Vous êtes sur la page 1sur 10

Australian Journal of Psychology, Vol. 62, No. 1, March 2010, pp. 513.

Concurrent and incremental validity of three trait emotional intelligence measures

KATHRYN JANE GARDNER & PAMELA QUALTER


School of Psychology, University of Central Lancashire, Preston, Lancashire, UK

Abstract This study explored concurrent and incremental validity of three trait emotional intelligence measures: the Schutte Emotional Intelligence Scale, Multidimensional Emotional Intelligence Assessment, and Trait Emotional Intelligence Questionnaire. A total of 307 participants were drawn predominantly from community and student populations. Concurrent criterion validity of the measures varied depending on whether emotional intelligence (EI) was assessed as a lower, middle or higher level construct, with validity coefcients being larger for the former. In all cases, the Trait Emotional Intelligence Questionnaire was the superior predictor of multiple psychological criteria. At the higher level of assessment, incremental validity beyond (a) age, gender and the Big Five, and (b) the remaining two EI measures, was also superior.

Keywords: Personality, trait emotional intelligence, validity

Trait emotional intelligence (EI) is a constellation of emotion-related self-perceptions at the lower levels of personality hierarchies (Petrides, Pita, & Kokkinaki, 2007) and is typically assessed via selfreport questionnaire. An abundance of trait EI measures now exist, with their validity being subjected to much empirical scrutiny (e.g., Petrides, Perez-Gonzalez, & Furnham, 2007; Saklofske, Aus tin, & Minski, 2003). The papers in this special issue provide further examples of these validation studies, covering the associations of trait EI measures with processing style and subjective wellbeing (Schutte et al., 2010), stress and coping (Austin, Saklofske, & Mastoras, 2010), problem behaviours, coping and academic success in adolescence (Downey, Johnston, Hansen, Birney, & Stough, 2010; Hogan et al., 2010), and also EI proles of university students (Sanchez-Ruiz, Perez-Gonzalez, & Petrides, 2010). Few studies, however, have compared the validity of multiple trait EI measures within one study (although some do exist e.g., Bastian, Burns, & Nettelbeck, 2005; Brackett & Mayer, 2003). This hinders conclusions regarding the validity of individual EI instruments due to methodological variation.

Our overriding aim was to conduct such a study. We focused on two pervasive and contentious validity issues. First, we examined concurrent criterion validity of the measures in predicting a range of theoretically relevant psychological constructs. Such ndings aid selection of one trait EI measure over another for predicting specic criteria. As noted by Petrides, Perez-Gonzalez, et al. (2007), this is important to establish the nomological network of trait EI. Second, incremental validity was tested to explore the range of criteria in which the measures make an incremental contribution. The usefulness of any test is questioned if it cannot account for additional variance in relevant criteria (Garb, 1984), ndings that impact on whether a test is selected over other available measures. When testing incremental validity of trait EI measures, the primary concern has been whether they are incrementally valid beyond theoretically similar constructs such as the Big Five: Neuroticism, Extraversion, Openness, Agreeableness and Conscientiousness (e.g., Saklofske et al., 2003). More recently Petrides, Perez-Gonzalez, et al. (2007) suggested that because trait EI is a lower order personality trait (lower

Correspondence: Dr K. J. Gardner, School of Psychology, University of Central Lancashire, Preston, Lancashire PR1 2HE, UK. E-mail: kjgardner@uclan.ac.uk ISSN 0004-9530 print/ISSN 1742-9536 online The Australian Psychological Society Ltd Published by Taylor & Francis DOI: 10.1080/00049530903312857

