Can Personality Factors Predict Intelligence

Personality and Individual Differences 38 (2005) 1021–1033 www.elsevier.com/locate/paid

Can personality factors predict intelligence? Joanna Moutafi, Adrian Furnham *, Laurence Paltiel Department of Psychology, University College London, 26 Bedford Way, London WCI OAP, UK Received 27 January 2004; received in revised form 29 May 2004; accepted 30 June 2004 Available online 7 October 2004

Abstract The present study investigated the relationship between personality traits and psychometric intelligence. A total of 4859 participants completed the Critical Reasoning Test Battery (GRT2) and the Fifteen Factor Questionnaire (15FQ). Of the second-order personality factors, Conscientiousness, Extraversion and Neuroticism were significant predictors of general intelligence (g). Regressing personality and demographic factors on g indicated that they accounted for 13% of its variance. The investigation of personality predictors of specific mental abilities (numerical, verbal and abstract reasoning) revealed that although some variables can be used to predict scores on all three abilities (e.g. Conscientiousness, Extraversion), other variables can be used to predict only specific abilities (e.g. Openness, Neuroticism). Regressing personality and demographic factors on specific abilities indicated that they accounted for 9–17% of the variance in intelligence scores.
2004 Elsevier Ltd. All rights reserved. Keywords: Intelligence; Personality; Sex; Age

1. Introduction The present study is an investigation of the extent to which personality traits can predict psychometric intelligence. Intelligence and personality are usually treated as relatively distinct in the

*

Corresponding author. Tel.: +44 207 3877050; fax: +44 207 4364276. E-mail address: [email protected] (A. Furnham).

0191-8869/$ - see front matter
2004 Elsevier Ltd. All rights reserved. doi:10.1016/j.paid.2004.06.023

1022

J. Moutafi et al. / Personality and Individual Differences 38 (2005) 1021–1033

research on individual differences, however, many studies have shown that there are consistent predictable correlations between these two constructs (Ackerman & Heggestad, 1997; Austin et al., 2002; Collis & Messick, 2001; Furnham, Forde, & Cotter, 1998; Goff & Ackerman, 1992; Kyllonen, 1997; Zeidner, 1995). These studies have been based on a variety of intelligence and personality measures, which provides a broader understanding of the relationships between the two constructs, but has led to inconsistencies in the findings, which make interpretations difficult. Here we will summarize the findings that relate general, fluid and crystallized intelligence to the traits of the Five Factor Model of personality, which was proposed by McCrae and Costa (1987). Out of the Big 5 personality factors, the one which has most consistently been found to correlate with intelligence is Openness (Zeidner & Matthews, 2000). Significant correlations have been found between Openness and general intelligence (Austin et al., 2002; Kyllonen, 1997; Moutafi, Furnham, & Crump, 2003), and between Openness and quantitative ability (Kyllonen, 1997). It has however been proposed that Openness more specifically correlates with crystallized intelligence (gc) (Brand, 1994). Indeed, correlations of the magnitude of r = 0.58 have been reported between Openness and verbal ability, which is a measure of gc (Kyllonen, 1997). Goff and Ackerman (1992) suggested that the relationship between gc and Openness may be mediated by Typical Intellectual Engagement (TIE), which is a measure of intellectual motives and interests. Another factor, which has been found to be correlated with intelligence, is Extraversion (Zeidner & Matthews, 2000). Findings on this relationship have been controversial, in that Extraversion has been found to be both positively (Ackerman & Heggestad, 1997; Austin et al., 2002; Lynn, Hampson, & Magee, 1984) and negatively (Furnham et al., 1998; Moutafi et al., 2003) linked to measures of intelligence. It has been proposed that the relationship between Extraversion and intelligence is mediated by the nature of the intelligence test, due to Introverts having a higher resting level of cortical arousal than Extraverts (Eysenck, 1967; Eysenck, 1994). This suggestion is in line with Robinson (1985), who claimed that Extraversion is associated with intellectual styles and intelligence profiles and not with actual ability. More specifically, Extraverts have been found to perform better on timed tasks (Rawlings & Carnie, 1989) whereas Introverts tend to perform better in tasks requiring insight and reflection (Matthews, 1992; Saklofske & Zeidner, 1995). Significant negative correlations have also been observed between Neuroticism and intelligence (Ackerman & Heggestad, 1997; Kyllonen, 1997; Moutafi et al., 2003). The negative sign indicates that individuals who score highly on Neuroticism tend to achieve lower intelligence scores. One of the sub-factors of Neuroticism, which seems to mediate the relationship between Neuroticism and intelligence, is anxiety. According to Eysenck (1979), high-anxiety individuals engage in significantly more task-irrelevant processing (worry) than low-anxiety individuals, which impairs their performance. Negative correlations between anxiety and intelligence have been reported (Moutafi et al., 2003) and experimental research has shown that anxiety can impair intellectual functioning in a variety of contexts, ranging from IQ tests to school achievement (Sarason, 1980; Hembree, 1988). It seems that Neuroticism is more related to intelligence test performance than to intelligence per se, in that it affects the test-taking experience, which in turn leads to lower scores (Moutafi, Furnham, & Tsaousis, submitted for publication). The findings on the relationship between Conscientiousness and intelligence have been controversial. Ackerman and Heggestad (1997) reported a near zero correlation between Conscientiousness and g in their meta-analysis, a finding which was supported by Kyllonen (1997). However, a marginally significant negative relationship has been reported between Conscientiousness and cre-

