ruYxo^ontA, 2008, 15 (2) a 200-218 PSYCHOLOGY, 2008, 15 (2) ) 200-218

Measuring trait emotional intelligence:development and psychometric properties

of the Greek Emotional Intelligence Scale (GEIS)

lonrururs Tsnousrsl

ABSTRACTThis art icle describes the development and val idation of a new self reportmeasure o f emot iona l in te l l igence in the Greek language based on thetheoretical framework proposed by Mayer and salovey (1g97). The Greek

Emotional Intel l igence Scale (GEIS) was developed to measure trait emotional intel l igence, and consistsof 52 i tems measuring four basic emotional ski l ls: Expression and Recognit ion of Emotions, Control ofEmot ions, Use of Emot ions for Fac i l i ta t ing Th ink ing, and Car ing and Empathy. In th is s tudy, 1387individuals part icipated in the various stages of the development and val idation of the test. A principalcomponent analysis was conducted on the data and four interpretable factors were rotated using directoblimin procedure. The Cronbach's a coefficients for the four factors ranged between 0.80 and 0.92. Thetest-retest correlat ion coeff icients ranged between 0.79 and 0.91. Results from f ive dif ferent studiessupported also the convergent and discriminant val idity of the GEIS scales, using for that goal twelvedifferent measures from the cognit ive as well as from the emotional/personali ty domain. For example,GEIS scales were found to be posit ively and signif icantly correlated with Extraversion, Agreeablenessand Conscientiousness, and negatively and signif icantly correlated with Neuroticism. GEIS scales werealso found not to be correlated with cognitive ability. We concluded that the psychometric features of theGEIS supported i ts feasibi l i ty as a research instrument to measure trait emotional intel l iqence in Greekpopulat ion

Keywords: Emotional intelligence, Test construction, Information-processing El, Trait El.

1. Introduction

The recent and widespread interest in theconstruct of Emot ional Intel l igence (El) has ledresearchers to focus on how this newlyintroduced concept has developed (Bar-On &

Parker, 2000. Ciarrochi , Forgas & Mayer, 2001.Dulewicz & Higgs, 2003. Law, Wong & Song,2004. Petr ides & Furnham, 2003). However,besides the popular i ty of the construct , there isst i l l some theoret ical confusion regarding theexact meaning and domain of the concept. This

1. Address: Department of Psychology, University of Crete, 74100, Gallos, Rethymnon. Tel.: 2831-077519, e-marr:tsaousis@psy.soc.uoc. gr

,---4

Measuring trait emotional intelligence ) 20I

confusion has resul ted in the development ofthree alternative theoretical frameworks forconceptualizing the construct.

The f i rst model has been developed bySalovey and Mayer (1990), who first introducedthe term "Emotional Intel l igence". According tothem, El is def ined as " the subset of socialintel l igence that involves the abi l i ty to monitorone's own and others ' feel ings and emot ions, todiscr iminate among them and to use thisinformat ion to guide one's th inking and act ions"(p. 189). The El construct ref lects a four- levelhierarchy ranging from basic to more complexpsychological processes. At the lowest levelstands the abi l i ty to perceive, appraise, andexpress emot ion; the next level up involves theabil ity to use emotions to facil i tate cognition. Thethird level reflects the abil ity to understand andreason about emot ions, and the fourth levelinvolves the abi l i ty to regulate emot ions tofacil i tate emotional and cognitive growth, whichreflects the most complex level of El. The aboveconceptual model is characterized by a two-component schema: at the higher level , there isa general processing of emot ional informat ion,and at the lower level there are specific skil ls thatare involved in such processing. One couldarg.ue that this perspective perceives El as amodel of intell igence. This perspective is furtherenhanced by the recent work of Mayer, Carouso,and Salovey (1999), in which they expl ic i t lydeclared that their model should be viewedwithin the context of mental abi l i ty , s ince i tsatisfies the three traditional classes of criteria forintel l igence: conceptual , correlat ional , anddevelopmental . .