K. J. Gardner & P. Qualter For example, the SEIS has been associated with loneliness and life satisfaction (Saklofske et al., 2003), the MEIA with emotional resilience and life satisfaction (Tett et al., 2005), and TEIQue with aggression and life satisfaction (Petrides, Perez-Gonzalez, et al., 2007). The SEIS and TEIQue have also been related to clinical variables such as personality disorder (Gardner & Qualter, 2009; Petrides, Perez-Gonzalez, et al., 2007). These relationships are typically moderate in strength, but are often higher for broadly dened measures such as the TEIQue. This pattern is consistent with the results of a recent meta-analysis (Schutte, Malouff, Thorsteinsson, Bhullar, & Rooke, 2007). There is often inconsistency, however, across studies regarding the use of lower level trait EI subscales or higher level global scores, making it difcult to compare ndings. Why such comparisons are made difcult is explained by Ones and Viswesvaran (1996): when higher level facets or global constructs are used, measurement reliability is maximised and there is a broader bandwidth. This broader bandwidth can result in the prediction of a wider range of criteria, but instrument delity (similar to validity) tends to be moderate. Global scores can also mask the importance of lower level EI facets and attenuate relationships. In contrast, lower level subscales have narrow bandwidths and utilise the nonrandom variance specic to these narrow traits; they therefore typically afford greater predictive precision due to increased instrument delity (e.g., Ashton, Jackson, Paunonen, Helmes, & Rothstein, 1995; Paunonen & Ashton, 2001). Clearly, it is important for any comparison of trait EI measures to consider the level at which each test is assessed. All three trait EI measures have also shown some evidence of incremental validity beyond personality (e.g., SEIS: Saklofske et al., 2003; MEIA: Tett et al., 2005; TEIQue: Petrides, Perez-Gonzalez, et al., 2007). For the SEIS in particular incremental power appears limited and may be criterion specic (e.g., Chapman & Hayslip, 2005). As with concurrent validity, there is evidence that validity estimates differ as a function of global versus subscale score usage (e.g., Chapman & Hayslip, 2005). Moreover, the use of EI subscales beyond the Big Five creates an imbalance between the bandwidth and delity at which each measure is assessed (Chapman & Hayslip, 2005). In sum, differential treatment of trait EI measures regarding global versus subscale score usage across studies hinders our understanding of the incremental validity of these tests. The focus on incremental validity beyond personality has meant that little attention has been paid to the incremental validity of particular trait EI measures beyond others. Such analyses are also important and test the relative strengths of these measures in predicting criteria.

order traits being ones that make up the highest level traits of a hierarchical personality taxonomy), it should share large proportions of variance with the Big Five and only be incrementally valid beyond emotionally laden variables. Selected for inclusion in the study were three trait EI measures: the Schutte Emotional Intelligence Scale (SEIS) (Schutte et al., 1998), the Multidimensional Emotional Intelligence Assessment (MEIA) (Tett, Fox, & Wang, 2005), and the Trait Emotional Intelligence Questionnaire version 1.50 (TEIQue) (Petrides, 2009). The SEIS is one of the most widely used trait EI measures based on the earlier ability model of EI (Salovey & Mayer, 1990) and some research supports the assessment of four facets: optimism/mood regulation, appraisal of emotions, social skills, and utilisation of emotion (e.g., Saklofske et al., 2003). The MEIA (Tett et al., 2005) targets all facets outlined in the Salovey and Mayer (1990) model. Ten subscales load on three broader factors: (a) self orientation, consisting of motivating emotions, recognition of emotion in the self, regulation of emotion in the self and intuition versus reason; (b) emotional sharing, consisting of non-verbal emotional expression, empathy and mood redirected attention; and (c) other orientation, consisting of creative thinking, recognition of emotion in others and regulation of emotion in others. The TEIQue (Petrides, 2009) is a broadly dened comprehensive measure that covers all facets of trait EI as postulated by the Petrides and Furnham (2001) framework. The TEIQue includes four compound scales that encompass 15 subscales: (a) wellbeing: happiness, optimism and self-esteem; (b) self-control: control/emotion regulation, stress management and impulsiveness (low); (c) emotionability: emotion expression, empathy, emotion perception (self and others) and relationship skills; and (d) sociability: social competence, assertiveness and emotion management (others). Self-motivation and adaptability do not belong to any of the four factors. From the perspective of trait EI theory the three measures should produce compatible ndings (Petrides, Perez-Gonzalez, et al., 2007), irrespective of the underlying model. Validity estimates, however, may be superior for the TEIQue because the SEIS and MEIA assess a specic and limited set of trait EI facets. There may be similarity in results for the latter two measures given the same underlying model, which begs the question of which, if any, is a superior measure. Concurrent and incremental validity of the SEIS, MEIA and TEIQue All three measures (although less so for the MEIA) have been related to a range of psychological criteria.

