Journal of Gerontology 1976, Vol. 31, No. 6, 663-669
Relations of Age and Personality Dimensions to Cognitive Ability Factors1 The relation between three cognitive ability factors — Information Processing Ability (IPA), Manual Dexterity (MD), and Pattern Analysis Capability (PAC) — and three personality dimensions — Anxiety, Extroversion, and Openness to Experience — were examined in three age groups. Subjects were 969 male volunteers ranging in age from 25 to 82. Subjects high in anxiety scored lower on all three cognitive factors; subjects open to experience scored higher on IPA and PAC; and introverted subjects scored higher on PAC. Most of these effects remained when the education and socio-economic status were held constant in covariance analyses. Older subjects performed less well than younger ones on MD and PAC, but not on IPA. While personality has some influence on cognitive performance, the declines with age in performance on some cognitive tasks are not mediated by personality.
EVELOPMENTAL psychologists have D a long-standing interest in the relations among measures of personality and ability. On the one hand, intellectual ability may contribute to the development of particular patterns of personality; on the other, motivational factors and personality traits may influence performance on tasks which require problem solving and manipulation of verbal and nonverbal symbols. Except for research in which individual differences in personality are defined in terms of performance on cognitive or perceptual tasks (e.g., Bruner & Postman, 1949; Witkin, Lewis, Hertman, Machover, Meisser, & Wapner, 1954), studies on young adults have generally found little relation between measures of personality and abilities (Eysenck, 1971). The present study explores the possibility that, when the entire adult age range is represented, subtle influences of personality on cognitive performance may be more evident. In studies of age differences in abilities, individual variations in personality have most frequently been used to account for part of the ob'From the Veterans Administration Normative Aging Study. Portions of this paper were presented at the meeting of the Gerontological Society, October, 1975, Louisville. Research supported in part by the Medical Research Service of the Veterans Administration, The Council for Tobacco Research - U. S. A. Grant 1085, and by NIH Grant 00467. 'VA Outpatient Clinic, 17 Court St., Boston 02108. 'Also at Univ. of Massachusetts at Boston, Dorchester, MA 02125 'Also at Harvard Univ. School of Medicine. 'Also at Hellenic College, Brookline, MA.
served decline in cognitive performance associated with age. For example, it has been postulated that poorer performance of older persons on cognitive tasks is a result of anxiety generated by concern about their abilities and by unfamiliarity with standardized test procedures (e.g., Baltes & Labouvie, 1973; Fozard & Thomas, 1975). While anxiety may result in inhibited performance on intellectual tasks, there is no satisfactory evidence to demonstrate that anxiety affects the performance of older persons relatively more than that of younger ones. One purpose of the present study is to evaluate the hypothesis that there is an interaction between anxiety and age on measures of abilities. A second way in which personality might affect test performance is through the social setting of the test situation. The introverted, taskoriented individual might be more comfortable with the formal aspects of the testing situation than the extravert whose outgoing mode of social interaction could be less appropriate, thereby interfering with task performance. While available evidence provides no strong basis for such an hypothesis in young adults, the effects of extraversion on styles of social interaction are so pervasive that its relation to intellectual ability over the adult age range should be investigated, particularly because there is evidence that greater introversion is associated with older age (Ames, 1960; Brozek, 1955). 663
Downloaded from http://geronj.oxfordjournals.org/ at The Australian National University on January 22, 2015
Paul T. Costa, Jr., PhD, 2 3 James L. Fozard, PhD, 2 3 Robert R. McCrae, PhD3 and Raymond Bosse.'PhD2 5
664
COSTA, FOZARD, McCRAE, ANDBOSSE
METHOD
Measure of Abilities and Personality The three cognitive ability factors resulted from a principal axis factor analysis of the twelve subtests of the GATB. The varimax rotated factor loadings for the 12 subtests and the correlations of the factors with age, socioeconomic status (SES; Warner, Meeker, & Eells, 1960), and years of educations are shown in Table 1. This pattern of factor loadings has been replicated by Nuttall and Fozard (1971) using two other sets of published data, and by Bates, Coppinger, and Oldham (1974) using a sample of elderly individuals ranging in age from 75 to 96. The first factor, Information Processing Ability (IPA), measures the facility with which familiar verbal and numerical materials are used. The tests involve numerical computation, finding the synonyms and antonyms of words,
Table 1. Varimax Rotated Factor Loadings for 12 GATB Subtests and Correlations with Age, SES, and Education. Factor Loading h1
Variable
Information Processing Ability (IPA)
Manual Dexterity (MD)
Pattern Analysis Capability (PAC)
IPA
MD
PAC
Arithmetic reason
81
07
18
69
Computation
82
19
13
73
Vocabulary
69
07
27
55
Name Comparison
66
18
29
55
Dissassemble
13
68
28
56
Place
05
66
11
45
Turn
12
72
14
55
Assemble
07
51
23
32
Mark Making
41
52
06
44
Three Dimensional Space
26
19
69
58
Form Matching
23
31
67
60 51
Tool Matching
38
29
53
Age
-09
-44
-30
Warner Level SES
-44
00
-12
53
02
13
Correlates
Education
identifying differences in the spelling of names, and solving arithmetic problems. The tests are most similar to what Cattell calls crystallized intelligence (Ross, 1968), although the present tests put more emphasis on application of verbal and arithmetic skills than on accumulated capacity as measured by definitions of words or availability of information. Scores on the Information Processing Ability factor are highly related to differences in education and SES. The second factor, Manual Dexterity (MD), is composed of tests of finger and hand movements. It is most closely associated with age and is uncorrelated with variations in social class and education. In the present sample, manual dexterity performance is highly correlated with response speed on conventional choice reaction tasks (Fozard, Nuttall, & Waugh, 1972). Because all of the tests require repetitive aiming and transport movements, good performance requires one to sustain rhythm and organization of movement. Accordingly, manual dexterity represents a more complex set of skills than is usually defined by simple motor speed tests. The third factor, Pattern Analysis Capability (PAC), consists of two tests which require the
Downloaded from http://geronj.oxfordjournals.org/ at The Australian National University on January 22, 2015
A third possibility is that differences associated with receptiveness to new experience could affect test performances. The more open, imaginative individual may be expected to show more interest in intellectual matters in general and in the test situation in particular. This could result in higher test scores for these individuals at all ages. The differences may become more pronounced with age because the person who is sensitive to his environment and open to new ideas is likely to accumulate a greater variety of cognitive experiences over his lifetime than his practical and unimaginative age-peer. The differences should be most pronounced on measures dependent upon accumulated knowledge and abilities, characterized by Cattell (1963) as crystallized intelligence. These measures may covary to some extent with social class and educational experience. To explore the three possible relationships between personality and abilities described above, the present study examined the relationships among three measures of cognitive abilities derived from the General Aptitude Test Battery (GATB), a group of tests developed for occupational aptitude testing (US Dept. of Labor, 1967) and three personality dimensions derived from the Cattell Sixteen Personality Factor Questionnaire or 16PF (Cattell, Eber, & Tatsuoka, 1970).
