Discussion: Intelligence

The Validity of a New, Self-report Measure of Multiple Intelligence

Adrian Furnham

Published online: 16 September 2009 # Springer Science + Business Media, LLC 2009

Abstract In all, 187 participants completed a new, self-report measure of eight multiple intelligences (Haselbauer 2005), a General Knowledge test (Irwing et al. Personality and Individual Differences 30:857–871, 2001), a measure of Approaches to Learning Styles (Biggs 1987), a measure of the Big Five personality traits (Costa and McCrae 1992), as well as gave their own estimated scores on the Gardner (1999) multiple intelligences. Alpha co-efficients were modest with only three of the eight test-derived, multiple intelligence scores being over .70. ‘Linguistic’ and Mathemat- ical intelligence alone were correlated with General Knowledge. Five of the eight ‘intelligences’ were correlated both with Extraversion and Openness. Regressions indicated that ‘Intrapersonal intelligence’ was closely linked with Stability and Conscientiousness; ‘Interpersonal intelligence’ with Extraversion; ‘Linguistic intel- ligence’ with Openness; ‘Mathematical intelligence’ with Agreeableness and Conscientiousness. Correlations between self-estimated and test-derived emotional intelligence showed correlations ranging from r=.18 to r=.56 for similar type ‘intelligences’. This study provides modest evidence for the concurrent and construct validity of this measure. It requires more psychometric evidence of validity before it is used.

Keywords Multiple intelligences . Self-report . Validity

This study aims to assess the concurrent validity of a new, self-report measure of Gardner’s (1983, 1999) multiple intelligences by relating scores on this test to

Curr Psychol (2009) 28:225–239 DOI 10.1007/s12144-009-9064-z

A. Furnham (*) Department of Clinical, Educational, and Health Psychology, University College London, 26 Bedford Way, London WC1H 0AP, UK e-mail: [email protected]

measures of personality, approaches to learning as well as crystallized intelligence and self-estimated scores of those ‘intelligences’. A web and literature search revealed two things. First, there are a number of different, but similar in design and background, self-report measures of multiple intelligence on the web. They have high face validity but no known psychometric properties, particularly reliability as well as construct and predictive validity. Second, there is no single power or ability test of multiple intelligence although attempts have been made to form a battery made up of validated power tests of each of the different intelligences (Visser et al. 2006a, b). This study will examine the psychometric properties of a self-report measure.

Gardner (1983), in many ways the modern ‘father of multiple intelligence’, defined intelligence as “the ability to solve problems or to create products that are valued within one or more cultural settings.” He maintained that linguistic/verbal and logical/ mathematical are those typically valued in schools. and involved sensitivity to spoken and written language and the ability to learn languages. Logical-mathematical intelligence, he defined as involving the capacity to analyze problems logically, solve maths problems and investigate issues scientifically. These two types of intelligence dominate traditional intelligence tests and have always done so.

Three of Gardner’s (1983) multiple intelligences are particularly apparent in the arts though useful in many avenues of life: musical intelligence, which refers to skill in the performance, composition and appreciation of musical patterns; bodily- kinesthetic intelligence, which is defined as using the whole or parts of the body to solve problems or fashion products; and spatial intelligence, which is the ability to recognize and manipulate patterns in space. Gardner (1983) also outlined two ‘personal intelligences’: interpersonal intelligence which is the capacity to understand the intentions, motivations and desires of other people and work effectively with them; and intrapersonal intelligence which is the capacity to understand oneself and to use this information effectively in regulating one’s life. These have more recently been popularized in the concept of emotional intelligence (Petrides and Furnham 2003).

Gardner (1999. pp. 33–34) more recently defined intelligence as a “biological potential to process information that can be activated in a cultural setting to solve problems or create products that are of value in a culture.” In this book, he introduced three possible new intelligences though he notes: “The strength of the evidence for these varies, and whether or not to declare a certain human capacity another type of intelligence is certainly a judgement call” (p. 47). His new intelligence is naturalist intelligence which is “expertise in the recognition and classification of the numerous species—the flora and fauna—of his or her environment” (p. 43). It is the capacity of taxonomization: to recognize members of a group, to distinguish among members of a species and to chart out the relations, formally or informally, among several species. Spiritual intelligence is supposedly the ability to master a set of diffuse and abstract concepts about being, but also mastering the craft of altering one’s consciousness in attaining a certain state of being. It is an “intelligence that explores the nature of existence in its multifarious guises” (p. 60). Existential intelligence is yet more difficult to define: “the capacity to locate oneself with respect to the furthest reaches of the cosmos—the infinite and infinitesimal—and the related capacity to locate oneself with respect to such existential features of the human condition as the significance of life, the meaning of

226 Curr Psychol (2009) 28:225–239

death, the ultimate fate of the physical and the psychological worlds and such profound experiences as love of another person or total immersion in a work of art” (p. 61). It should be pointed out that Gardner rejects both spiritual and existential intelligence, though others recognize their existence. Gardner (1999) considered, then rejected spiritual/existential intelligences as a further intelligences because they did not meet his criteria for being deemed an ‘intelligence’.