Validity of trait emotional intelligence measures Aims Although multiple measures of a construct are useful, they are unlikely to be equally valid indicators (Grucza & Goldberg, 2007). This study extends the literature by providing an empirical evaluation of concurrent and incremental validity of multiple trait EI measures in predicting a range of criteria. In conducting these analyses we also addressed the issue of bandwidth and delity to provide a fairer comparison of validity across measures. The study had three aims: the rst aim was to compare concurrent criterion validity of the trait EI instruments with each assessed at the same hierarchical level or approximately comparable levels of specicity (Chapman & Hayslip, 2005). We considered the range of criteria predicted and the ability to predict each criterion (i.e., the size of the validity coefcients). Second, we aimed to provide a stringent approach to testing incremental validity in predicting criteria by partialling out variance from demographic variables (age and gender because men, women and different age groups may differ on trait EI e.g., Mikolajczak, Luminet, Leroy, & Roy, 2007; Saklofske et al., 2003) and the Big Five. Global EI scores were used here so as to avoid imbalance between bandwidth and delity of the EI and personality measures. Third, as a further test of incremental validity, using global EI scores we examined the capacity of one of the three measures to increment the other two when predicting the criteria. Criterion selection In selecting the criterion measures, variables were chosen that may relate to emotional experience and have some affective relevance. For example, physical aggression may indicate a failure to manage emotions; drug abuse and eating pathology attempts to regulate emotion; and evaluations of how happy or satised one is may be inuenced by low levels of trait EI. In addition, at the time of selecting these criteria (2004), the majority had not been investigated in terms of relationships with our three chosen EI measures. Negative associations were anticipated between trait EI and aggression, loneliness, alcohol and drug abuse and eating pathology. Positive associations were anticipated with indicators of wellbeing, measured by life satisfaction and happiness (note that effect sizes may vary as a function of affective relevance of the criterion: Petrides, Perez-Gonzalez, et al., 2007).

measures of trait EI, the Big Five and psychological functioning. The MEIA was completed on a separate website hosted by the test developer (R. Tett). To reduce respondent burden, the study was conducted in two sessions whereby participants could return to the website to complete the second half minutes, hours or days after completing Session 1. Participants A mixed community and student sample was used. The sample consisted of 310 (74 male) UK native English speakers aged 1879 (M 36.70 years; SD 12.05). Participants identied as submitting multiple data sets, that is, those with duplicate email addresses, were deleted. A large percentage were from the northwest of England (53.2%), and reported themselves as white British (89.3%). Most participants were educated up to General Certicate in Secondary (high school) Education level (94.5%). Occupations were diverse, with highest percentages working in health services (8.3%), other public sector (17.9%), academic sector (31.9%), or were not employed (9.3%). A total of 35% were students. Measures Schutte Emotional Intelligence Scale. The SEIS (Schutte et al., 1998) includes 33 items that assess global trait EI and four facets (as previously described). Participants responded using a 5-point Likert scale, ranging from strongly disagree to strongly agree. Internal consistencies in this sample were .90 (total score) and .63.86 (four subscales). Multidimensional Emotional Intelligence Assessment. The 116-item MEIA (Tett et al., 2005) includes 10 facets and three higher order EI factors (as previously described), rated on a 6-point Likert scale, ranging from strongly disagree to strongly agree. Internal consistencies in this sample were .95 (total score), .82.90 (10 subscales) and 89.91 (three higher order factors). Trait Emotional Intelligence Questionnaire. The 153item TEIQue (Petrides, 2009) assesses 15 trait EI facets, four higher order factors and a global score (as previously described), rated using a 7-point Likert scale, ranging from completely disagree to 7 completely agree. Internal consistencies in this sample were .97 (total score), .71.94 (15 subscales) and .90.95 (four higher order factors). International Personality Item Pool. This 50-item questionnaire is a broad-bandwidth public-domain personality scale adapted from the International

Method Design The study used a Web-based approach to data collection, for which the participants completed

K. J. Gardner & P. Qualter from strongly disagree to strongly agree. Internal consistency in this sample was .90. Procedure All participants completed the questionnaires via a website and provided demographic information. Participants were provided with incentives for participating. After Session 1 these included scores and feedback on their levels of happiness and life satisfaction. After Session 2 participants received scores and feedback on their trait EI (using the SEIS), an EI workbook (Qualter, Whiteley, & Gardner, 2003) and entry into a 100 prize draw.

