-false questionnaire completed by the child's parent. The PIC scales selected for use in this study are the Adjustment Scale, a screening measure of overall psychological adjustment, and 9 of the 12 standard clinical scales (see Table I). In addition to ~cales assessing the child's adjustment, the prc contains three validity scales, which are designed as measures of distortion or bias. The Lie scale attempts to identify parents with a defensive response set that involves describing the child in extremely virtuous terms and denying minor, commonly occurring problems. The Defensiveness

268

scale assesses the parent's tendency to minimize the child's problems. The F scale is constructed to identify exaggeration of problems or random responding. The Coopersmith Self-Esteem Inventory School Form (Self-Esteem) (Coopersmith, 1981) is a 58-item like-me/unlike-me questionnaire completed by the child. It provides an overall measure of Self-Esteem as a well as a Lie scale score. The Revised Children's Manifest Anxiety Scale (Anxiety) (Reynolds & Paget, 1983) is a 37-item yes/no questionnaire also completed by the child. It provides an overall Anxiety score and a Lie scale score. The Family Environment Scale (FES) (Moos & Moos, 1981) is a 90-item true/false questionnaire completed by one or more family members. Only the Cohesion scale was selected for use in the present analyses. Cohesion is defined as "the degree of commitment, help, and support family members provide for one another" (Moos & Moos, 1981, p. 2). Parents (primarily mothers) competed the PIC, FES, and a family background questionnaire in group meetings held at school. They were given explicit instructions on having their children complete the Self-Esteem and Anxiety measures independently at home and return them in stamped, addressed envelopes provided for them.

RESULTS A series of simple correlations were calculated between IQ and each of the adjustment meas.ures (see Table I). Eleven of 12 correlations for the full sample of 83 children (IQ range 120-168) were in the direction of more favorable adjustment for higher IQ children, but only 3 were statistically significant. One possibility was that there were significant negative correlations between IQ and adjustment in the higher IQ ranges which were masked by correlating across the full 120-168-IQ range. As a result, additional correlations were calculated for successively smaller IQ ranges, omitting lower IQ groups. The second set of correlations was for the 65 children falling into the 130-168 range, and the third set was for 21 children in the 145-168-IQ range. None of these correlations were significant. Next, multiple-regression analyses on the relationship between IQ and adjustment were calculated after controlling for the effects of respondent bias (validity scales for each measure) and family Cohesion. Using the MIDAS (Fox & Guire, 1976) REGRESSION command, the appropriate validity scales and Cohesion were entered into the analyses

November 1988

as "fixed" variables, followed by IQ as a "free" variable. All 12 multiple-regression analyses were significant, with multiple rs ranging from .43 to .78 (see Table 2). For each adjustment measure, one or more of the validity scales were significant correlates, and for four adjustment measures, Cohesion also was a significant correlate. After controlling for these effects, IQ was partially correlated with four of the adjustment measures beyond the .05 level, and with three more of the adjustment measures beyond the .10 level.

DISCUSSION The present study finds modest support for the view that IQ is positively related to healthy personality adjustment within the gifted range. Concerns raised by the Hollingworth view that very high IQ is associated with poor adjustment were not supported. Children with higher IQs tended to be less anxious and nervous (Anxious, child report). According to parent report on the PIC, they were less likely to evidence .problems in physical or cognitive development (Development scale) or exhibit behavior and discipline problems (Delinquency scale). The IQ-adjustment relationship was reexamined after controlling for the influence of both respondent bias and family Cohesion through multiple-regression analyses. As reported in Table 2, this combination of predictor variables (validity scales and Cohesion entered together, followed by IQ) was significantly related to each of the 12 adjustment measures beyond the .01 level. Multiple correlations ranged from .43 to .78, indicating that these predictors account for 18.5% to 60.8% of the variance in adjustment for this sample. This procedure resulted in somewhat stronger support for the Terman hypothesis, with the addition of a significant partial correlation between IQ and PIC Adjustment.