There have been extensive and detailed critiques of various multiple intelligence theories as well as many investigations into lay perceptions of multiple intelligences (Furnham 2004; Furnham and Petrides 2006). Gottfredson (2003a, b) provided a devastating attack on the evidence for Sternberg’s multiple intelligence theory (called Practical Intelligence Theory). There have also been conceptual and theoretical reviews of Gardner’s theory (Chen 2004; Cuban 2004; Eisner 2004; Klein 1997; Kallenbach and Viens 2004; Morgan 1996; Shearer 2004), as well as empirical critiques (Visser et al. 2006a, b). Gardner (2006a, b) has replied to many of his critiques, though more with argument than data.

There are essentially three major controversies around the concept of multiple intelligences. First, the extent to which they are inter-correlated to provide evidence of general intelligence. Second, how they can or should be measured (i.e. whether preference vs power tests could be used). Third, whether such concepts as inter or intrapersonal intelligence could justifiably be described as an intelligence as opposed to a social skill or a personality trait.

There has also been an amazingly increases in the number of multiple intelligences ‘discovered’. Furnham (2005) counted over twenty, including ‘network’, ‘intuitive’ and ‘political’ intelligence. Certainly one of the most controversial issues is about how to measure some, each or all of the multiple intelligences though specific tests of abilities akin to these intelligences do exist and have been used (Visser et al. 2006a, b).

Petrides and Furnham (2003) who developed a self-report measure of emotional intelligence distinguished between trait and ability tests of emotional intelligence. They have often preferred to call this measure ‘trait emotional self-efficacy’. Some people would argue that all genuine intelligence tests must be power tests (timed tests with definite correct answers) in order that they measure some ability or intelligence. That is, they need to be timed tests with correct and incorrect answers to all questions (Brody 2004). Others have argued that self-report measures are acceptable, primarily because emotional intelligence is a personality trait and that emotional perception and regulation are essentially subjective issues not open to objective testing. Indeed this debate can be seen as a manifestation of the typical vs maximal distinction. At least in the highly popular area of emotional intelligence there has been a great deal of effort and energy devoted to devising tests. Indeed Perez et al. (2004) listed five ability and fifteen trait measures of emotional intelligence with highly variable, and sometimes very limited, psychometric data in support of them. There appears to be a considerable emotional intelligence test ‘industry’ but this has not yet spilled over to many of the other intelligences.

While Gardner (1983, 1999, 2006a, b) appears to be less interested in developing tests to assess his well known 7, 8 or 10 multiple intelligences, Sternberg et al. (2000) have attempted measurement of his less ambitious trilogy of multiple intelligence. The satisfactory nature of these measures of tacit intelligence are however very much in doubt (Gottfredson 2003a, b).

Curr Psychol (2009) 28:225–239 227227

This paper sets out to examine the concurrent and discriminant validity of a self- report measure of eight of Gardner’s (1999) multiple intelligences. This study demonstrates how one may begin to validate such a test constructed for people to self-evaluate their own ‘multiple intelligences’. This study therefore tests a self- report preference, ‘trait’ measure rather than an ability ‘power’ measure of multiple intelligence. There is a long history of objective personality test but a far shorter history of subjective ability tests (Furnham 2008). Whilst many may argue it is more appropriate to develop and test a power or ability measure of the multiple intelligences that is another and rather different goal.

This study will use three types of tests to gain some insight into the concurrent validity of this new test. First, most self-report measures need to establish their position in personality space (Petrides and Furnham 2006). A great effort by trait theorists over 50 years has attempted to map personality space. Most personality theorists believe that it is important for any new measure to evidence its discriminant validity and uniqueness by establishing itself in that space. Whilst there yet remain different personality models, this study will attempt to map this new measure in Big Five Personality space currently the dominant model but of course not without its critics (Furnham (2008). Hence the multiple intelligence test scores will be compared to a valid measure of the Big Five (Costa and McCrae 1992). Petrides and Furnham (2003, 2006) have argued that trait emotional intelligence is a second order (level II) trait within the ‘Big Five’. Second, a measure of crystallized intelligence (namely General Knowledge) will be used to establish the relationship between all eight ‘multiple intelligences’ and a validated, power test of “actual” intelligence (Chamorro-Premuzic et al. 2006; Furnham and Chamorro-Premuzic 2006). Third, because of the way multiple intelligence is defined and construed it may well be that they are related to learning/thinking styles which in turn have been shown to be logically related to intelligence. Fourth, despite well established problems of self-validational assessment this study will examine the relationship between self-reported behaviours, supposedly measures of multiple intelligence as well as participants actual self-estimates on the normal distribution.