Personality Item Pool (Golberg, 1999). The test assesses the higher order Big Five factors (Neuroticism, Extraversion, Openness, Agreeableness and Conscientiousness) on a 5-point Likert scale. The scale ranges from very inaccurate to very accurate. Internal consistencies in this sample were .80.91. Unlike the NEO, this test is brief, freely available and publicly accessible. Aggression Questionnaire. The Aggression Questionnaire (Buss & Perry, 1992) contains 29 items assessing self-reported physical aggression, verbal aggression, anger and hostility. A 5-point Likert scale is used, ranging from extremely uncharacteristic of me to extremely characteristic of me. Internal consistencies in this sample were .77.85. Social and Emotional Loneliness Scale for AdultsShort form. This questionnaire includes 15 items and three subscales assessing self-reported emotional loneliness (family and romantic) and social loneliness, rated using a 7-point Likert scale ranging from strongly disagree to strongly agree (DiTommaso, Brannen, & Best, 2004). Internal consistencies in this sample were .89.95. Eating Disorders Diagnostic Scale. The Eating Disorders Diagnostic Scale (Stice, Telch, & Rizvi, 2000) is a 22-item self-report questionnaire. The composite score is based on items 118 and item 21, but the latter item (addressing amenorrhea) was omitted from the scoring procedure in this study to produce a score suitable for both male and female subjects (E. Stice, personal communication, 14 July 2005). Response formats vary, for example, a 6-point Likert scale ranging from not at all to extremely, or a dichotomous yes/no format. Internal consistency in this sample was .89. Self-Administered Alcoholism Screening Test. This questionnaire includes 35 dichotomous yes/no items that screen for alcohol-related problems (Hurt, Morse, & Swenson, 1980). Internal consistency in this sample was .76. Subjective Happiness Scale. The Subjective Happiness Scale (Lyubomirsky & Lepper, 1999) is a fouritem measure of subjective happiness rated using a 7-point Likert scale. The statements at each end of the scale differ, but one example is not very happy person to very happy person. Internal consistency in this sample was .89. Satisfaction With Life Scale. This is a well-established questionnaire that includes ve items that assess life satisfaction (Diener, Emmons, Larsen, & Grifn, 1985). A 7-point Likert scale is used, ranging

Results Data screening and preliminary analyses Missing values were estimated using person mean substitution, as recommended in Hawthorne and Elliot (2005). Prior to analysis three cases identied as multivariate outliers were deleted, producing a revised N 307 (Tabachnick & Fidell, 2001). For all regression models multicollinearity among predictors did not appear to be present (r 5 .90; variance ination factors 5 4.36). For descriptive purposes we examined convergence among the three trait EI measures (restricted to global scores to keep the number of correlations to a minimum). There was high convergence among the measures: SEIS and MEIA (r .84), SEIS and TEIQue (r .73), MEIA and TEIQue (r .77; all ps 5 .001). Main analyses Concurrent validity. In interpreting our results, a more stringent alpha of .01 was used to control Type I error. Table I shows the standard multiple regression results at three hierarchical levels. (The correlation matrix between all components of the three trait EI measures [lower, middle and higher level] and the 11 criterion variables is available upon request from the rst author. At the component level [lower and middle level], for all criteria the strongest correlation was always with a component of the TEIQue.) Given that the purpose of these analyses was to compare concurrent criterion validity rather than determine which subscales explain the criteria, we compared adjusted multiple R2 (R2 ) to adjust for the number adj of predictors across all regression models. (The table of beta weights for Step 3 of all analyses in Table I is available upon request from the rst author.) Notably, unadjusted R2 were highly similar or identical. In these and the incremental validity analyses the average adjusted squared multiple correlation

Table I. Adjusted squared multiple correlation coefcients for predicting psychological functioning from EI measures

Inventory

Physical aggression Anger Hostility

Verbal aggression

Social loneliness

Family loneliness

Romantic loneliness

Eating disorder

Alcohol abusea

Happiness

Life Satisfactionb

Mean R2 adj .27** .26** .50** (.46**) .22** .20** .42** .11 .11 .21 (.19)

Mean signicant R2 adj .15 .15 .23 (.21)

Higher (global) level constructs SEIS MEIA TEIQue .04** .04** .07** .00 .01 .00 .02* .04** .15** .18** .21** .40** .28** .25** .35** .16** .14** .21** .05** .08** .11** 7.00 .00 .05** .01 .00 .05**

1 1 1

Middle level constructs MEIA TEIQue .10** .20** .06** .26** .21** .43** .29** .46** .25** .40** .18** .29** .11** .13**

3 4

.08** .11**

.07** .09**

.37** .64** (.55**) .06** .07** .28** .06** .11** .16** .42** .46** .74** (.63**)

.30** .52**

.18 .32 (.28) .32** .35** .57** .17 .25 .40 (.36)

.19

Lower level constructs SEIS MEIA TEIQue .06** .16** .27** .02 .22** .42** .10** .35** .47** .25** .36** .46** .34** .31** .45** .19** .21** .41**

4 10 15

.06** .12** .19**

Validity of trait emotional intelligence measures 9

Notes: EI emotional intelligence; MEIA Multidimensional Emotional Intelligence Assessment; SEIS Schutte Emotional Intelligence Scale; TEIQue Trait Emotional Intelligence Questionnaire; k no. subscales. (xx) Figures for models with TEIQue happiness items excluded due to high theoretical overlap with the criterion. a N 289 due to pairwise deletion of data for respondents who had made no attempt to complete the alcohol abuse questionnaire. b N 306 due to pairwise deletion of data for one respondent who had made no attempt to complete the life satisfaction questionnaire. *p 5 .01, **p 5 .001.