J. Moutafi et al. / Personality and Individual Differences 38 (2005) 1021–1033

1023

ativity (Furnham, 1999), and most importantly, recent studies have reported negative correlations between Conscientiousness and measures of intelligence (Allik & Realo, 1997; Demetriou, Kyriakides, & Avraamidou, 2003; Furnham, Chamorro-Premuzic, & Moutafi, submitted for publication; Moutafi et al., 2003; Moutafi, Furnham, & Paltiel, 2004). It has been suggested that less intelligent individuals, from competitive school, academic or business institutions, may become more conscientious in order to cope with their disadvantage. It is also possible that more intelligent individuals do not become so conscientious, as they can rely on their intelligence to accomplish most tasks. The only factor of the Big 5 that has not been found to be related to intelligence is Agreeableness (Ackerman & Heggestad, 1997; Furnham et al., submitted for publication; Austin et al., 2002; Kyllonen, 1997). The aim of this study was to investigate the relationship between personality and intelligence, using valid psychometric tests hitherto not reported in this specific literature. These tests were used not only to extend our understanding of the relationship between the two constructs, but also to provide psychologists in the occupational field, who use these tests for recruitment and selection, with an understanding of how the measures they use for assessment are theoretically interrelated. In order to obtain a more comprehensive picture, three measures of intelligence were used (numerical reasoning, verbal reasoning and abstract reasoning), in addition to a measure of general intelligence (which was derived from factor analysis of these three). These scores were related to a particular measure of the Big 5, which is used in occupational settings, but has not yet been widely used for research purposes. The first hypothesis (H1) is that Openness will be a significant positive predictor of verbal and numerical reasoning, and of general intelligence. This would support Kyllonen (1997), who found Openness to be positively correlated with verbal ability, quantitative ability, and g. The second hypothesis (H2) is that Extraversion will be a significant positive predictor of the three intelligence tests, as they are timed tests, which have been found to favor Extraverts (Rawlings & Carnie, 1989). The third hypothesis (H3) is that Neuroticism will be a significant negative predictor of general intelligence, in line with Ackerman and Heggestad (1997), and of the three intelligence measures (numerical, verbal, abstract reasoning), in line with the suggestion that Neuroticism is related to intelligence test performance (Moutafi et al., submitted for publication). The fourth hypothesis (H4) is that Conscientiousness will be a significant (negative) predictor of general intelligence, in line with Demetriou et al. (2003) and Furnham et al. (submitted for publication). In the statistical analysis of the data, gender and age of the subjects will be controlled for, as these demographic factors have been found to correlate with intelligence. There is a general consensus is that males and females do not differ in general intelligence (APA Public Affairs Office, 1997; Hyde & Linn, 1988; Loehlin, 2000; Matthews, Davies, Westerman, & Stammers, 2000), although some researchers oppose with this (Lynn, 1994, 1999). Sex differences have however been reported in the variance of IQ scores, with males showing greater variability (Feingold, 1992) and more importantly, researchers have found significant sex differences in specific abilities (Matthews et al., 2000). In the study of age-related differences in cognitive ability, there is a general consensus that scores on intelligence tests tend to decline with age (Schaie, 1994), with fluid intelligence peaking at the age of 16 or 17 and crystallized intelligence peaking within the age range of 45–54 (Ryan, Sattler, & Lopez, 2000). Researchers however also advocate for the stability of intelligence (Deary, Whalley, Lemmon, Crawford, & Starr, 2000), due to a change in the genetic over the environmental contribution across the lifespan (Plomin, Pedersen, Lichtenstein, & McClearn, 1994).