The second model has been introduced byDaniel Goleman (1998a), who was responsible forthe popularization of the concept. He has definedEl as " the capaci ty for recogniz ing our ownfeel ings and those of others, for mot ivat ingourselves, and for managing emot ions wel l inourselves and in our relat ionships" (p. 317). Heformulated his model in terms of a theory ofperformance since, as he suggested, his model

has direct applicabil ity to the domain of work andorganisational effectiveness, particularly inpredict ing excel lence ( i .e. job performance) injobs of a l l k inds, f rom sales to leadership(Goleman, 1998b).

Finally, Raven Bar-On (1997) has placed El inthe context of emotional and social competencies.His definit ion of El described it as "an array ofnoncognitive capabil it ies, competencies, and skil lsthat inf luence one's abi l i ty to succeed in copingwith environmental demands and pressures" (p.14). He proposed a model of noncognitiveintelligences that includes five broad areas of skillsor competencies from the personality domain, andwithin each, more speci f ic ski l ls that appear tocontribute to success. These include (a)intrapersonal skil ls, (b) interpersonal skil ls, (c)adaptabil ity, (d) stress management, and (e)general mood.

On the basis of the above descr ibedtheoretical development, it seems that, at present,there are two approaches in studying El. On theone hand, there is the abil ity E/, proposed byMayer and Salovey (1997) and Mayer, Carusoand Salovey (1999), who argue that El constitutesan addi t ional aspect of crystal l ized intel l igence.On the other hand, there is the trait E/ which refersto " . . .a constel lat ion of emot ion-related sel f -perceptions and dispositions located at the lowerlevels of personality hierarchies" (Petrides, Perez-Gonzalez & Furnham ,2007). According to Carroll(1993), the conceptualisation of El as a personalitytrait leads to a construct that lies wholly outside thetaxonomy of human cogni t ive abi l i ty . Goleman(19BBa), although he init ially supported the latterapproach, has recently tried to represent El withinthe competence domain.

Extensive research work has been produceddur ing the last few years on this debate, in anattempt to clarify and crystall ize which of the twomodels best explains El . Unfortunately, theoutcome of this attempt, instead of empoweringone or the other approach, has led to some degreeof theoret ical confusion since the resul ts werecontradictory. According to Petrides and Furnham

202 a loonnis Tsaousts

(2003), "these two are different constructs becausethe procedures used in their operational definit ionsare fundamentally different, even though theirtheoretical domains might overlap" (p, 40)

Almost s imultaneously wi th the developmentof theoretical models of El there was an inevitableinterest in the development of tests to measure theconcept. According to Mayer, Caruso, and Salovey(1999), the measurement of El plays an importantrole in the conceptualization of the concept, sincei f i t cannot adequately be measured then onecould argue that it might not exist as a meaningfulscientif ic construct. Existing measures of El maybe divided into two categories: (a) performancemeasures, and (b) self-report measures. The firstcategory operationalizes the information-processing El model whi le the second categoryooerationalizes the trait El model.

The most comprehensive performancemeasure of El is the Mayer-Salovey-CarusoEmotional Intel l igence Test (MSCEIT, V.2.0)developed by Mayer, Salovey, and Caruso(2002), which appears to be an improvement overi ts predecessor, the Mult i factor EmotionalIn te l l igence Sca le (MElS. Mayer e t a l . , 1999) ,which in turn, was designed to measure fourmajor hypothet ical components ( i .e. branches)that under l ie the hierarchical model theyproposed. Al though the MSCEIT demonstratesacceptable psychometr ic propert ies (Mayer,Caruso & Salovey, 1999. Mayer et a l . , 2003),some researchers express concerns regardingthe scor ing techn ique used ( i .e . consensusscoring) and its effectiveness to providemeaningful scores especial ly at the high end ofthe El cont inuum (Conte, 2005. Matthews,Zeidner & Roberts, 2002). Other abi l i ty-basedscales are the Emotional Accuracy ResearchScale (EARS. Geher, Warner & Brown, 2001.Mayer & Geher, 1996), and the Levels ofEmotional Awareness Scale (LEAS Lane et a l . ,1990). Both measures, a l though they belong toperformance El measures, have not gained muchscient i f ic at tent ion dur ing the last years(especially after the emergence of the MEIS and

MSCEIT) and are rarely used as typical Elmeasures.