AGE, PERSONALITY, AND ABILITY
Subjects and Procedure The subjects were 969 male volunteers, all participants in the Normative Aging Study (Bell, Rose, & Damon, 1972). These subjects are distinguished from the general population in that all were judged to be healthy at the time of their entry into the study between 1963 and 1968. In a single testing session, small groups of subjects were administered the GATB and the 16PF during a period from 1967 to 1968. Age
groups were formed on the basis of age at the time of psychological testing. The three age groups with sample size, mean and standard deviation of age are as follows: Young, N = 140, X = 31.7, SD = 2.28; Middle-aged, N = 711, X = 43.7, SD = 5.54; Old, N = 109, X = 60.3, SD = 5.24. Analyses Cluster dimension scores were calculated for each subject within each of the three age groups by summing the standardized scores of each of the component scales. Cognitive ability factor scores were least-square regression estimates of Varimax rotated factors. Four-way analyses of variance were performed for each of the cognitive factors, using age group, anxiety, extraversion, and openness to experience as the classifying variables. The "classic experimental approach" option of the SPSS (Nie, Hull, Jenkins, Steinbrenner, & Bent, 1974) analysis of variance program was used to accommodate unequal cell frequencies. Personality groups were formed by dichotomizing individuals according to whether their personality scores were higher or lower than the mean.for their age group. In order to assess the influence of educational level and SES on the relations between personality and cognitive abilities, the four-way analyses described above were also performed using years of education and Warner Level as covariates. INFORMATION
PROCESSING ABILITY
MANUAL DEXTERITY .6 .2 -.2 -.6 -1.0
PATTERN ANALYSIS
.6 %
CAPABILITY
.2
1 -, 30
40
50
60
30
40
50
60
30
40
50
60
MEAN AGE OF CROUPS (years)
Fig. 1. Unadjusted deviation contrasts by age and personality on three ability factors.
Downloaded from http://geronj.oxfordjournals.org/ at The Australian National University on January 22, 2015
subject to locate similar forms and one in which he is to identify the three dimensional realizations of two dimensional drawings. Performance on these three is moderately correlated both with age and with education and social class. While there is no direct evidence, Pattern Analysis Capability probably measures an ability similar to what Cattell (1963) would refer to as Fluid Intelligence. A cluster analysis (Tryon & Bailey, 1970) of Forms A and B of the Cattell 16 PF (Costa & McCrae, 1976) yielded three cluster dimensions: Anxiety-Adjustment, IntroversionExtraversion, and Open-Closed Experiential Style. The anxiety cluster was defined by the factors of low emotional stability (C-), suspiciousness (L), guilt-proneness (0), low selfsentiment integration (Q3-), and ergic tension (Q4). It represents a coping dimension of adjustment-maladjustment, which includes trait anxiety, but also other maladjustive emotional responses. The extraversion cluster is composed of the factors warmth (A), surgency (F), adventurousness (H), and group dependence (Q2-). These two clusters resemble the first two second order factors repeatedly found in analyses of the 16PF (Cattell, 1973). The third cluster identified was age specific in structure. Around the core element of imaginativeness (Factor M) three different clusters emerged in three age groups. In the young group, imaginativeness (M), and tendermindness (I) defined an openness to feelings cluster. In the middle group, imaginativeness and radicalism (Ql) seemed to form an openness to ideas cluster. In the old group, both tendermindedness (I) and radicalism (QI) joined imaginativeness (M) as did intellectual brightness (B). This combination was identified as a balanced openness to both feelings and ideas. Across the three age groups, the third cluster can be interpreted as an open-closed experiential style dimension.