The three tests chosen for this concurrent and predictive validity study were chosen for three reasons. First, to have both ability and preference measures. Second to have a measure of the Big Five, currently the most widely accepted theoretical structure for personality researches. Third, to have a learning styles measure given that multiple intelligence is seen to be a cognitive style or preference variable. In addition this study compares global self estimates of each of the eight multiple intelligences with the scores from each section of the questionnaire.

Method

Participants

In total, 187 students (77 male) from three British universities took part in this study. Their mean age was 21.71 years (SD=2.1 years). They were all fluent in English. Half were psychology students and half from a wide variety of other disciplines.

228 Curr Psychol (2009) 28:225–239

Materials

Multiple Intelligences Test (Haselbauer 2005) This is a 79-item test where items are responded to on an eight point scale where 8 = Strongly Agree is 8 and 1 = Strongly disagree. The author, founder of the International High IQ Society, has constructed various power tests of particular intelligence (i.e., verbal mathematical, visual, spatial). To illustrate the nature of the questionnaire, the 13 items making up verbal/ linguistic intelligence they are presented here: “I own a large collection of books; I enjoy playing with words and puns; Word games such as Scrabble are fun; I preferred school subjects such as English and Social Studies to Maths or Science; I enjoy a good debate about daily news items; I am critical of grammatical errors in others’ writing; I have a voluminous vocabulary and seek to augment it with further study; I like to impress people with my erudition; I would enjoy writing a grant proposal, or other lengthy essay to achieve a goal; I love to read; Poetry is highly rewarding when reading or writing it; I can fluently speak more than one language; I have received compliments from supervisors regarding the quality of my writing or speaking efforts”. Examples from the other scales are given below. Musical: “I love to listen to music”. Logical/Mathematical: “I like to argue with others about facts and ideas”; “I enjoy mazes and jigsaw puzzles”. Spatial: “I perform better in geometry than algebra”; “I am good at explaining mental images to other people”. Interpersonal: “I think of myself as a ‘people person’”; “I love to make others laugh”. Intrapersonal: “I set realistic goals for myself”; “I take pride in my identity”. Body Kinesthetic: “I regularly exercise”; “I think better when I pace back and forth”. Naturalistic: “I love walking outside”; “I intuitively recognise patterns and similarities.”

The NEO Personality Inventory—Revised (NEO-PI-R; Costa and McCrae 1992) This 60-item, non-timed questionnaire which measures 5 primary personality factors (domains), i.e., Neuroticism, Extraversion, Openness to Experience, Agreeableness, and Conscientiousness. Items involve questions about typical behaviours or reactions which are answered on a five-point Likert scale, ranging from Strongly disagree to Strongly agree. The manual shows impressive indices of reliability and validity.

General Knowledge Test (Irwing et al. 2001) This is a 72 item questionnaire that measures knowledge of six areas: literature, general science, medicine, games, fashion and finance. Each area is measured by ten items, and each correct response is awarded one point (in a few cases, there are two correct responses and not one). The internal reliability of the test for the present sample was a=.78. Various studies have shown scores on this test correlate significantly with other validated tests of both fluid and general intelligence (Chamorro-Premuzic et al. 2006; Furnham and Chamorro-Premuzic 2006).

Study Process Questionnaire (Biggs 1987) This is a 42-item questionnaire that yields six scores. There are three approaches and two components. The first component is learning motive (why students learn): the second learning strategy (how students learn). The three approaches are surface (a reproduction of what is

Curr Psychol (2009) 28:225–239 229229

taught to meet the minimum requirement), deep (a real understanding of what is learned), and achieving (designed specifically to maximise grade). Thus there are six subscales: surface-motive, surface-strategy, deep-motive, deep-strategy, achieving- motive, achieving-strategy; and each subscale has seven questions relating to it. The questionnaire has been repeatedly shown to have satisfactory internal reliability, content, construct and predictive validities. The questionnaire is frequently still used in many studies in applied, educational and personality psychology.

Self-Assessed Multiple Intelligence (Furnham 2001) This is a simple one page questionnaire that shows people the IQ bell curve with standard-deviations and description for various scores. Using the bell curve scores they are then requested to estimate their own score on overall IQ (general intelligence) as well as the 10 Gardner intelligences that each has a one-line description of that the ‘intelligence’ means. Thus verbal intelligence was described as “the ability to use words”, while interpersonal intelligence was described as “the ability to understand other people”. This questionnaire has been used in over 30 studies in the area (Furnham and Chamorro-Premuzic 2004). They have shown that overall self-assessed intelligence tends to correlate .30<r<.50 with validated, power tests of intelligence (Chamorro- Premuzic et al. 2004). Various tests of the validity of self-reported multiple intelligence have been conducted.