The second major category involves anumber of researchers who have at tempted todevelop sel f - report measures of El , One of theearliest attempts was put forth by Schutte,Malouff , Hal l , Haggerty, Cooper, Golden, andDornheim (1998) who have developed a sel f -report measure of El (The Schutte Self-ReportInventory-SSRI), based on Salovey and Mayer 's(1990) in i t ia l El model (expression and appraisal ,regulat ion, and ut i l izat ion of emot ion). Al thoughthe SSRI demonstrates acceptable rel iabi l i tyindices, i ts main l imi tat ion is related to i tsdimensionality; according to Schutte et al. (1998),the test was designed to measure a general Elfactor, however, later studies (Petrides &Furnham, 2000. Saklofske, Austin & Minski, 2003)using more powerful statistical techniques failedto repl icate that general El factor. Instead, theysuggested a mult i -d imensional model containingthree dist inct components (opt imism/moodregulat ion, appraisal of emot ions, social ski l ls ,and uti l ization of emotions), which does not maponto Salovey and Mayer's model of El. Moreover,Aust in, Saklofske, Huang, and McKenney (200a)added eight i tems into the or ig inal 33- i tem scaleand found that a modif ied 41- i tem version hadbetter psychometric characteristics.

Bar-On (1997), based on his theoret icalframework of "noncognitive" factors, has developedthe Emotional Quotient Inventory (EO-i) to assessEl, This 133-item self-report measure consists of15 dist inct scales, including (a) intrapersonal El(emotional self-awareness, assertiveness, self-regard, self-actualization, and independence), (b)interpersonal El (empathy, relationship skil ls, andsocial responsibi l i ty) , (c) adaptabi l i ty (problemsolving, real i ty test ing, and f lexibi l i ty) , and (d)stress management (stress to lerance andimpulse control) . EQ-i has shown good internalconsistency and test-retest re l iabi l i ty indices aswel l as evidence of convergent and divergentvalidity, mainly from the personality domain (Bar-On, 1996. Bar-On, Brown, Kirkcaldy & Thome,

Measuring trait emotional intelligence ) 203

2000. Dawda & Hart , 2000. Newsome, Day &Catano, 2000). The major crit ique regarding this Elmeasure comes from Mathews et al, (2002) whoargue that the theory behind this measure is rathervague, and that further research is needed toprove that EQ-i's sub-scales are related to El.Boyatzis, Goleman and Rhee (2000), workingwithin the competencies domain, have developedthe Emotional Competence Inventory (ECl), a 110-i tem instrument which has been designed tomeasure 20 competencies organized into fourclusters: Self-Awareness, Self-Management,Social Awareness, and Social Ski l l . The internalconsistency reliabil i ty of the ECI ranges from 0.61t.o 0.85, while discriminant and predictive validityevidence comes only f rom the Sel f -AssessmentQuesiionnaire (SAQ) which is the predecessor ofthe ECl. This lack of val id i ty evidence promptedConte (2005) to argue that, unti l peer-reviewedempir ical studies using this measure areconducted, ECI does not deserve ser iousconsideration.

Dulewicz and Higgs (2000) have alsodeveloped an instrument aiming to assess theconcept of El , The Emotional Intel l igenceQuest ionnaire (ElO) has been designed tospecifically assess through seltreport sevenelements of an individual's emotional intell igence:self-awareness, emotional resi I ience, motivation,inter-personal sensitivity, influence, intuit iveness,and conscientiousness. The authors presentevidence which supports the rel iabi l i ty and theval id i ty of their instrument (Dulewicz, Higgs &Slaski, 2003) and claim that EIQ is a parsimoniousmeasure of El, suitable for use within the workingand organizational framework. Additionally,Petrides and his associates (2005, 2006, 2007)have developed the Trait Emotional Intell igenceQuestionnaire (TElQue), a 153-item instrument oft ra i t El , which contains four components (wel l -being, self-control, emotionality, and sociabil ity)and fifteen subscales (including optimism, emotionregulat ion, emot ion management, socialcompetence, adaptabil ity, etc.). TElQue hasdemonstrated adequate reliability and validity data