10

K. J. Gardner & P. Qualter


Notes: EI emotional intelligence; MEIA Multidimensional Emotional Intelligence Assessment; SEIS Schutte Emotional Intelligence Scale; TEIQue Trait Emotional Intelligence Questionnaire. For each model, the variables on Steps 1 and 2 remain the same despite the variables on Step 3 changing (i.e., three different sets of EI variables), thus, results for Steps 1 and 2 are presented only once. (xx) Figures for models with TEIQue happiness items excluded due to high theoretical overlap with the criterion. a N 289 due to pairwise deletion of data for respondents who had made no attempt to complete the alcohol abuse questionnaire. b N 306 due to pairwise deletion of data for one respondent who had made no attempt to complete the life satisfaction questionnaire. *p 5 .01, **p 5 .001. Mean signicant DR2 adj

Adjusted squared multiple correlation coefcients for psychological functioning with gender, age (Step 1), the Big Five (Step 2) and global trait EI (Step 3)

(R2 ) validity coefcient across criteria is also shown, adj as an evaluative index that combines both bandwidth and delity (Grucza & Goldberg, 2007). As shown in Table I, there was some variation in validity coefcients at the three hierarchical levels, with larger numbers of predictors producing higher adjusted multiple R2. At each hierarchical level and on average the TEIQue was the strongest predictor of all 11 criteria (one exception includes the higher level analyses in which no EI measure predicted signicant variance in verbal aggression), and signicantly predicted the most criteria as a higher level construct. This was still the case when the TEIQue happiness items were removed when predicting the happiness criterion (TEIQue optimism items were not removed due to a lack of direct conceptual overlap, although these items also inate validity estimates). Both the MEIA and TEIQue signicantly predicted all 11 criteria when assessed at the middle or lower level. The SEIS and MEIA were comparable when assessed as higher level constructs, in terms of both predictive strength and number of criteria signicantly predicted. The MEIA, however, was a slightly stronger predictor of some criteria (particularly all components of aggression and romantic loneliness) when lower level subscales were used, and was a better predictor on average. Incremental validity. First, incremental validity beyond age, gender (Step 1) (Cohen & Cohen, 1983) and the Big Five (Step 2) was explored. The use of global scores ensured consistency with the Big Five in terms of the specicity at which each are assessed. As shown in Table II, the SEIS and MEIA signicantly predicted modest degrees of unique variance in the same six criteria: hostility, social, family and romantic loneliness, happiness and life satisfaction. Which measure was a stronger predictor, however, was a function of the criterion predicted. The TEIQue performed slightly better still, predicting larger amounts of signicant variance in these six criteria, and alcohol abuse, although the adjusted R2 change (DR2 ) for the latter was small. adj Second, we tested the relative strengths of the three measures by examining the ability of each to increment the other two (Table III). As shown, global SEIS predicted small amounts of signicant variance in social loneliness and eating disorder. The MEIA only predicted signicant variance in alcohol abuse, but again the amount was small. In contrast, the TEIQue signicantly predicted small to moderate amounts of variance in all criteria except verbal aggression. Discussion This study examined concurrent and incremental validity of three trait EI measures, taking into account

Mean DR2 adj

Life Satisfactionb

.02 .26** .09** 23.11** .07** 21.30** .17** .02 .06** .01 4.22** 7.00 3.82** .02* .04** .06** 7.00 5.15** 7.00 5.16** .01 .02 .01 .03* 3.10* .07** 4.78** .11** .02 .20** .09** 17.49** .06** 15.53** .17** .01 .37** .03** 26.93** .03** 26.74** .08** .00 .36** 7.00 22.58** 7.00 25.58** .00 .02 .17** 7.00 10.11** 7.00 10.12** .00 .06** .08** .01 7.81** .00 7.39** .01 Step 1 R2 adj Step 2 DR2 adj Step 3 SEIS: DR2 adj Overall F Step 3 MEIA: DR2 adj Overall F Step 3 TEIQue DR2 adj .02 .09** .07** 9.17** .05** 8.06** .14**

.07 (.06) Overall F 7.53** 10.29** 22.94** 32.24** 25.50** 13.36** 6.85** 5.95** 4.81** 32.16**

.04

Happiness

Variables

Table II.