Limitations of the Present Study Several qualifications to the present findings are in order. The sample consists of children 7 to II years old and therefore the results do not exclude the possiblity of high-lQ-related problems at other ages. Also, the 83 children ranged from 120 to l68 IQ, including 4 children scoring at the ceiling for the WISC-R or Stanford Binet norm tables. It is possible that the serious adjustment problems hypothesized to be associated with high IQ in fact only occur in children with IQs beyond the norm tables (essentially 4 or more standard deviations from the mean). Scores at this level, however, should be regarded with great

Exceptional Children

TABLE 1 Correlations Between IQ and Adjustment Correlation Coefficients by IQ Range Adjustment Measure

120-/68 /30-168 /68 (n = 83) (n = 65) (n =21)

Self-Esteem (self-report) .15 Anxiety (self-report) -.23* Adjustment (PIC) -.16 Development (PIC) -.26* Somatic Concern (PIC) .01 Depression (PIC) .14 -.22* Delinquency (PIC) Withdrawal (PIC) -.10 Anxiety (PIC) -.16 -.11 Psychosis (PIC) -.13 Hyperactivity (PIC) -.07 Social Skills (PIC)

.11

-.13

-.11

-.10

-.17

.07 .13

-.12 .11 -.07 -.22 -.07 -.01

-.06 -.17 -.14

.23

.02 -.20 -.08 .34 -.22 .16 .07

Note. PIC = Personality Inventory for Children. *p < .05.

caution. Standard lQ tests are not designed to discriminate at such a high level, and these scores may be unreliable and of questionable meaning. Moreover, it should be noted that, in principle, only a fraction of the gifted population should fall beyond the range included in this study. If one takes an IQ score of 2 standard deviations above the mean as an arbitrary cutoff, then in accord with the normal distribution, about 2.27% of the population falls into the gifted range. Those 4 or more standard deviations above the mean would make up only .00003171.0227 or 0.14% of those in the gifted IQ range. Another limitation to the present study is that all subjects were already enrolled in gifted programs. To the extent that adjustment problems arise from the high-TQ child's difficulties in coping in a regular classroom, these problems would be ameliorated in this sample. Only further work comparing giftedlevel children in regular and gifted programs (or before and after placement in a gifted program) can adequately address this question. There is reason to believe, however, that placement in a gifted program is not entirely facilitative of the child's self-esteem. Studies by both Rodgers (1979) and Coleman and Fults (1982) reported a decline in self-concept scores following placement in a gifted program.

269

TABLE 2 Regression Analyses to Predict Adjustment From IQ After Controlling for Response Bias and Family Cohesion

Adjustment Measure, Predictors

Self-Esteem (self-report) Lie Scale Cohesion IQ Anxiety (self-report) Lie Scale Cohesion lQ Adjustment (PIC) Lie Scale F Scale Defensiveness Cohesion IQ Development (PIC) Lie Scale F Scale Defensiveness Cohesion IQ Somatic Concern (PIC) Lie Scale F Scale Defensiveness Cohesion IQ Depression (PIC) Lie Scale F Scale Defensiveness Cohesion IQ Delinquency (PIC) Lie Scale F Scale Defensiveness Cohesion IQ

Partial r Correlation

t Value

.30 .34 .18

2.76*** 3.18*** 2.65

-.22 -.34 -.26

2.00** 3.17*** 2.41 * *

-.43 .45 -.12 -.32 -.27

4.14*** 4.46*** 1.07 2.94*** 2.44*"

.01 .43 .00 -.18 -.30

0.05 4.23**"' 0.03 1.62 2.76*" '"

-.17 .53 .04 -.04 -.00

1.49 5.47*** 0.37 0.37 0.03

-.26 .55 -.23 -.03 -.21

2.38*' 5.72*** 2.06** 0.29 1.85*

-.10 .37 -.06 -.36 -.28

0.90 3.47*** 0.51 3.42*** 2.51 * *

Multiple r Correlation

.48

7.68***

.48

7.66***

.74

19.01 '**

.56

6.94**"

.59

8.24***

.69

14.11*"'*

.62

9.53***

Continued

Why Do High-IQ Children Suffer Adjustment Problems? Although high IQ appears to be associated with better adjustment, it clearly does not render the child invulnerable to adjustment problems. Several hypotheses concerning the reasons why high-IQ children suffer adjustment problems can be considered. The most parsimonious explanation is that high-IQ children develop adjustment problems for 270

Overall F Value

Oil

next page.