665
666
COSTA, FOZARD, McCRAE, ANDBOSSE
RESULTS
DISCUSSION
The hypotheses concerning the relations between personality and cognitive abilities were partially confirmed in this study. Results of the analysis of variance showed that anxious subjects consistently scored significantly lower than adjusted subjects on all three intelligence factors, and in all age groups. While anxiety may be related to intrinsic intellectual ability (or competence), it seems most probable that the adverse effects of anxiety on test performance accounts for the differences between
groups. Trait anxiety predisposes individuals to state anxiety, and state anxiety interferes with performance on these cognitive and perceptual motor tasks. With regard to Information Processing, the relation between cognitive performance and anxiety seems to be mediated by the environmental influences of education and SES. When these effects were statistically controlled, anxiety showed no relation to performance. This may help to explain why studies conducted on certain populations (where there is typically little variation in education and SES) have shown no relation between anxiety and intellectual ability (Eisdorfer, 1968). It was hypothesized that extraverted subjects might be at a disadvantage because of the formality of the testing situation. If this were so, they would be expected to perform uniformly less well on all three types of ability factors. In fact, extraverts show significantly lower scores only on the Pattern Analysis Capability factor. Pattern analysis seems to involve the comparison of perceptions with internal schemata, and this suggests that perhaps it is the introspective capacity of introverts rather than their social orientation which affects cognitive performance. It was also hypothesized that persons open to experience would show greater cognitive abilities, particularly in the area of information processing, where the cumulative effect of acquiring novel experience would be most beneficial. While open subjects did show higher scores on both IPA and PAC, the relation between openness and IPA appears to have been accounted for by the effects of education and SES. Greater amounts of education may lead to more tolerance for new experience, and also contribute to the development of information processing ability. On the other hand, openness to experience was found to be related to Pattern Analysis Capability even when levels of education and SES were controlled. Perhaps the metaphorical explanation that open-minded people are more adept at "looking into things" contains an element of literal truth. The present study provides additional confirmation of the interpretation of IPA as crystallized intelligence. Crystallized intelligence is supposed to be unaffected by the aging process; and the present results show that this factor alone is unrelated to age group differences. Moreover, crystallized intelligence is sensitive
Downloaded from http://geronj.oxfordjournals.org/ at The Australian National University on January 22, 2015
Findings relating personality and age to cognitive factors are presented in Fig. 1. The left hand panel of the figure shows that anxious subjects performed less well than adjusted subjects across all age groups on all cognitive ability factors (F = 6.62, 7.08, and 15.44 for IPA, MD, and PAC, respectively). The degrees of freedom for the F values reported here was 943. Subjects open to experience (right hand panel) showed significantly better performance on the Information Processing Ability and Pattern Analysis Capability factors than did subjects closed to experience (F = 6.63 for IPA and 6.22 for PAC). Extraverted subjects (middle panel) did significantly less well only on Pattern Analysis Capability (F = 5.27). Finally, older subjects did less well than younger subjects on Manual Dexterity (F = 31.48), but did not differ on Information Processing Ability (F = .24). The interactions between personality variables and age groups were not significant in any of the above comparisons. When these analyses were repeated using education and SES as covariates, a somewhat different pattern of results emerged. Anxious subjects did not differ from adjusted subjects on Information Processing (F = 1.46) but did score lower on Manual Dexterity (F = 8.13) and Pattern Analysis Capability (F = 8.59). Extraverted subjects again did less well on Pattern Analysis Capability (F = 6.62). Subjects open to experience did not differ significantly from closed subjects on Information Processing (F = .01) but did score higher on Pattern Analysis Capability (F = 4.50). When education and SES were covaried out, older subjects still did less well on Manual Dexterity and Pattern Analysis tasks (F = 55.28 and 28.54 respectively).
AGE, PERSONALITY, AND ABILITY
detrimental at all ages, and the effects of openness are uniformly beneficial for IPA and PAC at all ages. Neither can the declines in cognitive performance be attributed to individual differences in education or social class, since they remain even when these variables are controlled. Although this study provides useful information on the relations between personality and cognitive domains across a wide age range, there are three reasons for using caution in interpreting the results. First, while statistically significant effects were found, their magnitude was quite small. Second, the variables in this study by no means exhaust the domains from which they are drawn. Only 3 cognitive factors are examined here; theorists have proposed from 7 (Thurstone, 1938) to 120 (Guilford, 1967) distinguishable mental abilities. Similarly, the three personality dimensions employed here are merely three prominent selfreport factors. The equally important personality variables of self-concept, defensiveness, or need for achievement might show larger associations with cognitive abilities. Third, the nomothetic strategy of research pursued here necessarily provides only a limited view of the relations between personality and intelligence. As Heim (1970) points out, the functioning of intelligence in the individual is affected by mood, attitude, temperament, character, and aspirations. The abstractions which are measured by standardized tests may have maximum generality as descriptions of independent traits, but they cannot provide extensive information concerning the interplay of psychological processes in the individual. SUMMARY
Traditionally the two psychological domains of intelligence and personality have been treated separately, and some empirical studies suggest that they are statistically independent. Yet it seems reasonable to argue that particular patterns of personality should affect performance on intelligence tests, and perhaps even the development or maintainance of certain forms of cognitive ability. In the present study, the effects on intelligence of three personality dimensions were investigated in a large sample of adult males. It was hypothesized that Anxiety (reflecting the
Downloaded from http://geronj.oxfordjournals.org/ at The Australian National University on January 22, 2015
to variation in education and social class, and of the three cognitive factors, IPA showed the most substantial correlations with these two variables. Since educational opportunities have increased over the past half century, Schaie (1970) has argued that cross-sectional research on the general population would show agerelated declines in crystallized intelligence. Such a decline is not observed in our population since the selection of subjects resulted in an older group who are better educated than their age peers in the general population. Fluid intelligence has been interpreted as reflecting the psychomotor or gross CNS integrity of the organism (Horn, 1970; Horn & Cattell, 1972), and thus corresponds to MD and PAC. These factors are shown in the present study to be relatively independent of education and SES, but substantially affected by age differences. Since all GATB subtests are positively intercorrelated, factor scores generated by leastsquare regression methods represent a somewhat artificial splitting of cognitive abilities into independent dimensions. The IPA factor scores used in the present analyses, for example, represent not the ability to perform well on the vocabulary, arithmetic reason, computation, and name-comparison tasks, but only the portion of that ability which is independent of the MD and PAC factors. This distinction is of some practical importance because the sum of scores on the four subtests which load the first factor is significantly correlated with age. Thus, while the IPA factor may be considered a relatively pure measure of crystallized intelligence, the sum of the IPA subtests is not. Perhaps the most interesting results of the present study concern the negative findings. One hypothesis considered was that the effects of anxiety would be more pronounced in the elderly, since they, as a group, are less familiar with standardized test procedures. Another hypothesis was that older subjects with a longer history of openness to experience, might have accumulated more knowledge and would therefore show a differential superiority over their practical and unimaginative age-peers (this hypothesis assumes that openness to experience is a longitudinally stable trait). In fact, neither of these age-related predications was confirmed. The effects of anxiety are uniformly
667
668
COSTA, FOZARD, McCRAE, ANDBOSSE
REFERENCES
Ames, L. B. Age changes in the Rorschach responses of a group of individual elderly subjects. Journal of Genetic Psychology, 1960, 67, 287-315.