Procedure

Participants were tested in large groups soon after arriving at university. They were given 20 min to complete the General Knowledge test but allowed to complete all other preference tests, presented in a booklet, in their own time.

Results

Because the sample consisted of students it is important to show they are in the normal range on all measures. Table 1 shows that this is indeed the case. Table 2 shows the basic data from the eight scales of the Multiple Intelligence Test. Table 1 also shows the mean scores for males and females. There was no significant difference except on Intrapersonal intelligence, F (1, 189) = 9.13, p<.01. Over half had alpha coefficients lower than usual traditional cut off score of .70. None fall below .50, however. Psychometricians differ in the importance they attach to alpha co-efficients and their agreement on a minimum desirable cut-off point. For many, .70 is the minimum cut-off point to ensure sufficient internal reliability, while for others a high alpha becomes more a measure of redundancy.

It could be argued that the alphas reflect the homogeneous-heterogeneous or coherent nature of the different multiple intelligence concepts. Thus, while linguistic and musical intelligence seem both well known concept and internally coherent and consistent this is much less true of a much vaguer and incoherent concept like intrapersonal intelligence. It may therefore be more logical to measure test-retest rather than internal reliability to get a good measure of the tests replicability.

230 Curr Psychol (2009) 28:225–239

In all, 18 participants completed the multiple intelligence test twice over a 6 week period. The correlations between the two test periods ranged from r=.73 to r.85 for the different scales. This is probably adequate but these need replication on a bigger sample for a longer interval between testing sessions.

Overall then the internal and rest-retest reliability for this test suggests that its overall poor reliability might well threaten its validity. Table 3 shows the intercorrelation matrix. Of the 28 correlations five were r>.40 and seven .30<r<.20. This indicated that the scales overlapped albeit not to a great extent. ‘Naturalistic intelligence’ appeared to correlate with all intelligences while correlations with ‘Linguistic intelligence’ were modest. To some extent this may be seen as evidence of general, rather than multiple intelligence. Indeed a principle components analysis (available from the author) did yield one very large single factor.

Table 1 Means and standard deviations and norms for the three major tests used in this study

This Study Student Norms

Big Fivea M SD M SD

Neuroticism 23.59 8.73 24.56 7.87

Extraversion 30.03 6.35 30.49 5.84

Openness 28.99 6.23 27.82 5.85

Agreeableness 36.48 7.45 30.14 5.40

Conscientiousness 30.25 6.68 3.07 6.79

Learning styles

Surfaceb 42.85 6.50 43.7 4.31

Deep 45.41 7.85 43.7 5.11

Achieving 46.48 7.45 41.9 5.11

General Knowledgec 30.08 9.72 29.74 9.70

a NEO Professional Manual, p. 78 b Manual Norms c Furnham et al. (2007)

Table 2 Means, standard deviations and internal reliabilities of the eight scales

“Intelligence” N of Q Alpha M SD

Linguistic 13 .83 59.69 15.59

Musical 11 .80 58.20 15.17

Mathematical 7 .60 36.14 7.52

Spatial 9 .51 47.96 7.43

Interpersonal 11 .75 61.02 9.89

Intrapersonal 9 .51 46.84 7.53

Bodily Kinesthetic 10 .61 47.13 10.18

Naturalistic 9 .64 41.42 9.58

Each question was responded to on a scale where 8 is Strongly Agree and 1 is Strongly Disagree

Curr Psychol (2009) 28:225–239 231231

Table 4 shows the result of the Varimax rotated factor analysis, which attempted to explore the factor structure. This analysis was based on the assumption that these subscale exist. Prior to this a factor analysis was done at the item level both exploratory and confirmatory. However, the participant to item ratio of under 3 to 1 mean the analysis was probably too unstable. Hence analysis was done at the subscale level. There is a debate and disagreement amount the optimal or more often minimum participants to ‘item’ ratio in order to ensure factor analytic results are stable. Various ‘rules of thumb’ are applied like 10 to 1. In this analysis the ratio was 23 to 1 because the subscale scores were used. However it is freely admitted that it would have been most desirable to have a much larger N, one approaching 250 to 300 participants.

The factor analysis (Table 4) certainly did not confirm Gardner’s three-fold higher order classification. The three fold factor structure is difficult to interpret. It fits neither the higher order classification of Gardner’s theory (Linguistic and Mathematical; Musical, Bodily-Kinesthetic, Spatial, Natural, Inter-, and Intraper- sonal) nor the extensive factor analytic work on self-rated multiple intelligences which tend to show three factors (Linguistic, Mathematical, Spatial; Bodily- Kinesthetic, and Musical; Inter -and Intra-personal). Further exploration of the factor structure using oblique and orthogonal exploratory or confirmatory methods did little to help interpretation. This result could not be attributable to sex differences

Table 3 Intercorrelations between the eight scales

L M Mm S Ie Ia Bk N

Linguistic (L)