1024

J. Moutafi et al. / Personality and Individual Differences 38 (2005) 1021–1033

2. Method 2.1. Participants A total of 4859 participants were recruited for this study. Of these, 3944 were male and 903 were female (12 did not specify their gender). Their age ranged from 14 to 63, with a mean of 35.28 and a standard deviation of 8.67. In the statistical analysis of the intelligence measures, participants who scored below 4 (out of 35), on any test were excluded from the analysis. This meant that in total 4639 participants were included, of which 3765 were male and 865 were female (9 did not specify their gender). Their age ranged from 14 to 63 with a mean of 35.23 and a standard deviation of 8.61. Participants were all applying for jobs in the human resource management field. 2.2. Materials 2.2.1. The general reasoning test battery (GRT2) (Budd, 1993) This is a timed (28 min), speeded ability test, measuring numerical (NR) (25 items), verbal (VR) (35 items) and abstract (AR) (25 items) reasoning. Numerical reasoning measures the ability to use numbers in a logical and efficient way. Verbal reasoning measures basic vocabulary, verbal fluency, and the ability to reason using words. Abstract reasoning measures the ability to understand abstract logical problems, and use new information outside the range of previous experience. Examination of the alpha coefficients for all three sub-tests of the GRT2 showed that they were all above 0.8, demonstrating a high level of reliability of the test. Furthermore, test–retest coefficients were all above 0.7. In order to test the validity of the GRT2, its sub-scales and total score were compared to the sub-scales and total score of the Alice Heim reasoning test (AH5). Correlation coefficients ranged from 0.56 to 0.76 for the sub-scales, and for the total scores of the two tests it was 0.82 (N = 81), demonstrating that the GRT2 measures the same trait of reasoning ability which is assessed by the AH5, although the discriminant validity of the sub-scales is not very high. These results are presented in Table 1. 2.2.2. Fifteen factor questionnaire (15FQ) (Budd, 1992) This is an un-timed questionnaire, taking approximately 30 min to complete, measuring 15 bipolar personality dimensions. The 15FQ was developed in the UK on a large sample of applicants drawn from a wide range of occupational groups, and is used for personnel assessment and selection (Budd, 1992). The items were selected with the criteria that they cover the construct adequately, while maintaining acceptable levels of scale cohesiveness and minimum overlap with other scales. The dimensions measured by the test are Outgoing, Stable, Assertive, Enthusiastic, Conscientious, Socially bold, Intuitive, Suspicious, Conceptual, Restrained, Self-doubting, Radical, Self-sufficient, Disciplined, Tense driven. Factor analysis has shown that a further five broad underlying characteristics can be derived from the 15FQ, which have been found to be favorably compared to the Big 5 factors of personality. These are Extraversion (corresponding to NEO Extraversion, r = 0.77), Anxiety (corresponding to NEO Neuroticism, r = 0.71), Tough-mindedness (corresponding to NEO Openness,

J. Moutafi et al. / Personality and Individual Differences 38 (2005) 1021–1033

1025

Table 1 Reliability and validity coefficients for GRT2 and 15FQ Reliability

Validity

Alpha coef.

Test–retest coef.

Alice Heim 5

r

N

r

N

Num.

NR VR AR

0.84 0.83 0.83

135 135 135

0.78 0.79 0.74

70 70 70

0.76

FA FC FE FF FG FH FI FL FM FN FO FQ1 FQ2 FQ3 FQ4

0.70 0.67 0.64 0.73 0.72 0.79 0.69 0.74 0.72 0.65 0.77 0.73 0.66 0.80 0.66

618 618 618 618 618 618 618 618 618 618 618 618 618 618 618

0.80 0.71 0.78 0.88 0.68 0.85 0.80 0.76 0.70 0.87 0.71 0.78 0.79 0.79 0.78

83 83 83 83 83 83 83 83 83 83 83 83 83 83 83

16PF Ver.

Per.

0.63 0.56 0.63 0.69 0.57 0.64 0.29 0.75 0.73 0.59 0.72 0.15 0.82 0.59 0.78 0.32 0.25

r = 0.64), Independence (corresponding to NEO Agreeableness, r = 0.55) and Control (corresponding to NEO Conscientiousness, r = 0.36). The technical manual provides evidence for the testÕs reliability and validity (Budd, 1992). All the 15FQ dimensions were found to have reliability coefficients above 0.64, which compare favourably to the reliability coefficients of the 16PF (Smith, 1994). Long-term data also showed test–retest reliability coefficients ranging from 0.70 to 0.88. Evidence for the testÕs construct validity comes from comparisons between the 15FQ and other personality measures such as the 16PF and the NEO short form, which show that the dimensions of the 15FQ are consistent with similar measures. The Alpha coefficients, the test–retest reliability coefficients and the correlations between the 15FQ factors and their corresponding factors of the 16PF are presented in Table 1. Although most scales have satisfactory levels of validity, some scales show rather low levels. The validity of the 15FQ has been however further supported by looking at the relationship of the 15FQ with a number of tests, including the Myers-Briggs Type Indicator, the Jung Type Indicator, the Professional Personality Questionnaire, the Occupational Personality Profile and the FIRO-B (Budd, 1992). 2.3. Procedure Participants were all job applicants tested by Psytech International as part of an assessment center.