Physical aggression

Verbal aggression

Anger

Hostility

Social loneliness

Family loneliness

Romantic loneliness

Eating disorder

Alcohol abusea

.01 .48** .06** 48.72** .04** 44.18** .09** (.07**) 55.24** (50.10**)

.03

.11 (.09)

.06

.05

Table III. Adjusted squared multiple correlation coefcients predicting psychological functioning with two trait EI measures on Step 1 and one on Step 2

Physical aggression Anger .17** .01 .18** .00 .04* .14** 23.09** 69.73** 59.95** 29.77** .19** .08** .06** .04** 13.96** .22** .29** .16** .08** 7.00 .00 .00 7.00 7.00 .00 .10** 12.26** .40** .37** .22** .11** .10** .00 .02* .01 .00 .02* .00 .05** .03* .02 .07** 9.66** .40** .35** .21** .11** .08** .08** Hostility .01 7.00 7.00 .01 .01 7.00 1.91

Verbal aggression

Social loneliness

Family loneliness

Romantic loneliness

Eating disorder

Alcohol abusea

Happiness

Life Satisfactionb .42**s .00 .41** .01 .23** .19** 74.46**

Mean DR2 adj

Mean signicant DR2 adj

Step 1: MEIA, TEIQue: R2 adj

.07**

Step 2: SEIS: DR2 adj

7.00

.01 (.01)

.02 (.02)

Step 1: SEIS, TEIQue: R2 adj

.07**

Step 2: MEIA: DR2 adj

7.00

.00 (.00)

.03 (.03)

adj Step 1: SEIS, MEIA R2

.04**

Step 2: TEIQue: DR2 adj

.02*

.10 (.08)

.11 (.09)

Overall Fc

7.73**

.51** (.46**) .00 (.01) .50** (.46**) .00 (.00) .29** (.29**) .22** (.18**) 106.07** (88.11**)

Validity of trait emotional intelligence measures 11

Notes: EI emotional intelligence; MEIA Multidimensional Emotional Intelligence Assessment; SEIS Schutte Emotional Intelligence Scale; TEIQue Trait Emotional Intelligence Questionnaire. (xx) Figures for models with TEIQue happiness items excluded due to high theoretical overlap with the criterion. a N 289 due to pairwise deletion of data for respondents who had made no attempt to complete the alcohol abuse questionnaire. b N 306 due to pairwise deletion of data for one respondent who had made no attempt to complete the life satisfaction questionnaire. c For each criterion variables the overall F is equivalent across the three sets of predictors. *p 5 .01, **p 5 .001.

12

K. J. Gardner & P. Qualter measure was redundant and unable to increase predictions (this is not to say that the measures have little explanatory value). It is worth noting too, that these were stringent tests of incremental validity given the use of (a) multiple control variables, and (b) a global trait EI score. In the latter case trait EI carries only one degree of freedom and is being pitted against ve personality factors that carry ve degrees of freedom and are measured at similar bandwidths. Differential incremental validity was also documented via tests of the capacity of one EI measure to increment the other two, where again the TEIQue performed best. There was also some suggestion that the SEIS and MEIA (although more so the former) assess something unique in specic criteria and enhance predictions beyond the other EI measures. These ndings show important information regarding what psychological criteria each measure is incrementally associated with. Future research should assess incremental validity of these EI measures at lower levels of assessment and beyond lower level personality facets. In addition, the prediction of non-self-report-based criterion measures will avoid the problems of common method variance, which may inate validity coefcients. Related to this point, although self-report may be an appropriate assessment method for some criteria (e.g., happiness and life satisfaction), for others such as aggression it is only a substitute for measuring the real construct. Conclusion For researchers interested in using trait EI inventories for prediction, these data suggest that the TEIQue should be used, at least out of the three trait EI measures given and in the context of the present criterion variables. For researchers wishing to assess a limited and specic set of components, the MEIA and SEIS (both based on the Salovey & Mayer [1990] model) appear to be equally valid measures of global trait EI, but the former is more useful when using lower level facets to predict criteria. References
Ashton, M. C., Jackson, D. N., Paunonen, S. V., Helmes, H., & Rothstein, M. G. (1995). The criterion validity of broad factor scales versus specic facet scales. Journal of Research in Personality, 29, 432442. Austin, E. J., Saklofske, D. H., & Mastoras, S. M. (2010). Emotional intelligence, coping, and exam-related stress in Canadian undergraduate students. Australian Journal of Psychology, 62, 4250. Bastian, V. A., Burns, N. R., & Nettelbeck, T. (2005). Emotional intelligence predicts life skills, but not as well as personality and cognitive abilities. Personality and Individual Differences, 39, 11351145.