Musical (M) .17*

Mathematical (Mm) .00 .22**

Spatial (S) .22** .11 .12

Interpersonal (Ie) .17* .31** .14 .41**

Intrapersonal (Ip) .11 −.01 .34** .41** .20** Bodily-K (Bk) .18* .35** .32** .36** .40** .29**

Naturalistic (N) .24** .32** .40** .32** .23** .25** .49**

**p<.01; * p<.05

Mathematical .76

Naturalistic .72

Bodily-K .63

Musical .63

Linguistic .67

Interpersonal .66

Spatial .62 .57

Intrapersonal .83

Eigenvalue 2.82 1.17 1.11

Variance 35.51 14.67 13.96

Table 4 Factor analytic results (VARIMAX rotation) of eight multiple intelligences

232 Curr Psychol (2009) 28:225–239

in the factor scores. This factor solution however adds yet more doubt to the scientific merit of this instrument.

Table 5 shows the concurrent validity between the eight scales and the nine scores derived from the General Knowledge, Learning Styles, and Personality Tests. Only two scales were significantly correlated with General Knowledge, albeit modestly. Linguistic intelligence was correlated with General Knowledge, all three learning styles and two Big Five variables. Those with high Linguistic intelligence were Open, preferred a Deep Learning Style and had, wide General Knowledge. It is the profile of a ‘bright’ student.

Musical intelligence was correlated only with Neuroticism and Openness which makes sense given the personality profile of successful musicians (Wilson 2002). Mathematical intelligence was correlated positively with General Knowledge, Deep and Achieving learning styles as well as Agreeableness and Conscientiousness. Spatial intelligence was correlated significantly only with Extraversion and Openness, while Interpersonal Intelligence was correlated only with Extraversion. There were five significant correlates of Intra-personal intelligence. The more Interpersonally intelligent were Agreeable, Conscientious, Stable, Extraverts with a preferred Deep Learning Style. Both Bodily-Kinesthetic and Naturalistic intelligence had similar correlates namely Extraversion, Openness, Agreeableness, and Deep Learning Style.

Table 6 shows correlations believed the eight test-derived scores (the columns) and the ten self-estimates (rows). Those correlations in bold show where the two should be logically most closely related. The were all positive and significant ranging from r=.18 for Naturalistic to r=.56 for Musical intelligence. Though for some test-derived scores more than one self estimated intelligence was significant (particularly Linguistic and Intrapersonal intelligence) for others (Musical and Mathematical intelligence) the two overlapping measures where the only significant correlation.

Table 5 Correlates of the eight scales with general knowledge, learning style and Big Five traits

Learning Styles Personality

GK Surface Deep Achiev N E O A C

Linguistic .23** −.22** .39** .17* .02 .06 .51** .17* .05 Musical −.03 −.09 .10 .01 .30** .11 .27** .01 −.09 Mathematical .18** .00 .24** .32** −.12 .05 .11 .32** .25** Spatial .09 .03 .13 .10 −.10 .27** .21** .10 −.07 Interpersonal .07 .07 .08 .01 −.06 .58** .14 −.02 .09 Intrapersonal .05 −.07 .26* .21 −.37** .19** .07 .22** .31** Bodily K .09 −.12 .18* .14 −.08 .31** .21** .14 .06 Naturalistic .03 −.07 .32** .17 .05 .19** .32** .17* −.03

N Neuroticism, E Extraversion, O Openess, A Agreeableness, C Conscientiousness

**p<.01; *p < .05

Curr Psychol (2009) 28:225–239 233233

Table 7 shows two sets of regressions both with General Knowledge (GK) as the criterion variable. In the first the test scores of the eight ‘intelligences’ were predictor variables while in the second it was the self-estimates that were predictor variables in the regression. Both showed fairly similar results. Those with higher Linguistic and Mathematical Intelligence received higher GK scores. The same two factors were significant predictors when using self-estimated scores. However this analysis also showed that those who were low on Bodily Kinesthetic and Interpersonal intelligence also scored higher on GK.

Table 8 shows the results of five regressions. In each the Big Five were the criterion variable and the eight multiple intelligence scores the predictor variables. All were significant and indicated that between a third and a sixth of the variance

Table 6 Partial correlations (Sex, Age, GK) partialled out between self-estimated and test derived multiple intelligence

Test derived

Self-Estimate Linguistic Musical Mathemat Spatial Interper Intraper Bodily-K Nat

Verbal .44** −.05 .04 .21** .14 .23** .06 .05 Mathematical −.18 .08 .51** .03 −.02 .16* .06 .09 Spatial .15 −.05 .10 .37** −.04 .28** .13 .16*` Musical .07 .56** .12 .08 .04 .08 .13 .21**