1026

J. Moutafi et al. / Personality and Individual Differences 38 (2005) 1021–1033

3. Results 3.1. Intelligence measures Scores on all three intelligence measures were approximately normally distributed with a slight positive skew. Scores on abstract reasoning ranged from 4 to 25 (M = 17.67, SD = 4.67), with 41.5% of the participants responding to all items. Scores on numerical reasoning ranged from 4 to 25 (M = 17.45, SD = 5.31), with 30.4 of the participants responding to all items. Scores on verbal reasoning ranged from 4 to 35 (M = 23.97, SD = 5.43), with 46.2% of the participants responding to all items. Pearson product moment correlations were computed to investigate the relationship between the three sub-tests of the General Reasoning Test 2 (verbal, numerical and abstract reasoning). These are presented in Table 2. Partial correlations were also computed, controlling for gender and age, these are presented in brackets. In order to obtain a measure of general intelligence, principal component analysis was performed on the three intelligence tests. This yielded one factor (g), with loadings of 0.87 (numerical reasoning), 0.84 (abstract reasoning) and 0.83 (verbal reasoning). 3.2. Second-order factors of the 15FQ It has been established by factor analysis that the 15FQ dimensions can be summarized in five broad dimensions of personality (Budd, 1992), which correspond to each of the five dimensions of the Five Factor Model. Correlations between the Big 5 and their corresponding dimensions are given in the method section. These second order factors were computed for this sample using the following formulae, provided by the manual of the 15FQ (Budd, 1992): • • • • •

Anxiety (Neuroticism) = (Suspicious + Self-doubting + Tense driven-Stable)/4 Extraversion (Extraversion) = (Outgoing + Enthusiastic + Socially bold-Self-sufficient)/4 Tough Minded (Openness) = (Intuitive + Conceptual + Radical)/3 Independence (Agreeableness) = (Assertive + Suspicious-Restrained + Radical + Enthusiastic)/5 Control (Conscientiousness) = (Conscientious + Restrained + Disciplined)/3

3.2.1. Factor analysis In order to investigate the validity of the second order factors, factor analysis with varimax rotation was performed on the fifteen personality factors of the 15PF. Five factors were extracted, accounting for 68% of the variance. The rotated factor matrix is presented in Table 3. Table 2 Correlations between GRT2-NR, GRT2-VR and GRT2-AR (and partial correlations) GRT2-NR GRT2-AR *p < 0.001, N = 4639 (N = 4625)

GRT2-VR

GRT2-AR

0.58* (0.59*) 0.53* (0.52*)

0.61* (0.61*)

J. Moutafi et al. / Personality and Individual Differences 38 (2005) 1021–1033

1027

Table 3 Rotated component matrix for second order factors of 15FQ Component Extraversion (E) Outgoing Self-sufficient Socially bold Enthusiastic Self-doubting Tense driven Stable Suspicious Conceptual Intuitive Radical Conscientious Disciplined Restrained Assertive

Anxiety (N) 0.19 0.19 0.13 0.00 0.77 0.73 0.71 0.46 0.00 0.00 0.00 0.18 0.00 .14 0.00

0.80 0.72 0.64 0.59 0.11 0.00 0.19 0.22 0.13 0.00 0.00 0.00 0.00 0.14 0.12

Tough-minded (O)

Control (C)

0.00 0.00 0.23 0.22 0.00 0.00 0.00 0.00 0.78 0.67 0.42 0.16 0.15 0.00 0.00

0.00 0.00 0.11 0.14 0.12 0.19 0.24 0.19 0.00 0.10 0.25 0.67 0.60 0.46 0.00

Independence (A) 0.00 0.00 0.28 0.50 0.22 0.00 0.10 0.31 0.00 0.00 0.32 0.00 0.00 0.41 0.66

As can be seen from the table, Outgoing, Self-sufficient, Socially bold and Enthusiastic load on Extraversion (E), Self-doubting, Stable, Tense driven and Suspicious load on Anxiety (N), Conceptual, Intuitive and Radical load on Tough Minded (O), Conscientious, Disciplined and Restrained load on Control (C), and Enthusiastic, Suspicious, Radical, Restrained and Assertive load on Independence (A). Thus, the results of the factor analysis are in line with the structure of the second-order factors as reported in the manual. The eigenvalues and the variance accounted for by these factors are presented in Table 4. Pearson product moment correlation coefficients between the five personality factors and the four intelligence measures, and partial correlations, controlling for gender and age are presented in Table 3. Multiple regressions were performed for each of the dependent variables (NR, VR, AR and g). The independent variables used were the five factors extracted from the factor analysis, gender and age. These results are summarized in Table 5. 3.2.2. Numerical reasoning The model that used NR as the dependent variable was significant accounting for 9% of the variance in intelligence scores. Significant predictors of NR were Anxiety (–), Independence, Control (–), gender (–) and age (–). Table 4 Eigenvalues and % of variance accounted for by factor analysis components Extraversion (E) Anxiety (N) Tough-Minded (O) Control (C) Independence (A)