bandwidth and delity to conduct a fairer comparison across measures. In support of concurrent validity, all three measures were related to a range of criteria, regardless of whether they were assessed as lower, middle or higher level constructs. Validity coefcients were maximised for lower levels of assessment, results that are consistent with past literature in the personality eld (e.g., Ashton et al., 1995; Paunonen & Ashton, 2001) and the bandwidthdelity trade-off. Of all measures, the TEIQue was a superior predictor at each level, either signicantly predicting a broader range of criteria than the MEIA and SEIS (mainly at the higher level of assessment), or being a consistently stronger predictor (as evidenced by higher average validity coefcients). Statistically speaking, for the lower and middle levels of assessment this is not surprising because measures with more subscales have a predictive edge, although there is no guarantee that more variables will add substantially to prediction, as in this case (Paunonen & Ashton, 2001). Moreover, even at the higher levels of assessment (global scores), where each EI measure carried only one degree of freedom and the TEIQue was unable to capitalise on chance associations, it was still stronger. This may result from the broader theoretical domain of the TEIQue. It is interesting that average validity coefcients were comparable for the SEIS and MEIA when assessed as higher level constructs with only slight differences across individual criteria. Thus, there may be little advantage in using one of these measures over the other if the aim is to assess global trait EI. When assessing lower level facets, however, EI may be better assessed by the MEIA for some criteria given stronger predictive power. This may arise, however, because the MEIA includes more subscales than the SEIS, which carries only four degrees of freedom. It is important for future research to determine the range of criteria in which the MEIA is a superior predictor, and criteria for which the shorter and freely available SEIS is an equally valid and alternative indicator of trait EI (brief scales save time in testing and also avoid participant boredom and fatigue). For criteria for which predictive power of both measures is comparable, under the law of parsimony it is preferable to use fewer factors (the SEIS) if they are equally good indicators of trait EI (Gorsuch, 1983). Regarding incremental validity, the three measures were incrementally associated with approximately half of the criteria. The MEIA and SEIS produced fairly similar-sized validity coefcients and the TEIQue demonstrated superior incremental power. These positive ndings regarding the TEIQues incremental validity are in accordance with past studies (e.g., Petrides, Perez-Gonzalez, et al., 2007). It is also clear, however, that in some cases each

Validity of trait emotional intelligence measures


Brackett, M. A., & Mayer, J. D. (2003). Convergent, discriminant, and incremental validity of competing measures of emotional intelligence. Personality and Social Psychology Bulletin, 29, 1147 1158. Buss, A. H. & Perry, M. (1992). The Aggression Questionnaire. Journal of Personality and Social Psychology, 63, 452459. Chapman, B. P. & Hayslip B. Jr. (2005). Incremental validity of a measure of emotional intelligence. Journal of Personality Assessment, 85, 154169. Cohen, J., & Cohen, P. (1983). Applied multiple regression/correlation for the behavioral sciences (2nd ed.). Hillsdale, NJ: Lawrence Erlbaum Associates. Diener, E., Emmons, R. A., Larsen, R. J., & Grifn, S. (1985). The Satisfaction With Life Scale. Journal of Personality Assessment, 49, 7175. DiTommaso, E., Brannen, C., & Best, L. A. (2004). Measurement and validity characteristics of the short version of the Social and Emotional Loneliness Scale for Adults. Educational and Psychological Measurement, 64, 99119. Downey, L. A., Johnston, P. J., Hansen, K., Birney, J., & Stough, C. (2010). Investigating the mediating effects of emotional intelligence and coping on problem behaviours in adolescents. Australian Journal of Psychology, 62, 2029. Garb, H. N. (1984). The incremental validity of information used in personality assessment. Clinical Psychology Review, 4, 641 655. Gardner, K., & Qualter, P. (2009). Emotional intelligence and borderline personality disorder. Personality and Individual Differences, 47, 9498. Golberg L. R. (1999). A broad-bandwidth, public-domain, personality inventory measuring the lower-level facets of several Five-Factor models. In I. Mervielde, I. J. Deary, F. De Fruyt, & F. Ostendorf (Eds.), Personality psychology in Europe (pp. 7 28). Tilburg, The Netherlands: Tilburg University Press. Gorsuch, R. L. (1983). Factor analysis (2nd ed.). Hillsdale: NJ: Erlbaum. Grucza, R. A., & Goldberg, L. R. (2007). The comparative validity of 11 modern personality inventories: Predictions of behavioral acts, informant reports, and clinical indicators. Journal of Personality Assessment, 89, 167187. Hawthorne, G., & Elliott, P. (2005). Imputing cross-sectional missing data: Comparison of common techniques. Australian and New Zealand Journal of Psychiatry, 39, 583590. Hogan, M. J., Parker, J. D. A., Wiener, J., Watters, C., Wood, L. M., & Oke, A. (2010). Academic success in adolescence: Relationships among verbal IQ, social support and emotional intelligence. Australian Journal of Psychology, 62, 3041. Hurt, R. D., Morse, R. M., & Swenson, W. M. (1980). Diagnosis of alcoholism with a Self-Administered Alcoholism Screening Test: Results with 1,002 consecutive patients receiving general examinations. Mayo Clinic Proceedings, 55, 365370. Lyubomirsky, S., & Lepper, H. S. (1999). A measure of subjective happiness: Preliminary reliability and construct validation. Social Indicators Research, 46, 137155. Mikolajczak, M., Luminet, O., Leroy, C., & Roy, C. (2007). Psychometric properties of the Trait Emotional Intelligence Questionnaire (TEIQue; Petrides & Furnham, 2003): Factor structure, reliability, construct, and incremental validity in a French-speaking population. Journal of Personality Assessment, 88, 338353.