Bodily-Kinectic .14 .08 .00 .21** .21** .12 .38** .21

Inter-Personal .28** .10 .14 .30** .35** .21** .18** .11

Intra-Personal .22** .00 .03 .28 .12 .31** .11 .07

Existential .33** .06 .17* .25** .14 .34** .16* .17**

Spiritual .27* .06 .11 .18 .12 .25** .17 .25**

Naturalistic .09 .07 .12 .16* −.01 .28* .03 .18*

**p<.01; *p < .05

Table 7 Regressions of the multiple intelligences onto the criterion variable general knowledge

A) Test derived Beta t B) Self estimates Beta t

Linguistic .27 3.59*** Verbal .37 5.80***

Musical −.13 1.58 Mathematical .12 2.05* Mathematical .25 2.94** Spatial −.03 0.50 Spatial .06 0.72 Musical .04 0.65

Interpersonal .00 0.28 Bodily-K −.14 2.30** Intrapersonal −.11 1.29 Interpersonal −.14 2.02* Bodily-K .06 0.68 Intrapersonal .09 1.21

Naturalistic −.11 1.15 Exist −.02 0.29 F (8, 179) = 2.74** Adj R2=.07 Spiritual .00 0.01

Naturalistic −.03 0.58 F (10, 372) = 6.79*** Adj R2=.14

234 Curr Psychol (2009) 28:225–239

was accounted for. In most regressions it was one or two of the multiple intelligences that was most closely related to each of the Big Five.

Those with low Intrapersonal but high Naturalistic intelligence tended to score high on Neuroticism. The only predictor of Extraversion was Interpersonal intelligence. Openness was predicted by three multiple intelligences all positively: Linguistic, Musical and Naturalistic. Those high on Mathematical and Linguistic Intelligence tended to score high on Agreeableness. Those scoring high on Intrapersonal and Mathematical intelligence but low on Naturalistic Intelligence tended to score high on Conscientiousness.

Discussion

This study attempted possibly the first concurrent validity study of a self-report measure of multiple intelligence. There remains those who insist that the term ‘multiple intelligence’ is always put in inverted commas to denote that we are not talking about or measuring intelligence in the conventional manner. Thus these are seen as specific aptitudes or broad abilities even preferences rather than what might be seen as accurate measures of Carroll’s (1993) level 2 intelligence factors. Some suggest that any intelligence or ability must be measured by a power test and that any preference test is little more than a personality type measures (Brody 2004). In this sense this is a trait multiple intelligence measure not an ability intelligence test measure.

There is a growing literature on people’s ability and accuracy in estimating their own ability and personality scores (Chamorro-Premuzic et al. 2004; Furnham and Chamorro-Premuzic 2004). This study used self-estimates of multiple intelligences as evidence of construct and concurrent validity for this new measure. However the test-derived eight scores should not be seen as self-estimates but self-reports on

Table 8 Regression of the Five Traits onto the eight multiple intelligences

N E O A C

Beta t Beta t Beta t Beta t Beta t

Lingusitic .04 0.45 −.07 1.07 .44 6.63*** .17 2.24* .09 1.22 Musical −.07 0.93 −.08 1.22 .14 1.96* −.09 1.08 −.09 1.18 Math −.04 0.49 −.14 1.99 .02 0.29 .25 3.06** .23 2.87** Spatial .02 0.23 .05 0.72 .09 1.10 −.03 0.35 .02 0.26 Inter-Pers −.01 0.14 .59 8.46*** −.05 0.67 −.11 1.32 −.11 1.34 Intra-Pers −.41 4.97*** .12 1.01 −.06 0.87 .13 1.48 .29 3.50*** Bodily-K .04 0.47 .08 1.06 .00 0.78 .08 0.86 .04 0.47

Natural .21 2.32* .10 1.28 .17 2.06* .05 0.52 −.17 1.96* F(8,179) 4.47*** 13.63** 10.25** 3.73* 4.83*

Adj R2 0.13 0.36 0.29 0.11 0.15

N Neuroticism, E Extraversion, O Openess, A Agreeableness, C Conscientiousness

***p<.001 **p<.01 *p< .05

Curr Psychol (2009) 28:225–239 235235

attitudes and behaviours. In this sense it is strictly a trait rather than an ability measure.

The results shown in Tables 1 to 4 indicate serious problems and limitations with this measure. Over half of the eight scales failed to reach the generally accepted level internal reliability (Alpha=.70). Limited evidence of test-retest reliability did not suggest the test was particularly reliable over time. Further, the scales were nearly all positively inter-correlated with only five of the 28 correlations failing to be significant. Indeed those who support the concept of general, as opposed to multiple, intelligence may argue that even for self-report (trait) measures of intelligence there is evidence of a general (highly intercorre- lated) factor, rather than of multiple intelligence. Third, the factor analytic results in Table 4 fail to support Gardner’s three-fold classification of the multiple intelligences. Furthermore the factor structure was not easily interpretable. As noted this maybe a function of the size of the sample and factor analysis of this data on a bigger sample would certainly be desirable. This suggests as Visser et al (2006a, b) have demonstrated using power tests that there is little evidence for multiple intelligences using preference tests. Scores are quite simply too highly correlated to argue for distinct and unique abilities/intelligences.