(Mikolajczak et a l . , in press. Perez, Petr ides &Furnham, 2005. Petr ides & Furnham, 2006.Petrides, P6rez-GonzAlez & Furnham, 2007) andhas already been translated in many differentlanguages ( i .e. Spanish, Greek, Pol ish,Portuguese, l ta l ian, French, Dutch, Chinese,Norwegian, Croat ian, Malay and German). Themain l imitation of both measures (ElQ and TElQue)is that most of their scales either overlap withpersonal i ty dimensions (e.9. conscient iousness,intuit iveness, sociabil ity) and other psychologicalconcepts (e.9. motivation, well-being) or might bebest described as meta-cognitive constructs (e.9.seltawareness, self-control). Consequently, itseems olausible these instruments are notassessing a separate trait of emotional intell igencebut rather a particular combination of existingperson-specific characteristics that mostly appearunrelated to emotion.

Finally, there are two other self-reportmeasures, namely the Wong and Law's Emotionallntell igence Scale (WLEIS. Wang & Law, 2002),and the Mult id imensional Emot ional Intel l igenceAssessment (MEIA. Tett, Fox & Wang, 2005)which are based on the or ig inal Salovey andMayer (1990, 1997) El model. Both instrumentspresent evidence to support the theoret icalframework introduced by Salovey and Mayer(1990), but th is t ime from the trai t perspect ive,enhancing the argument made by Schutte et a l .(1998) that th is model "seems to be an excel lentprocess-oriented model that emphasizes stages ofdevelopment in emot ional intel l igence, potent ia lfor growth and the contributions emotions make tointellectual growth" (p. 169).

The preceding review of El measuresindicates that although there are many measures-either performance or self-reports- that claim tomeasure the El concept, there is st i l l a need forvalidated measures that are based on acomprehensive and parsimonious model. AsSchutte et a l . (1998) suggest, th is can be themodel proposed by Mayer and Salovey (1997),s ince i t "seems to be an excel lent orocess-or iented model that emphasizes stages of

, - - - 4

202 4 loannis Tsaousis

(2003), "these two are different constructs becausethe procedures used in their operational definit ionsare fundamentally different, even though theirtheoretical domains might overlap" (p. 40).

Almost simultaneously with the developmentof theoretical models of El there was an inevitableinterest in the development of tests to measure theconcept. According to Mayer, Caruso, and Salovey(1999), the measurement of El plays an importantrole in the conceptualization of the concept, sincei f i t cannot adequately be measured then onecould argue that it might not exist as a meaningfulscientif ic construct. Existing measures of El maybe divided into two categories: (a) performancemeasures, and (b) self-report measures. The firstcategory operationalizes the information-processing El model whi le the second categoryooerationalizes the trait El model.

The most comprehensive performancemeasure of El is the Mayer-Salovey-CarusoEmotional Intel l igence Test (MSCEIT, V.2.0)developed by Mayer, Salovey, and Caruso(2002), which appears to be an improvement overi ts predecessor, the Mult i factor EmotionalIn te l l igence Sca le (MElS. Mayer e t a l . , 1999) ,which in turn, was designed to measure fourmajor hypothet ical components ( i .e. branches)that under l ie the hierarchical model theyproposed. Al though the MSCEIT demonstratesacceptable psychometr ic propert ies (Mayer,Caruso & Salovey, 1999. Mayer et a l . , 2003),some researchers express concerns regardingthe scor ing techn ique used ( i .e . consensusscoring) and its effectiveness to providemeaningful scores especial ly at the high end ofthe El cont inuum (Conte, 2005. Matthews,Zeidner & Roberts, 2002). Other abi l i ty-basedscales are the Emotional Accuracy ResearchSca le (EARS. Geher , Warner & Brown, 2001.Mayer & Geher, 1996), and the Levels ofEmotional Awareness Scale (LEAS. Lane et a l . ,1990). Both measures, a l though they belong toperformance El measures, have not gained muchscient i f ic at tent ion dur ing the last years(especially after the emergence of the MEIS and

MSCEIT) and are rarely used as typical Elmeasures.