reasons (social, familial, intrapsychic) no different than do other children. The child's high IQ is such a remarkable attribute that it may be falsely perceived as associated with any adjustment problems the child develops. The "contrast effect" of high IQ coupled with maladjustment has an ironic, compelling quality that draws attention and may lead to superstitious reasoning about a causal connection between them. Society's ambivalent attitudes toward high-IQ individuals also may foster a willingness to cast November 1988

TABLE 2 Continued

Adjustment Measure, Predictors

Withdrawal (PIC) Lie Scale F Scale Defensiveness Cohesion IQ Anxiety (PIC) Lie Scale F Scale Defensiveness Cohesion 10 Psychosis (PIC) Lie Scale F Scale Defensiveness Cohesion 10 Hyperactivity (PIC) Lie Scale F Scale Defensiveness Cohesion IQ Social Skills (PIC) Lie Scale F Scale Defensiveness Cohesion IQ

Partial r Correlation

.06 .42 -.23 .03

t Value

-.11

0.53 4.05*** 2.09* * 0.29 1.0 I

-.20 .49 -,12 -.03 -.21

1.77' 4.92**' 1.07 0.28 1.92*

-.28 .68 -.22 -.05 -.19

2.52 8.14**'" 2.02** 0.45 1.73*

-.33 .00 -.03 -.09 -.16

3.06** 0.09 0.28 0.81 1.46

-.28 .38 -.23 .06 -.11

2.53** 3.62*** 2.03** 0.53 0.97

Multiple r Correlation

Overall FValue

.51

5.28*'*

.61

9.20***

.78

23.30***

.43

3.45***

.58

7.73***

Note IQ was entered into regression analyses after all other variables, which were entered simultaneously. PIC = Personality Inventory for Children. "p < .10. **p < .05. ***p < .01.

giftedness in a negative light (Cornell, 1984). Finally, it may be that a troubled child would emphasize his or her talents defensively-for example, affecting an overintellectualized, arrogant manner-thereby creating a link between maladjustment and intelligence that is secondary, not primary. An alternative explanation is that high intelligence does place some children at risk for adjustment problems. Here again, however, it is not necessry to assume that intelligence per se directly fosters maladjustment. Family members, teachers, peers, or others may respond to the child's intelligence with rejection, jealousy, fear, or other negative attitudes. The absence of an appropriate school program or lack of intellectual peers also may have adverse effects. Not high intelligence, but rather the consequences

Exceptional Children

of high intelligence in some social environments, may have a negative effect on personality adjustment. These factors need to be studied systematically and ruled out before accepting the hypothesis that high intelligence is a direct, primary source of maladjustment.

REFERENCES Austin, A. B., & Draper, D. C. (1981). Peer relationships of the academically gifted: A review. Gifted Child Quarterly, 25. 129-133. Coleman, J. M., & Fults, B. A. (1982). Self concept and the gifted classroom: The role of social comparisons. Gijied Child Quarterly, 26(3), 116-120, Coopersmith, S. (1981). Self-esteem inventories. Palo Alto, CA: Consulting Psychologists Press, Inc.

271

Cornell, D. G. (1984). Families or girted children. Ann Arbor, Ml: UMI Research Press. Cornell, D. G., & Grossberg, L N. (1986). Siblings of children in gifted programs. Journal jlJr the Education of the Gifted. 9(4), 253,264. Cornell, D. G, & Grossberg, L N. (1987). Family environment and personality adjustment in gifted program children. Gijied Child Quarterly, 31(2), 59-64. Feldman, D. H. (1984) A follow-up of subjects scoring above 180 IQ in Terman's 'Genetic Studies of Genius.' Exceptional Children, 50, 518-523.

& Guire, K, E. (1976) Documentation for MIDAS (3rd ed.) Ann Arbor, MI: Statistical Research Laboratory, University of Michigan.