Eigenvalues

% of Variance

Cumulative %

3.67 2.26 1.86 1.41 1.11

24.48 15.09 12.42 9.41 7.41

24.48 39.57 51.99 61.40 68.81

1028

J. Moutafi et al. / Personality and Individual Differences 38 (2005) 1021–1033

Table 5 Beta coefficients for multiple regressions of personality and demographic factors on intelligence easures

Anxiety (N) Extraversion (E) Tough-Minded (O) Independence (A) Control (C) Gender Age Regression model Adj. R2

GRT2-NR

GRT2-VR

GRT2-AR

G

Beta 0.08* 0.09* 0.05 0.06 0.20* 0.16* 0.14* F(7,4621) = 63.81* 0.09

Beta 0.05 0.08* 0.07* 0.02 0.24* 0.00 0.16* F(7,4621) = 68.79* 0.09

Beta 0.07* 0.08* 0.04 0.05 0.017* 0.08* 0.39* F(7,4621) = 137.35* 0.17

Beta 0.08* 0.10* 0.01 0.05 0.24* 0.09* 0.29* F(7,4621) = 102.76* 0.13

*p < 0.001

3.2.3. Verbal reasoning The model that used VR as the dependent variable was significant accounting for 9% of the variance in intelligence scores. Significant predictors of VR were Anxiety (–), Tough-minded, Independence, Control (–) and age (–). 3.2.4. Abstract Reasoning The model that used AR as the dependent variable was significant accounting for 17% of the variance in intelligence scores. Significant predictors of AR were Anxiety (–), Independence, Control (–), gender (–) and age (–). 3.2.5. General intelligence The model that used g as the dependent variable was significant accounting for 13% of the variance in intelligence scores. Significant predictors of g were Anxiety (–), Independence, Control (–), gender (–) and age (–). 3.2.6. Curve estimation Non-linear relationships were also tested for each of the five personality factors (as independent variables), and each of the intelligence measures (as dependent variables). These models were viewed in comparison to linear models in order to estimate whether a non-linear model fitted the data better than a linear one. The curve models that were tested were logarithmic, quadratic and cubic. For Anxiety, Extraversion and Control, the model that best fitted the data of all intelligence tests was the linear model. For Independence, none of the models tested were significant at the 0.001 significance level. Finally, for Openness, verbal intelligence was best predicted by the linear model, abstract reasoning was not significantly predicted by any model, numerical reasoning was best predicted by as logarithmic model (F(1,4627) = 12.74, p < 0.001, Adj. R2 = 0.00, Beta = 0.05) and g was best predicted by a quadratic model (F(2,4636) = 8.98, p < 0.001, Adj. R2 = 0.00, Beta1 = 0.23, Beta2 = 0.26), although Openness accounted for less than 1% of the variance in NR and g scores.

J. Moutafi et al. / Personality and Individual Differences 38 (2005) 1021–1033

1029

4. Discussion The aim of this study was to investigate the extent to which personality trait scores can predict psychometric intelligence. Together, the second-order factors of the 15FQ accounted for 13% of the variance in gs, and 9–17% of the variance in specific abilities. A series of hypotheses were tested, with respect to the individual contribution of specific personality factors. The second-order factors of the 15FQ will be referred to in the discussion by their names that correspond to the Five-Factor model, to make comparisons with previous studies clearer. The first hypothesis (H1), which was that Openness (O) would be a significant positive predictor of verbal, numerical reasoning, and of g, was only partly supported by the results. This hypothesis was based on the findings of Kyllonen (1997), who found O to be positively correlated with numerical reasoning and with g, as well as with verbal reasoning. The present results however showed that O was a significant predictor of verbal reasoning only. It has been proposed that Open individuals are more inclined to engage in intellectual activities, especially those of a verbal and cultural nature, which may lead them to develop their verbal reasoning (Zeidner & Matthews, 2000). Based on the suggestion that tests of verbal reasoning are generally believed to measure gc (Horn, 1988), this finding seems to support the suggestion that O specifically and exclusively correlates with crystallized intelligence (gc) (Brand, 1994; Goff & Ackerman, 1992). However, it should be noted that the sub-tests of GRT2 have not been explicitly linked to either gf or gc, and therefore conclusions on the relationship between O and gc cannot be reached by the present study. The second hypothesis (H2), which was that Extraversion (E) would be a significant positive predictor of numerical, verbal and abstract reasoning was not supported by the results. Contrary to expectations, Extraversion was found to be a significant negative predictor of AR, VR, NR and g. The hypothesis proposed here was based on the suggestion that Extraverts have an advantage over Introverts for short and speeded tasks due to their lower resting level of cortical arousal (Rawlings & Carnie, 1989). This is because E has been found to mediate the inverted U-shaped relationship between arousal and intelligence, causing Extraverts to perform better under conditions of medium and high arousal (such as completing short and speeded tests), and Introverts to perform better under conditions of low arousal (Bates & Rock, 2004). The hypothesis was therefore made on the assumption that GRT2 is a short and speeded task, as the test comprises of 85 multiple-choice questions, which must be completed in 28 min. It could however be argued that although the test is definitely a speeded task, it may not necessarily be considered as a short one. This would mean that Extraverts may outperform Introverts initially, but may then get bored (under-aroused), which would cause them to perform less efficiently. This could be tested by future studies, by administering an IQ test on a pc, which would be programmed to calculate achieved scores at different times, i.e. this would allow comparison of ExtravertsÕ and IntrovertsÕ performance at time1 versus time2, even though the psychometric properties of the test would not be valid when only a part of the test would be completed. The third hypothesis (H3) was that Neuroticism (N) would be a significant negative predictor of general intelligence and of all three specific abilities. This hypothesis was partly supported by the results, as Neuroticism was a significant predictor of NR, AR and g, but failed to reach significance for VR at the 0.001 significance level. However, all correlations and partial correlations, including those for VR were significant. It has been proposed that the relationship between N