Baltes, P. B., & Labouvie, G. V. Adult development of intellectual performance: Description, explanation, modification. In C. Eisdorfer & M. P. Lawton (Eds.), The psychology of adult development and aging, American Psychological Ass., Washington, 1973. Bates, H. D., Coppinger, N. W., & Oldham, M. J. Psychometric differences between elderly athletes and non-athletes. Gerontologist, 1974,14, (abstract) 54. Bell, B., Rose, C. L., & Damon, A. The Normative Aging Study: An interdisciplinary and longitudinal study of health and aging. Aging & Human Development, 1972, 3, 5-17. Brozek, J. Personality changes with age: An item analysis of the Minnesota Multiphasic Personality Investory. Journal of Gerontology, 1955,10, 194-206. Bruner, J. S., & Postman, L. Perception, cognition, and behavior. Journal of Personality, 1949,18, 14-31. Cat tell, R. B. Theory of fluid and crystallized intelligence: A critical experiment. Journal of Educational Psychology, 1963,54, 1-22. Cattell, R. B. Personality and modd by questionnaire. Jossey-Bass, San Francisco, 1973. Cattell, R. B., Eber, H. W., & Tatsuoka, M. M. Handbook for the Sixteen Personality Factor Questionnaire. Institute for Personality and Ability Testing, Champaign, IL, 1970. Costa, P. T., Jr., & McCrae, R. R. Age differences in personality structure: A cluster analytic approach. Journal of Gerontology, 1976,31, 564-570. Eisdorfer, C. Arousal and performance: Experiments in verbal learning and a tentative theory. In G. A. Talland (Ed.), Human aging and behavior. Academic Press, New York, 1968. Eysenck, H. J. Relation between intelligence and personality. Perceptual & Motor Skills. 1911.32. 637-638. Fozard, J. L., Nuttall, R. L., & Waugh, N. C. Age differences in mental performance. Aging & Human Development, 1972,5, 19-43. Fozard, J. L., & Thomas, J. C , Jr. Psychology of aging: Basic findings and their psychiatric applications. In J. G. Howells (Ed.), Modern perspectives in the psychiatry of old age. Brunner-Mazel, New York, 1975. Guilford, J. P. The nature of human intelligence. McGrawHill, New York, 1967. Heim, A. Intelligence and personality. Penguin Books, Baltimore, 1970. Horn, J. L. Organization of data on life-span development of human abilities. In L. R. Goulet & P. B. Baltes (Eds.), Life-span developmental psychology: Research and theory. Academic Press, New York, 1970. Horn, J. L. & Cattell, R. B. Age differences in fluid and crystallized intelligence. Ada Psychologica, 1972, 26, 103-129. Nie, N. H., Hull, C.H., Jenkins, J. G., Steinbrenner, K., & Bent, D. H. SPSS: Statistical package for the social sciences. McGraw-Hill, New York, 1975. Nuttall, R. L., & Fozard, J. L. A re-examination of the structure of the General Aptitude Test Battery aptitudes. Industrial Gerontology, 1971,5, 1-18. Ross, J. E. A measure of crystallized intelligence in the General Aptitude Test Battery. American Psychologist, 1968,25, (abstract) 820. Schaie, K. W. A reinterpretation of age related changes in cognitive structure and functioning. In L. R. Goulet &
Downloaded from http://geronj.oxfordjournals.org/ at The Australian National University on January 22, 2015
over-emotionality or maladjustment of subjects) would interfere with cognitive performance; that Introversion (implying taskorientation and seriousness as well as low sociability) would facilitate performance in the formal test-taking situations; and that Openness to Experience would stimulate interest in the tests. In order to test these hypotheses, three distinct forms of intellectual or cognitive ability were examined in all subjects. These three were concerned with such areas as vocabulary and arithmetic reasoning; manual dexterity; and the ability to visualize three-dimensional objects. Most such measures of intelligence have shown cross-sectional declines in performance with increasing age. One purpose of the present study was to determine whether these declines might be mediated by personality factors. Specifically, an age by personality interaction was hypothesized such that anxiety might be much more debilitating in older subjects, for whom intelligence tests may be particularly unfamiliar and threatening. Also with regard to age, it seemed possible that older persons who had spent a lifetime being receptive to new experience might show an accumulated superiority in cognitive functioning over their closedminded peers. Results showed small but significant relations between personality and cognitive ability factors, most of which could not be accounted for solely by education or social class. As predicted, anxious subjects scored lower than adjusted subjects in all three forms of intelligence, while subjects more open to experience scored higher than closed-minded subjects in two of the three. Introverts showed some superiority over extraverts, but only in one form of cognitive ability. Consistent with other studies, older subjects generally did less well than younger subjects; but the hypothesized interactions of age and personality were not found. This suggests that while personality and intelligence are subtly related, the decline in cognitive functioning with age cannot be attributed to the influence of personality variables, or at least those variables with which the present study was concerned.
AGE, PERSONALITY, AND ABILITY
669
Security. Manual for the use of the General Aptitude P. B. Baltes (Eds.). Life-span developmental Test Battery, Section III. Development. U. S. Governpsychology: Research and theory. Academic Press, New ment Printing Office, Washington, 1967. York, 1970. Thurstone, L. L. Primary mental abilities. Psychometric Warner, W. L., Meeker, M., & Eells, K. Social class in Monograph, No. 1. Univ. of Chicago Press, Chicago, America: The evaluation of status. Harper, New York, 1938. 1960. Tryon, R. C , & Bailey, D. E. Cluster analysis. McGraw- Witkin, H. A., Lewis, H. B., Hertman, M., Machover, K., Hill, New York, 1970. Meisser, P. B., & Wapner, S. Personality through perception. Harper, New York, 1954. United States Dept. of Labor, Bureau of Employment Downloaded from http://geronj.oxfordjournals.org/ at The Australian National University on January 22, 2015
Relations of Age and Personality Dimensions to Cognitive Ability Factors1 The relation between three cognitive ability factors — Information Processing Ability (IPA), Manual Dexterity (MD), and Pattern Analysis Capability (PAC) — and three personality dimensions — Anxiety, Extroversion, and Openness to Experience — were examined in three age groups. Subjects were 969 male volunteers ranging in age from 25 to 82. Subjects high in anxiety scored lower on all three cognitive factors; subjects open to experience scored higher on IPA and PAC; and introverted subjects scored higher on PAC. Most of these effects remained when the education and socio-economic status were held constant in covariance analyses. Older subjects performed less well than younger ones on MD and PAC, but not on IPA. While personality has some influence on cognitive performance, the declines with age in performance on some cognitive tasks are not mediated by personality.