Fox, D, J

Freeman, J. (1979). Gifted children: Lancaster, England: MTP Press Limited. Gallagher, J. J. (1958) Peer acceptance of highly gifted children in elementary school. The Elementary School Journal, 58, 465-470. Gallagher, J: J., & Crowder, T (1957). The adjustment of gifted children in the regular classroom. exceptional Children, 23, 306-312, 317-319. Grossberg, I. N. (1985), Relationships be/ween lQ and emotional adjustment in gifted, latency-age children. Unpublished doctoral dissertation. Department of Counseling (Education), Wayne Stale University. Hollingshead, A B. (1975). FOllr [actor index (~r social status. Unpublished manuscript. Yale University. Hollingworth, L. S. (1942). Childrell above 180 lQ, Stanford-Binet: Origin and development, Yonkers, NY: World Book. Janos, P. (1983). The psychosocial adjustment of children of very superior intellectual ability. Unpublished doctoral dissertation, Ohio State University. Janos, P. M., & Robinson, N. M, (1985). Psychosocial development in intellectually gifted children. In Frances D. Horowitz & Marion O'Brien (Eds.), The gijied and talented: Drvelopmentot perspectives. Washington, DC: American Psychological Association, Kelly, K. R" & Colangelo, N. (1984). Academic and social self-concepts of gifted, general and special students, Exceptional Children, 50(6), 551,554, Kincaid, D. (1969). A study of highly gifted elementary pupils, Gifted Child Quarterlv, 13(4), 264,267, Lajoie, S, P" & Shore, B, M. (1981), Three myths? The overrepresentation of the gifted among dropouts, delinquents, and suicides. Gifted Child Quarterly, 25, 136,i43. Lehman, E, B., & Erdwins, C. J. (1981). The social and emotional adjustment of young intellectually gifted children. Gifted Child Quarterly, 25, 134-137. Lewis, W. D. (1943). Some charactcrisitics of very superior children. Journal of Genetic Psychology, 62, 301-309, Lombroso, C. (1891). The man oj genius. London: Scott. Moos, R. H" & Moos, B. S. (1981) Family Environment Scale manual. Palo Alto, CA: Consulting Psychologists

Powell, PM" & Haden, T (1984), The intellectual and psychosocial nature of extreme giftedness, Roeper Reveiw, 6(3),131-136. Reynolds, C. R., & Bradley, M. (1983, Spring), Emotional stability of intellectually superior children versus nongil'ted peers as estimated by chronic arix iety levels, School Psychology Review, 12,(2), Spring, 190,194 Reynolds, C. R" & Paget, K. D. (1983). National normative and reliability data for the revised Children's Manifest Anxiety Scale. School Psychology Review, 12(3),324-336. Rodgers, B, S. (1979). Effects or an enrichmen/ program screening process on the selj-concept and other-concept

of gifted elementary children. Unpublished doctoral dissertation. University of Cincinnati Roedell, W, C. (1984). Vulnerabilities of highly gifted cbildren. Roeper Review, 6(3), 127-130. Schauer, G, H, (1976, Winter). Emotional disturbance and giftedness. Gifted Child Quarterly, 20(4),470-477, Selig, K. (1951), Personality structure as revelli2d by /he Rorschach Technique on a grtJliP o{ children wIVJ!e:I'/ (1/ or above 170 IQ on the 193,7 revision of the Stlli/FtdBinet (Vol, I-V). Unpublished doctoral dissertation. New York University. Terman, L. M. (Ed.). (1925,1959), Genetic studies of genius, Vol. I-V, Stanford, CA: Stanford University Press. Terman, L. M., & Merrill, M. A. (1973). Stanford-Binet Intelligence Scale: Manual for rile thirt] revision, form L-M (1972 norms edition). Boston: Houghton Mifflin, Wirt, R. D., Lacher, D., Kiinedirst, J. K., & Scat, P D. (1977). Multidimensional description of child personality: A manual for The Personality lnveniorvfcn: Children, Los Angeles: Western Psychological Services, Zorbaugh, H., Boardman, R., & Sheldon, P. (195l). Some observations on highly gifted children, In P, Willy (Ed.), The gified child, Lexington, MA: Heath, Manuscript received November 1986; revision accepted April 1987.

INDEX TO ADVERTISERS American Guidance Service, cover 2, p. 193 American Printing House for the Blind, p. 279 Charles C Thomas, Publisher, p. 221 Council for Exceptional Children, pp. 202, 239, 248, 265, cover 3 DLM Teaching Resources, cover 4 Foundation for Exceptional Children, p. 277 Lake Grove, p. 214 McCarron-Dial Systems, p. 229 National Scholarship Quest, Inc., p, 285 School District of Independence, MO, p. 279 Vocational Research Institute, p, 198

Press,

272

November 1988