1030

J. Moutafi et al. / Personality and Individual Differences 38 (2005) 1021–1033

and intelligence is mediated by anxiety (Eysenck & Eysenck, 1985; Moutafi et al., submitted for publication; Sarason, 1980; Zeidner & Matthews, 2000). This is because Neurotic individuals experience higher levels of test (state) anxiety, and anxiety causes individuals to engage in significantly more task-irrelevant processing (worry), which interferes with their performance (Eysenck, 1979). This implies that N is actually more systematically related to intelligence test performance than to intelligence per se. This suggestion cannot actually be argued here, as a measure of state anxiety was not administered in order to test this. However, participants would be expected to experience anxiety, on the basis that their performance on the tests would have major consequences for them, as it would affect their chances to be selected or promoted. It would perhaps be advisable for organizations that use intelligence tests for any form of selection, to also administer a questionnaire measuring anxiety. This is because if N is negatively linked to intelligence test performance due to state anxiety, IQ tests will underestimate the intelligence of Neurotics. This will be so for IQ tests administered mostly under any context, as conditions of testing are generally believed to induce state anxiety (Dobson, 2000), but especially if the results of the tests will have consequences on the stestees, as this will elevate state anxiety further. The fourth hypothesis (H4), which was that Conscientiousness (C) would be a negative predictor of g was clearly supported by the results. C was further a significant negative predictor of numerical, verbal as well as abstract reasoning, and Pearson product moment and partial correlations were also significant for all intelligence measures. These results are in line with recent studies that have investigated the relationship of intelligence and personality (Allik & Realo, 1997; Demetriou et al., 2003; Furnham et al., submitted for publication; Moutafi et al., 2003), which have reported a negative link between C and intelligence. It has been proposed C specifically correlates with fluid intelligence (Moutafi et al., 2004). This has been explained in that this trait may be adaptive, in the sense that low gf may lead to the development of C as a compensatory effect, or that high gf may discourage its development, as gf may be sufficient to cope with situations that would otherwise be efficiently dealt with by utilizing characteristics of C. For example in order to complete an assignment, an individual low on gf would need to be organized, methodical and persistent (which are characteristics of C), but an individual high on gf may have the ability to accomplish the same task with less effort and less organization. What was curious about C, was that it was also a significant predictor of verbal reasoning; in the sense that if VR is indeed found to be a valid measure of gc, it would not be expected to correlate with C as highly as gf does (Moutafi et al., 2004). This is because characteristics of C, such as being organized, methodical and persistent, would be expected to positively affect the development of gc during primary education, as gc is influenced by the environment. On the other hand, the correlation is not totally unexpected, due to the high correlation between gf and gc (Cattell, 1971). Future studies should investigate the relationship between gc and C using established measures of gc, and ideally controlling for gf, in order to reach a conclusion on whether there is an actual relationship between gc and C, or whether this is mediated by gf. In summary, significant predictors of g were Conscientiousness (negative), Neuroticism (negative), age (negative) and sex (indicating an advantage of males). Together, personality and demographic factors accounted for 9–17% of the variance in general intelligence scores. The present findings attest to the link between the constructs of personality and intelligence, emphasizing the importance in investigating their relationship, instead of investigating them as independent