EVELOPMENTAL psychologists have D a long-standing interest in the relations among measures of personality and ability. On the one hand, intellectual ability may contribute to the development of particular patterns of personality; on the other, motivational factors and personality traits may influence performance on tasks which require problem solving and manipulation of verbal and nonverbal symbols. Except for research in which individual differences in personality are defined in terms of performance on cognitive or perceptual tasks (e.g., Bruner & Postman, 1949; Witkin, Lewis, Hertman, Machover, Meisser, & Wapner, 1954), studies on young adults have generally found little relation between measures of personality and abilities (Eysenck, 1971). The present study explores the possibility that, when the entire adult age range is represented, subtle influences of personality on cognitive performance may be more evident. In studies of age differences in abilities, individual variations in personality have most frequently been used to account for part of the ob'From the Veterans Administration Normative Aging Study. Portions of this paper were presented at the meeting of the Gerontological Society, October, 1975, Louisville. Research supported in part by the Medical Research Service of the Veterans Administration, The Council for Tobacco Research - U. S. A. Grant 1085, and by NIH Grant 00467. 'VA Outpatient Clinic, 17 Court St., Boston 02108. 'Also at Univ. of Massachusetts at Boston, Dorchester, MA 02125 'Also at Harvard Univ. School of Medicine. 'Also at Hellenic College, Brookline, MA.
served decline in cognitive performance associated with age. For example, it has been postulated that poorer performance of older persons on cognitive tasks is a result of anxiety generated by concern about their abilities and by unfamiliarity with standardized test procedures (e.g., Baltes & Labouvie, 1973; Fozard & Thomas, 1975). While anxiety may result in inhibited performance on intellectual tasks, there is no satisfactory evidence to demonstrate that anxiety affects the performance of older persons relatively more than that of younger ones. One purpose of the present study is to evaluate the hypothesis that there is an interaction between anxiety and age on measures of abilities. A second way in which personality might affect test performance is through the social setting of the test situation. The introverted, taskoriented individual might be more comfortable with the formal aspects of the testing situation than the extravert whose outgoing mode of social interaction could be less appropriate, thereby interfering with task performance. While available evidence provides no strong basis for such an hypothesis in young adults, the effects of extraversion on styles of social interaction are so pervasive that its relation to intellectual ability over the adult age range should be investigated, particularly because there is evidence that greater introversion is associated with older age (Ames, 1960; Brozek, 1955). 663
Downloaded from http://geronj.oxfordjournals.org/ at The Australian National University on January 22, 2015
Paul T. Costa, Jr., PhD, 2 3 James L. Fozard, PhD, 2 3 Robert R. McCrae, PhD3 and Raymond Bosse.'PhD2 5
664
COSTA, FOZARD, McCRAE, ANDBOSSE
METHOD
Measure of Abilities and Personality The three cognitive ability factors resulted from a principal axis factor analysis of the twelve subtests of the GATB. The varimax rotated factor loadings for the 12 subtests and the correlations of the factors with age, socioeconomic status (SES; Warner, Meeker, & Eells, 1960), and years of educations are shown in Table 1. This pattern of factor loadings has been replicated by Nuttall and Fozard (1971) using two other sets of published data, and by Bates, Coppinger, and Oldham (1974) using a sample of elderly individuals ranging in age from 75 to 96. The first factor, Information Processing Ability (IPA), measures the facility with which familiar verbal and numerical materials are used. The tests involve numerical computation, finding the synonyms and antonyms of words,
Table 1. Varimax Rotated Factor Loadings for 12 GATB Subtests and Correlations with Age, SES, and Education. Factor Loading h1
Variable
Information Processing Ability (IPA)
Manual Dexterity (MD)
Pattern Analysis Capability (PAC)
IPA
MD
PAC
Arithmetic reason
81
07
18
69
Computation
82
19
13
73
Vocabulary
69
07
27
55
Name Comparison
66
18
29
55
Dissassemble
13
68
28
56
Place
05
66
11
45
Turn
12
72
14
55
Assemble
07
51
23
32
Mark Making
41
52
06
44
Three Dimensional Space
26
19
69
58
Form Matching
23
31
67
60 51
Tool Matching
38
29
53
Age
-09
-44
-30
Warner Level SES
-44
00
-12
53
02
13
Correlates
Education
identifying differences in the spelling of names, and solving arithmetic problems. The tests are most similar to what Cattell calls crystallized intelligence (Ross, 1968), although the present tests put more emphasis on application of verbal and arithmetic skills than on accumulated capacity as measured by definitions of words or availability of information. Scores on the Information Processing Ability factor are highly related to differences in education and SES. The second factor, Manual Dexterity (MD), is composed of tests of finger and hand movements. It is most closely associated with age and is uncorrelated with variations in social class and education. In the present sample, manual dexterity performance is highly correlated with response speed on conventional choice reaction tasks (Fozard, Nuttall, & Waugh, 1972). Because all of the tests require repetitive aiming and transport movements, good performance requires one to sustain rhythm and organization of movement. Accordingly, manual dexterity represents a more complex set of skills than is usually defined by simple motor speed tests. The third factor, Pattern Analysis Capability (PAC), consists of two tests which require the
Downloaded from http://geronj.oxfordjournals.org/ at The Australian National University on January 22, 2015
A third possibility is that differences associated with receptiveness to new experience could affect test performances. The more open, imaginative individual may be expected to show more interest in intellectual matters in general and in the test situation in particular. This could result in higher test scores for these individuals at all ages. The differences may become more pronounced with age because the person who is sensitive to his environment and open to new ideas is likely to accumulate a greater variety of cognitive experiences over his lifetime than his practical and unimaginative age-peer. The differences should be most pronounced on measures dependent upon accumulated knowledge and abilities, characterized by Cattell (1963) as crystallized intelligence. These measures may covary to some extent with social class and educational experience. To explore the three possible relationships between personality and abilities described above, the present study examined the relationships among three measures of cognitive abilities derived from the General Aptitude Test Battery (GATB), a group of tests developed for occupational aptitude testing (US Dept. of Labor, 1967) and three personality dimensions derived from the Cattell Sixteen Personality Factor Questionnaire or 16PF (Cattell, Eber, & Tatsuoka, 1970).