13

Ones, D. S., & Viswesvaran, C. (1996). Bandwidth-delity dilemma in personality measurement for personnel selection. Journal of Organizational Behavior, 17, 609626. Paunonen, S. V., & Ashton, M. C. (2001). Big ve factors and faces and the prediction of behavior. Journal of Personality and Social Psychology, 81, 524539. Petrides, K. V. (2009). Technical manual for the Trait Emotional Intelligence Questionnaires (TEIQue). London: London Psychometric Laboratory. Petrides, K. V., & Furnham, A. (2001). Trait emotional intelligence: Psychometric investigation with reference to established trait taxonomies. European Journal of Personality, 15, 425448. Petrides, K. V., & Furnham, A. (2003). Trait emotional intelligence: Behavioural validation in two studies of emotion recognition and reactivity to mood induction. European Journal of Personality, 17, 3957. Petrides, K., Perez-Gonzalez, J. C., & Furnham, A. (2007). On the criterion and incremental validity of trait emotional intelligence. Cognition and Emotion, 21, 2655. Petrides, K. V., Pita, R., & Kokkinaki, F. (2007). The location of trait emotional intelligence in personality factor space. British Journal of Psychology, 98, 273289. Qualter, P., Whiteley, H. E., & Gardner, K. (2003). How to make emotional intelligence work for you: A workbook. Unpublished manuscript. Saklofske, D. H., Austin, E. J., & Minski, P. S. (2003). Factor structure and validity of a trait emotional intelligence measure. Personality and Individual Differences, 34, 707721. Salovey, P., & Mayer, J. (1990). Emotional intelligence. Imagination, Cognition and Personality, 9, 185211. Sanchez-Ruiz, M. J., Perez-Gonzalez, J. C., & Petrides, K. V. (2010). Trait emotional intelligence proles of students from different university faculties. Australian Journal of Psychology, 62, 5157. Schutte, N. S., Malouff, J. M., Hall, L. E., Haggerty, D. J., Cooper, J. T., Golden, C. J., et al. (1998). Development and validation of a measure of emotional intelligence. Personality and Individual Differences, 25, 167177. Schutte, N. S., Malouff, J. M., Thorsteinsson, E. B., Bhullar, N., & Rooke, S. E. (2007). A meta-analytic investigation of the relationship between emotional intelligence and health. Personality and Individual Differences, 42, 921933. Schutte, N. S., Thorsteinsson, E. B., Hine, D. W., Foster, R., Cauchi, A., & Binns, C. (2010). Experiential and rational processing styles, emotional intelligence and wellbeing. Australian Journal of Psychology, 62, 1419. Stice, E., Telch, C. F., & Rizvi, S. L. (2000). Development and validation of the Eating Disorder Diagnostic Scale: A brief selfreport measure of anorexia, bulimia, and binge-eating disorder. Psychological Assessment, 12, 123131. Tabachnick, B. G., & Fidell, L. S. (2001). Using multivariate statistics (4th ed.). Boston: Allyn and Bacon. Tett, R. P., Fox, K. E., & Wang, A. (2005). Development and validation of a self-report measure of emotional intelligence as a multidimensional trait domain. Personality and Social Psychology Bulletin, 31, 130.

Copyright of Australian Journal of Psychology is the property of Taylor & Francis Ltd and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.

Vous aimerez peut-être aussi