J. Moutafi et al. / Personality and Individual Differences 38 (2005) 1021–1033

1031

constructs. The present study also indicates the importance of investigating the relationship of personality variables in relation to specific types of intelligence. Although several factors were significant predictors of all measures of intelligence (Extraversion, Conscientiousness), others were predictors only of specific types of intelligence. For example Neuroticism was a predictor of NR and AR but not of VR, while Openness was a predictor of VR but not of NR or AR. These relationships with specific measures of intelligence are not discussed to a great extent, as the measures used have not been directly linked to either fluid or crystallized intelligence, and therefore these findings will not contribute greatly to existing knowledge. However, merely the finding that personality variables differentially correlate with specific types of intelligence should caution researchers to specify the types of intelligence they are measuring when investigating their relationship with personality, as it will affect their findings. Although this study was based on a large sample and of a wide age range, there is however a limitation which should be considered. The Big 5 personality factors, which are discussed above were not measured by the NEO PI-R, but were derived by the primary factors of the 15FQ. Although previous research has provided evidence of high correlations between the second order factors of the 15FQ and the Big 5 factors, they are still not identical. However, most of the results replicated previous research findings which used the Big 5 factors as assessed by the NEO PI-R, therefore using the second order factors of the15FQ does not seem to be an important limitation. Furthermore, using a different test which measures the same factors is a way to provide further support for previous research and moreover the primary factors of the 15FQ are conceptually very close to the 16PF, which has thus far not been explored with respect to its relationship to intelligence. The investigation of the relationship between intelligence and personality is important, not only for extending the research on individual differences, but also because it has important implications in the applied field of psychology. Both personality and intelligence are individually used as predictors of different types of performance, such as academic and job performance (Salgado, 1997; Carretta & Doub, 1998; Hunton, Wier, & Stone, 2000; Barrick, Mount, & Judge, 2001). Therefore the understanding of the underlying relationship between these two constructs can be used to improve their predictive validity, and shows that it would be most useful to use both measures in conjunction instead of either individually.

References Ackerman, P. L., & Heggestad, E. D. (1997). Intelligence, personality and interests: evidence for overlapping traits. Psychological Bulletin, 121, 219–245. Allik, J., & Realo, A. (1997). Intelligence, academic abilities, and personality. Personality and Individual Differences, 23, 809–814. APA Public Affairs Office (1997). Intelligence: knowns and unknowns. Washington, DC: American Psychological Association. Austin, A. J., Deary, I. J., Whiteman, M. C., Fowkes, F. G. R., Padersen, N. L., Rabbitt, P., Bent, N., & McInnes, L. (2002). Relationships between ability and personality: Does intelligence contribute positively to personal and social adjustment?. Personality and Individual Differences, 32, 1391–1411. Barrick, M. R., Mount, M. K., & Judge, T. A. (2001). Personality and performance at the beginning of the new millennium: What do we know and where do we go next? International Journal of Selection and Assessment, 9, 9–30.

1032

J. Moutafi et al. / Personality and Individual Differences 38 (2005) 1021–1033

Bates, T., C & Rock, A. (2004). Personality and information processing speed: independent influences on intelligence performance. Intelligence, 32, 33–46. Brand, C. (1994). Open to experience-closed to intelligence: Why the ‘‘Big Five’’ are really the ‘‘Comprehensive Six’’. European Journal of Personality, 8, 299–310. Budd, R. J. (1992). 15FQ Technical Manual. Letchworth: Psytech International Ltd. Budd, R. J. (1993). General, critical and graduate test battery: the technical manual. Letchworth: Psytech International Ltd. Carretta, T. R., & Doub, T. W. (1998). Group differences in the role of g and prior job knowledge in the acquisition of subsequent job knowledge. Personality and Individual Differences, 24, 585–593. Cattell, R. B. (1971). Abilities: their structure, growth, and action. Boston: Houghton Mifflin. Collis, J. M., & Messick, S. (2001). Intelligence and Personality: Bridging the gap in theory and measurement. NJ: Lawrence Erlbaum Associates. Deary, I. J., Whalley, L. J., Lemmon, H., Crawford, J. R., & Starr, J. M. (2000). The stability of individual differences in mental ability from childhood to old age: follow-up of the 1932 Scottish Mental Survey. Intelligence, 28, 49–55. Demetriou, A., Kyriakides, L., & Avraamidou, C. (2003). The missing link in the relations between intelligence and personality. Journal of Research in Personality, 37, 547–581. Dobson, P. (2000). Neuroticism, extraversion and cognitive test performance. International Journal of Selection and Assessment, 8, 99–109. Eysenck, H. (1967). The biological basis of personality. Springfiled, IL: Thomas. Eysenck, H. (1979). The structure and measurement of intelligence. Berlin: Springer-Verlag. Eysenck, H. (1994). Personality and Intelligence: Psychometric and experimental approaches. In R. J. Sternberg & P. Ruzgis (Eds.), Personality and Intelligence (pp. 23–31). Cambridge: Cambridge University Press. Eysenck, H., & Eysenck, M. (1985). Personality and individual differences. New York: Plenum. Feingold, A. (1992). Sex differences in variability in intellectual abilities: A new look at an old controversy. Review of Educational Research, 62, 61–84. Furnham, A. (1999). Personality and creativity. Perceptual and Motor Skills, 88, 407–408. Furnham, A., Chamorro-Premuzic, T., & Moutafi, J. (submitted for publication). Personality and intelligence: gender, the Big five, self-estimated and psychometric intelligence. Furnham, A., Forde, L., & Cotter, T. (1998). Personality and intelligence. Personality and Individual Differences, 24, 187–192. Goff, M., & Ackerman, P. (1992). Personality-intelligence relations: Assessment of typical intellectual engagement. Journal of Educational Psychology, 84, 537–552. Hembree, R. (1988). Correlates causes, effects, and treatment of test anxiety. Review of Educational Research, 58, 47–77. Horn, J. L. (1988). A basis for research on age differences in cognitive abilities. In J. J. McArdle & R. W. Woodcock (Eds.), Human cognitive abilities in theory and in practice. Mahwah, NJ: Lawrence Erlbaum Associates Inc. Hunton, J. E., Wier, B., & Stone, D. N. (2000). Succeding in managerial accounting. Pert 2: A structural equations analysis. Accounting, Organizations and Society, 25, 751–762. Hyde, J. S., & Linn, M. C. (1988). Gender differences in verbal ability: A meta-analysis. Psychological Bulletin, 104, 53–69. Kyllonen, P. (1997). Smart testing. Handbook on testing. In R. Dillon (Ed.) (pp. 347–368). Westport, CT, US: Greenwood Press/Greenwood Publishing Group, Inc. Loehlin, J. C. (2000). Group differences in intelligence. In R. J. Sternberg (Ed.), Handbook of intelligence (pp. 176–193). Cambridge: Cambridge University Press. Lynn, R. (1994). Sex differences in intelligence and brain size: A paradox resolved. Personality and Individual Differences, 17, 257–271. Lynn, R. (1999). Sex differences in intelligence: A developmental theory. Intelligence, 27, 1–12. Lynn, R., Hampson, S., & Magee, M. (1984). Home background, intelligence, personality and education as predictors of unemployment in young people. Personality and Individual Differences, 5, 549–557. Matthews, G. (1992). Extraversion. In: Smith, A. P., & Jones D. M. (Eds.), Handbook of human performance. Vol. 3: state and trait (pp. 95–126). London: Academic. Matthews, G., Davies, R. D., Westerman, S. J., & Stammers, R. B. (2000). Human Performance. East Sussex: Psychology Press.