AGE, PERSONALITY, AND ABILITY
Subjects and Procedure The subjects were 969 male volunteers, all participants in the Normative Aging Study (Bell, Rose, & Damon, 1972). These subjects are distinguished from the general population in that all were judged to be healthy at the time of their entry into the study between 1963 and 1968. In a single testing session, small groups of subjects were administered the GATB and the 16PF during a period from 1967 to 1968. Age
groups were formed on the basis of age at the time of psychological testing. The three age groups with sample size, mean and standard deviation of age are as follows: Young, N = 140, X = 31.7, SD = 2.28; Middle-aged, N = 711, X = 43.7, SD = 5.54; Old, N = 109, X = 60.3, SD = 5.24. Analyses Cluster dimension scores were calculated for each subject within each of the three age groups by summing the standardized scores of each of the component scales. Cognitive ability factor scores were least-square regression estimates of Varimax rotated factors. Four-way analyses of variance were performed for each of the cognitive factors, using age group, anxiety, extraversion, and openness to experience as the classifying variables. The "classic experimental approach" option of the SPSS (Nie, Hull, Jenkins, Steinbrenner, & Bent, 1974) analysis of variance program was used to accommodate unequal cell frequencies. Personality groups were formed by dichotomizing individuals according to whether their personality scores were higher or lower than the mean.for their age group. In order to assess the influence of educational level and SES on the relations between personality and cognitive abilities, the four-way analyses described above were also performed using years of education and Warner Level as covariates. INFORMATION
PROCESSING ABILITY
MANUAL DEXTERITY .6 .2 -.2 -.6 -1.0
PATTERN ANALYSIS
.6 %
CAPABILITY
.2
1 -, 30
40
50
60
30
40
50
60
30
40
50
60
MEAN AGE OF CROUPS (years)
Fig. 1. Unadjusted deviation contrasts by age and personality on three ability factors.
Downloaded from http://geronj.oxfordjournals.org/ at The Australian National University on January 22, 2015
subject to locate similar forms and one in which he is to identify the three dimensional realizations of two dimensional drawings. Performance on these three is moderately correlated both with age and with education and social class. While there is no direct evidence, Pattern Analysis Capability probably measures an ability similar to what Cattell (1963) would refer to as Fluid Intelligence. A cluster analysis (Tryon & Bailey, 1970) of Forms A and B of the Cattell 16 PF (Costa & McCrae, 1976) yielded three cluster dimensions: Anxiety-Adjustment, IntroversionExtraversion, and Open-Closed Experiential Style. The anxiety cluster was defined by the factors of low emotional stability (C-), suspiciousness (L), guilt-proneness (0), low selfsentiment integration (Q3-), and ergic tension (Q4). It represents a coping dimension of adjustment-maladjustment, which includes trait anxiety, but also other maladjustive emotional responses. The extraversion cluster is composed of the factors warmth (A), surgency (F), adventurousness (H), and group dependence (Q2-). These two clusters resemble the first two second order factors repeatedly found in analyses of the 16PF (Cattell, 1973). The third cluster identified was age specific in structure. Around the core element of imaginativeness (Factor M) three different clusters emerged in three age groups. In the young group, imaginativeness (M), and tendermindness (I) defined an openness to feelings cluster. In the middle group, imaginativeness and radicalism (Ql) seemed to form an openness to ideas cluster. In the old group, both tendermindedness (I) and radicalism (QI) joined imaginativeness (M) as did intellectual brightness (B). This combination was identified as a balanced openness to both feelings and ideas. Across the three age groups, the third cluster can be interpreted as an open-closed experiential style dimension.