Tables 5 and 6 also provide some evidence of concurrent and construct validity. Correlations between self-estimated, specific, multiple intelligence and trait multiple intelligence scores were positive and significant. The clearest results were for Mathematical and Musical intelligence while for Intrapersonal intelligence estimate six of the eight correlations were significant. To some extent these results may be seen as evidence of method and item overlap rather than evidence of the existence of these measures.

The current literature on the correlation between ability intelligence test scores (i.e. traditional IQ power tests) and personality test scores (i.e. the big five) shows that Openness is the strongest correlate (Chamorro-Premuzic et al. 2004). They also show that Neuroticism tends to correlate negatively with test scores. The data in Table 5 is not dissimilar because it shows Openness positively and significantly correlated with five of the eight multiple intelligence scales. However the table did illustrate that Extraversion is positively correlated with five of the eight multiple intelligences. Whilst this makes good sense for Interpersonal intelligence it is much less clear what the explanation might be for Spatial or Naturalistic intelligence.

Perhaps the most interesting finding for the test validity is also in Table 4 namely the correlations between General Knowledge and the eight multiple intelligence scores. General knowledge has long been thought of, and empirically demonstrated to be, a good index of (crystallised) intelligence (Chamorro- Premuzic et al. 2006; Irwing et al. 2001). Only two of the eight measures: verbal and mathematical were significant correlates. The regression analysis supported this finding. That is, only two of the self-report multiple intelligence test scores predicted ‘actual’ test-derived intelligence. Interestingly it was the two scores— namely Linguistic and Mathematical intelligence—that Gardner (1999) himself notes were the types of intelligence that traditionally dominate intelligence tests. Whilst these two measures only accounted for 7% of the variance it does seem clear that self-report measures of linguistic/verbal and logical/mathematical ability do predict performance on crystallised intelligence test scores.

236 Curr Psychol (2009) 28:225–239

To some extent this data also reflects the self-estimated intelligence literature. Thus if people make an overall (g) as well as multiple intelligence self-estimate and the former is regressed onto the latter consistently the finding is that only two— namely Mathematical and Verbal—are the only significant predictors. That appears to suggest that people think Mathematical and Verbal intelligence are “real” intelligences and the remainder of the Gardner list not strictly related to IQ as measured by intelligence tests. The data in this study gives credence to this suggestion.

The final analysis sought to determine the relationship between the Big Five and the multiple intelligence eight. The results suggested those with low Intrapersonal intelligence tended to be Neurotic, while those with high Interpersonal intelligence tended to be Extraverted. Openness was most closely associated with Linguistic or Verbal intelligence. The interpretation for Agreeableness and Conscientiousness was much less clear.

A central question for all researchers interested in developing new self-report measures of behaviours is where to locate it in Big Five factor space and whether the new scale is essentially simply measuring one or more of the Big Five factors using different terminology (Petrides and Furnham 2003). This study certainly suggested that Interpersonal intelligence overlapped considerably with Extraver- sion and Openness with Linguistic intelligence. It would certainly be very interesting to have examined multiple intelligence correlates with the Big Five at the facet level. It may well be that like emotional intelligence some of these multiple intelligences are essentially profiles of various combinations of Big Five facets. However the size of the correlations shown in Table 4 would suggests that this measure is tapping into something more than just the Big Five personality traits.

Overall these results suggest there is weak support for the internal reliability and validity of this new self-report measure of the eight multiple intelligence. The instrument requires further refinement and empirical testing if it is to be in research. Two recommendations are made: first that an item analysis be used to remove items that reduce the overall alpha per scale; and second, that the predictive validity of the measure be assessed by examining its ability to predict behaviour or observer data that is conceptually related to each of the eight scale scores. Ideally this will be non self-report data, preferably test/behavioural data but even observational data would be desirable. The first was in fact tested but failed to improve reliabilities noticeably which does not bode well for this instrument. This may be because they were relatively so few items per scale and new items have to be written.

However it should be recognised that there will be those who argue that because there is such poor empirical evidence in support of multiple intelligent theory it is wasted effort to develop self-report measures of multiple intelligence (Gottfredson 2003a, b; Visser et al. 2006a, b). Whilst it may be acknowledged that most of the empirical evidence on multiple intelligence suggests they are highly intercorrelated and that some are little more than personality traits the idea of a valid self-report measure of intelligence that has predictive validity in the same sense of power-tests of intelligence remains a clearly worthwhile goal.

Curr Psychol (2009) 28:225–239 237237

References

Biggs, J. (1987). Study process questionnaire manual. Hawthorne: Australian Council for Educational Research.