J. Moutafi et al. / Personality and Individual Differences 38 (2005) 1021–1033

1033

McCrae, R., & Costa, P. (1987). Validation of the five-factor model of personality across instruments and observers. Journal of Personality and Social Psychology, 52, 81–90. Moutafi, J., Furnham, A., & Crump, J. (2003). Demographic and personality predictors of intelligence: A study using the NEO-Personality Inventory and the Myers-Briggs Type Indicator. European Journal of Personality, 17, 79–94. Moutafi, J., Furnham, A., & Paltiel, L. (2004). Why is conscientiousness negatively correlated with intelligence. Personality and Individual Differences, 37, 1013–1022. Moutafi, J., Furnham, A., & Tsaousis, I. (submitted for publication). Is the relationship between intelligence and trait neuroticism mediated by test anxiety? Plomin, R., Pedersen, N. L., Lichtenstein, P., & McClearn, G. E. (1994). Variability and stability in cognitive abilities are largely genetic in later life. Behavior Genetics, 24, 207–215. Rawlings, D., & Carnie, D. (1989). The interaction of EPQ extraversion and WAIS subtest performance under timed and untimed conditions. Personality and Individual Differences, 10, 453–458. Robinson, D. (1985). How personality relates to intelligence test performance: implications for a theory of intelligence, aging research, and personality assessment. Personality and Individual Differences, 6, 203–216. Ryan, J. J., Sattler, J. M., & Lopez, A. J. (2000). Age effects on Wechsler Adult Intelligence Scale III subtests. Archives of Clinical Neuropshycholoyg, 15, 311–317. Saklofske, D. H., & Zeidner, M. (1995). International Handbook of Personality and Intelligence. New York: Plenum Press. Salgado, J. F. (1997). The five-factor model of personality and job performance in the European community. Journal of Applied Psychology, 82, 30–43. Sarason, I. G. (Ed.). (1980). Test anxiety: theory, research and applications. Hillsdale NJ: Lawrence Erlbaum. Schaie, K. W. (1994). The course of adult intellectual development. American Psychologist, 49, 304–313. Smith, P. (1994). The UK standardization of the 16PF5: A supplement of norms and technical data. Windsor: NFER Publishing Co. Zeidner, M. (1995). Personality trait correlates of intelligence. In D. Saklofske & M. Zeidner (Eds.), International handbook of personality and intelligence Perspectives on individual differences (pp. 299–319). New York, NY, US: Plenum Press. Zeidner, M., & Matthews, G. (2000). Intelligence and personality. In R. Sternberg (Ed.), Handbook of intelligence (pp. 581–610). New York, NY, US: Cambridge University Press.