665
666
COSTA, FOZARD, McCRAE, ANDBOSSE
RESULTS
DISCUSSION
The hypotheses concerning the relations between personality and cognitive abilities were partially confirmed in this study. Results of the analysis of variance showed that anxious subjects consistently scored significantly lower than adjusted subjects on all three intelligence factors, and in all age groups. While anxiety may be related to intrinsic intellectual ability (or competence), it seems most probable that the adverse effects of anxiety on test performance accounts for the differences between
groups. Trait anxiety predisposes individuals to state anxiety, and state anxiety interferes with performance on these cognitive and perceptual motor tasks. With regard to Information Processing, the relation between cognitive performance and anxiety seems to be mediated by the environmental influences of education and SES. When these effects were statistically controlled, anxiety showed no relation to performance. This may help to explain why studies conducted on certain populations (where there is typically little variation in education and SES) have shown no relation between anxiety and intellectual ability (Eisdorfer, 1968). It was hypothesized that extraverted subjects might be at a disadvantage because of the formality of the testing situation. If this were so, they would be expected to perform uniformly less well on all three types of ability factors. In fact, extraverts show significantly lower scores only on the Pattern Analysis Capability factor. Pattern analysis seems to involve the comparison of perceptions with internal schemata, and this suggests that perhaps it is the introspective capacity of introverts rather than their social orientation which affects cognitive performance. It was also hypothesized that persons open to experience would show greater cognitive abilities, particularly in the area of information processing, where the cumulative effect of acquiring novel experience would be most beneficial. While open subjects did show higher scores on both IPA and PAC, the relation between openness and IPA appears to have been accounted for by the effects of education and SES. Greater amounts of education may lead to more tolerance for new experience, and also contribute to the development of information processing ability. On the other hand, openness to experience was found to be related to Pattern Analysis Capability even when levels of education and SES were controlled. Perhaps the metaphorical explanation that open-minded people are more adept at "looking into things" contains an element of literal truth. The present study provides additional confirmation of the interpretation of IPA as crystallized intelligence. Crystallized intelligence is supposed to be unaffected by the aging process; and the present results show that this factor alone is unrelated to age group differences. Moreover, crystallized intelligence is sensitive
Downloaded from http://geronj.oxfordjournals.org/ at The Australian National University on January 22, 2015
Findings relating personality and age to cognitive factors are presented in Fig. 1. The left hand panel of the figure shows that anxious subjects performed less well than adjusted subjects across all age groups on all cognitive ability factors (F = 6.62, 7.08, and 15.44 for IPA, MD, and PAC, respectively). The degrees of freedom for the F values reported here was 943. Subjects open to experience (right hand panel) showed significantly better performance on the Information Processing Ability and Pattern Analysis Capability factors than did subjects closed to experience (F = 6.63 for IPA and 6.22 for PAC). Extraverted subjects (middle panel) did significantly less well only on Pattern Analysis Capability (F = 5.27). Finally, older subjects did less well than younger subjects on Manual Dexterity (F = 31.48), but did not differ on Information Processing Ability (F = .24). The interactions between personality variables and age groups were not significant in any of the above comparisons. When these analyses were repeated using education and SES as covariates, a somewhat different pattern of results emerged. Anxious subjects did not differ from adjusted subjects on Information Processing (F = 1.46) but did score lower on Manual Dexterity (F = 8.13) and Pattern Analysis Capability (F = 8.59). Extraverted subjects again did less well on Pattern Analysis Capability (F = 6.62). Subjects open to experience did not differ significantly from closed subjects on Information Processing (F = .01) but did score higher on Pattern Analysis Capability (F = 4.50). When education and SES were covaried out, older subjects still did less well on Manual Dexterity and Pattern Analysis tasks (F = 55.28 and 28.54 respectively).
AGE, PERSONALITY, AND ABILITY
detrimental at all ages, and the effects of openness are uniformly beneficial for IPA and PAC at all ages. Neither can the declines in cognitive performance be attributed to individual differences in education or social class, since they remain even when these variables are controlled. Although this study provides useful information on the relations between personality and cognitive domains across a wide age range, there are three reasons for using caution in interpreting the results. First, while statistically significant effects were found, their magnitude was quite small. Second, the variables in this study by no means exhaust the domains from which they are drawn. Only 3 cognitive factors are examined here; theorists have proposed from 7 (Thurstone, 1938) to 120 (Guilford, 1967) distinguishable mental abilities. Similarly, the three personality dimensions employed here are merely three prominent selfreport factors. The equally important personality variables of self-concept, defensiveness, or need for achievement might show larger associations with cognitive abilities. Third, the nomothetic strategy of research pursued here necessarily provides only a limited view of the relations between personality and intelligence. As Heim (1970) points out, the functioning of intelligence in the individual is affected by mood, attitude, temperament, character, and aspirations. The abstractions which are measured by standardized tests may have maximum generality as descriptions of independent traits, but they cannot provide extensive information concerning the interplay of psychological processes in the individual. SUMMARY
Traditionally the two psychological domains of intelligence and personality have been treated separately, and some empirical studies suggest that they are statistically independent. Yet it seems reasonable to argue that particular patterns of personality should affect performance on intelligence tests, and perhaps even the development or maintainance of certain forms of cognitive ability. In the present study, the effects on intelligence of three personality dimensions were investigated in a large sample of adult males. It was hypothesized that Anxiety (reflecting the
Downloaded from http://geronj.oxfordjournals.org/ at The Australian National University on January 22, 2015
to variation in education and social class, and of the three cognitive factors, IPA showed the most substantial correlations with these two variables. Since educational opportunities have increased over the past half century, Schaie (1970) has argued that cross-sectional research on the general population would show agerelated declines in crystallized intelligence. Such a decline is not observed in our population since the selection of subjects resulted in an older group who are better educated than their age peers in the general population. Fluid intelligence has been interpreted as reflecting the psychomotor or gross CNS integrity of the organism (Horn, 1970; Horn & Cattell, 1972), and thus corresponds to MD and PAC. These factors are shown in the present study to be relatively independent of education and SES, but substantially affected by age differences. Since all GATB subtests are positively intercorrelated, factor scores generated by leastsquare regression methods represent a somewhat artificial splitting of cognitive abilities into independent dimensions. The IPA factor scores used in the present analyses, for example, represent not the ability to perform well on the vocabulary, arithmetic reason, computation, and name-comparison tasks, but only the portion of that ability which is independent of the MD and PAC factors. This distinction is of some practical importance because the sum of scores on the four subtests which load the first factor is significantly correlated with age. Thus, while the IPA factor may be considered a relatively pure measure of crystallized intelligence, the sum of the IPA subtests is not. Perhaps the most interesting results of the present study concern the negative findings. One hypothesis considered was that the effects of anxiety would be more pronounced in the elderly, since they, as a group, are less familiar with standardized test procedures. Another hypothesis was that older subjects with a longer history of openness to experience, might have accumulated more knowledge and would therefore show a differential superiority over their practical and unimaginative age-peers (this hypothesis assumes that openness to experience is a longitudinally stable trait). In fact, neither of these age-related predications was confirmed. The effects of anxiety are uniformly
667
668
COSTA, FOZARD, McCRAE, ANDBOSSE
REFERENCES
Ames, L. B. Age changes in the Rorschach responses of a group of individual elderly subjects. Journal of Genetic Psychology, 1960, 67, 287-315.