Brody, N. (2004). What cognitive intelligence is and what emotional intelligence is not. Psychological Inquiry, 15, 234–238.

Carroll, J. (1993). Human cognitive abilities: A survey of factor analytic studies. Cambridge: Cambridge University Press.

Chamorro-Premuzic, T., Furnham, A., & Moutafi, J. (2004). The relationship between estimated and psychometric personality and intelligence scores. Journal of Research in Personality, 38, 505–513.

Chamorro-Premuzic, T., Furnham, A., & Ackerman, P. (2006). Ability and personality correlates of general knowledge. Personality and Individual Differences, 41, 419–429.

Chen, J.-Q. (2004). Theory of multiple intelligences: is it a scientific theory. Teachers’ College Record, 106, 17–23.

Costa, P., & McCrae, R. (1992). Revised NEO personality inventory manual. Odessa: PAR. Cuban, L. (2004). Assessing the 20-year impact of multiple intelligences on schooling. Teachers’ College

Record, 106, 140–146. Eisner, E. (2004). Multiple intelligences: its tensions and possibilities. Teachers’ College Record, 106, 31–

39. Furnham, A. (2001). Self-estimates of intelligence. Personality and Individual Differences, 31, 1381–

1405. Furnham, A. (2004). Are lay people lumpers or splitters? Learning and Individual Difference, 14, 153–

168. Furnham, A. (2005). Gender and personality differences in self and other ratings of business intelligence.

British Journal of Management, 16, 91–103. Furnham, A. (2008). Personality and intelligence at work. London: Routledge. Furnham, A., & Chamorro-Premuzic, T. (2004). Estimating one’s own personality and intelligence scores.

British Journal of Psychology, 95, 146–160. Furnham, A., & Chamorro-Premuzic, T. (2006). Personality, intelligence and general knowledge. Learning

and Individual Differences, 16, 79–90. Furnham, A., & Petrides, K. (2006). Parental estimate of five types of intelligence. Australian Journal of

Psychology, 56, 10–17. Furnham, A., Christopher, A., Garwood, J., & Martin, A. (2007). Approaches to learning and the

acquisition of general knowledge. Personality and Individual Differences, 43, 1563–1571. Gardner, H. (1983). Frames of mind. New York: Basic Books. Gardner, H. (1999). Intelligence reframed. New York: Basic Books. Gardner, H. (2006a). On failing to grasp the core of MI theory. Intelligence, 34, 503–505. Gardner, H. (2006b). Responses to critiques of multiple intelligence theory. In J. Schaler (Ed.), Gardner

under fire. Chicago: Open Court. Gottfredson, L. (2003a). Dissecting practical intelligence theory: its claims and evidence. Intelligence, 31,

343–397. Gottfredson, L. (2003b). On Sternberg’s “Reply to Gottfredson”. Intelligence, 31, 415–424. Haselbauer, N. (2005). What’s your IQ?. New York: Barnes and Noble Books. Irwing, R., Cammock, T., & Lynn, R. (2001). Some evidence for the existence of a general factor of

semantic memory and its components. Personality and Individual Differences, 30, 857–871. Kallenbach, S., & Viens, J. (2004). Open to interpretation: multiple intelligences theory in adult literacy

education. Teachers’ College Record, 106, 58–66. Klein, P. (1997). Multiplying the problems of intelligence by eight: a critique of Gardner’s Theory.

Canadian Journal of Education, 22, 377–394. Morgan, H. (1996). An analysis of Gardner’s Theory of multiple intelligence. Roeper Review, 18, 263–

269. Perez, J., Petrides, K., & Furnham, A. (2004). Measuring trait emotional intelligence. In R. Schultz & R.

Roberts (Eds.), Emotional intelligence: An international handbook (pp. 181–201). Gottingen: Hogrefe and Huber.

Petrides, K., & Furnham, A. (2003). Trait emotional intelligence. European Journal of Personality, 17, 39–52.

Petrides, K. V., & Furnham, A. (2006). On the dimensional structure of emotional intelligence. Personality and Individual Differences, 29, 311–320.

238 Curr Psychol (2009) 28:225–239

Shearer, B. (2004). Multiple intelligences theory after 20 years. Teachers’ College Record, 106, 2–16. Sternberg, R., Forsythe, G., Hedlund, J., Horvath, J., Wagner, R., Williams, W., et al. (2000). Practical

intelligence in everyday life. New York: Cambridge University Press. Visser, B., Ashton, M., & Vernon, P. (2006a). Beyond g: putting multiple intelligences theory to the test.

Intelligence, 34, 487–502. Visser, B., Ashton, M., & Vernon, P. (2006b). g and the measurement of multiple intelligences: a response

to Gardner. Intelligence, 34, 507–510. Wilson, G. (2002). The psychology of performing artists. London: Whurr.

Curr Psychol (2009) 28:225–239 239239

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.