Baltes, P. B., & Labouvie, G. V. Adult development of intellectual performance: Description, explanation, modification. In C. Eisdorfer & M. P. Lawton (Eds.), The psychology of adult development and aging, American Psychological Ass., Washington, 1973. Bates, H. D., Coppinger, N. W., & Oldham, M. J. Psychometric differences between elderly athletes and non-athletes. Gerontologist, 1974,14, (abstract) 54. Bell, B., Rose, C. L., & Damon, A. The Normative Aging Study: An interdisciplinary and longitudinal study of health and aging. Aging & Human Development, 1972, 3, 5-17. Brozek, J. Personality changes with age: An item analysis of the Minnesota Multiphasic Personality Investory. Journal of Gerontology, 1955,10, 194-206. Bruner, J. S., & Postman, L. Perception, cognition, and behavior. Journal of Personality, 1949,18, 14-31. Cat tell, R. B. Theory of fluid and crystallized intelligence: A critical experiment. Journal of Educational Psychology, 1963,54, 1-22. Cattell, R. B. Personality and modd by questionnaire. Jossey-Bass, San Francisco, 1973. Cattell, R. B., Eber, H. W., & Tatsuoka, M. M. Handbook for the Sixteen Personality Factor Questionnaire. Institute for Personality and Ability Testing, Champaign, IL, 1970. Costa, P. T., Jr., & McCrae, R. R. Age differences in personality structure: A cluster analytic approach. Journal of Gerontology, 1976,31, 564-570. Eisdorfer, C. Arousal and performance: Experiments in verbal learning and a tentative theory. In G. A. Talland (Ed.), Human aging and behavior. Academic Press, New York, 1968. Eysenck, H. J. Relation between intelligence and personality. Perceptual & Motor Skills. 1911.32. 637-638. Fozard, J. L., Nuttall, R. L., & Waugh, N. C. Age differences in mental performance. Aging & Human Development, 1972,5, 19-43. Fozard, J. L., & Thomas, J. C , Jr. Psychology of aging: Basic findings and their psychiatric applications. In J. G. Howells (Ed.), Modern perspectives in the psychiatry of old age. Brunner-Mazel, New York, 1975. Guilford, J. P. The nature of human intelligence. McGrawHill, New York, 1967. Heim, A. Intelligence and personality. Penguin Books, Baltimore, 1970. Horn, J. L. Organization of data on life-span development of human abilities. In L. R. Goulet & P. B. Baltes (Eds.), Life-span developmental psychology: Research and theory. Academic Press, New York, 1970. Horn, J. L. & Cattell, R. B. Age differences in fluid and crystallized intelligence. Ada Psychologica, 1972, 26, 103-129. Nie, N. H., Hull, C.H., Jenkins, J. G., Steinbrenner, K., & Bent, D. H. SPSS: Statistical package for the social sciences. McGraw-Hill, New York, 1975. Nuttall, R. L., & Fozard, J. L. A re-examination of the structure of the General Aptitude Test Battery aptitudes. Industrial Gerontology, 1971,5, 1-18. Ross, J. E. A measure of crystallized intelligence in the General Aptitude Test Battery. American Psychologist, 1968,25, (abstract) 820. Schaie, K. W. A reinterpretation of age related changes in cognitive structure and functioning. In L. R. Goulet &
Downloaded from http://geronj.oxfordjournals.org/ at The Australian National University on January 22, 2015
over-emotionality or maladjustment of subjects) would interfere with cognitive performance; that Introversion (implying taskorientation and seriousness as well as low sociability) would facilitate performance in the formal test-taking situations; and that Openness to Experience would stimulate interest in the tests. In order to test these hypotheses, three distinct forms of intellectual or cognitive ability were examined in all subjects. These three were concerned with such areas as vocabulary and arithmetic reasoning; manual dexterity; and the ability to visualize three-dimensional objects. Most such measures of intelligence have shown cross-sectional declines in performance with increasing age. One purpose of the present study was to determine whether these declines might be mediated by personality factors. Specifically, an age by personality interaction was hypothesized such that anxiety might be much more debilitating in older subjects, for whom intelligence tests may be particularly unfamiliar and threatening. Also with regard to age, it seemed possible that older persons who had spent a lifetime being receptive to new experience might show an accumulated superiority in cognitive functioning over their closedminded peers. Results showed small but significant relations between personality and cognitive ability factors, most of which could not be accounted for solely by education or social class. As predicted, anxious subjects scored lower than adjusted subjects in all three forms of intelligence, while subjects more open to experience scored higher than closed-minded subjects in two of the three. Introverts showed some superiority over extraverts, but only in one form of cognitive ability. Consistent with other studies, older subjects generally did less well than younger subjects; but the hypothesized interactions of age and personality were not found. This suggests that while personality and intelligence are subtly related, the decline in cognitive functioning with age cannot be attributed to the influence of personality variables, or at least those variables with which the present study was concerned.
AGE, PERSONALITY, AND ABILITY
669
Security. Manual for the use of the General Aptitude P. B. Baltes (Eds.). Life-span developmental Test Battery, Section III. Development. U. S. Governpsychology: Research and theory. Academic Press, New ment Printing Office, Washington, 1967. York, 1970. Thurstone, L. L. Primary mental abilities. Psychometric Warner, W. L., Meeker, M., & Eells, K. Social class in Monograph, No. 1. Univ. of Chicago Press, Chicago, America: The evaluation of status. Harper, New York, 1938. 1960. Tryon, R. C , & Bailey, D. E. Cluster analysis. McGraw- Witkin, H. A., Lewis, H. B., Hertman, M., Machover, K., Hill, New York, 1970. Meisser, P. B., & Wapner, S. Personality through perception. Harper, New York, 1954. United States Dept. of Labor, Bureau of Employment Downloaded from http://geronj.oxfordjournals.org/ at The Australian National University on January 22, 2015