The second major category involves anumber of researchers who have attemoted todevelop sel f - report measures of El . One of theearliest attempts was put forth by Schutte,Ma lou f f , Ha l l , Hagger ty , Cooper , Go lden, andDornheim (1998) who have developed a sel f -report measure of El (The Schutte Sel f -ReportInventory-SSRI), based on Salovey and Mayer's(1990) in i t ia l El model (expression and appraisal ,regulat ion, and ut i l izat ion of emot ion). Al thoughthe SSRI demonstrates acceptable rel iabi l i tyindices, i ts main l imi tat ion is related to i tsdimensionality; according to Schutte et al. (1998),the test was designed to measure a general Elfactor, however, later studies (Petrides &Furnham, 2000. Saklofske, Austin & Minski, 2003)using more powerful statistical techniques failedto repl icate that general El factor. lnstead, theysuggested a mult i -d imensional model containingthree dist inct components (opt imism/moodregulat ion, appraisal of emot ions, social ski l ls ,and uti l ization of emotions), which does not maponto Salovey and Mayer's model of El. Moreover,Aust in, Saklofske, Huang, and McKenney (2004)added eight i tems into the or ig inal 33- i tem scaleand found that a modif ied 41- i tem version hadbetter psychometric characteristics.

Bar-On (1997), based on his theoret icalframework of "noncognitive" factors, has developedthe Emotional Quotient Inventory (EO-i) to assessEl, This 133-item self-report measure consists of15 dist inct scales, including (a) intrapersonal El(emotional self-awareness, assertiveness, self-regard, self-actualization, and independence), (b)interpersonal El (empathy, relationship skil ls, andsocial responsibi l i ty) , (c) adaptabi l i ty (problemso lv ing , rea l i t y tes t ing , and f lex ib i l i t y ) , and (d )stress management (stress to lerance andimpulse control) . EQ-i has shown good internalconsistency and test-retest re l iabi l i ty indices aswel l as evidence of convergent and divergentvalidity, mainly from the personality domain (Bar-On, 1996. Bar-On, Brown, Kirkcaldy & Thome,

204 4 loannis Tsaousis

development in emotional intell igence, potentialfor growth and the contributions emotions maketo intellectual growth" (p. 169). Furthermore, onecould argue that most dimensions of allthe othermodels can, more or less, be integrated into thismodel.

Furthermore, the lack of a psychometricinstrument measur ing the construct of El in theGreek language necessitated the construction ofa rel iable and val id measure in Greek. Betweenthe two al ternat ive methodologies; namely,adopt ing an imported measure by t ranslat ing i tinto the Greek language ("et ic" approach) ordeveloping a new one taking into account thespecific ethnic and cultural characteristics of theGreek populat ion ("emic" approach), the lat terwas deemed more appropr iate, Developing analtogether new measure provides severalconceptual as well as methodological advantages(Benet-Martinez, 2006),

Under th is perspect ive, the purpose of th isstudy was to (a) develop a reliable and valid self-report measure of the construct of emotionalintell igence, and (b) to provide validity evidencewhich justify that the newly developed instrumentis a measure of the trait emotional intell igence, atheoretical perspective which assumes that El isa disposi t ional tendency, c losely related topersonality domain.

2. Method

Participants

Four different samples were used in this study.Sample 1 was used to test the factor structure ofthe initial version of the GEIS, and consisted of 246individuals of whom 94 (38,2/"\ were males. Themean age of the total sample was 31.76(SD= 10.42) years of age. All participants wereemployees from private companies whoparticipated in emotional intell igence seminars.Sample 2 was used to replicate the factor structureas well as the convergent and discriminant validityof the f inal version of the GEIS. This sample

consisted of 511 individuals of whom 156 (30.5%)were males and 346 (67.7%) were females (nineindividuals did not report their gender). The meanage of the participants was 30.53 years (SD=9.97),and they were also employees from privatecompanies who participated in emotionalintel l igence seminars. Sample 3 was used toinvestigate the latent structure as well as theconvergent and discriminant validity of the finalversion of the GEIS, and consisted of 699individuals. Of these, 251 (35.9'/") were males (sixparticipants did not report their gender). The meanage of the sample was 30.47 (SD= 11.85). Final ly,Sample 4 was used to examine the test-retestreliabil i ty coefficients of the GEIS scales, andconsisted of 83 individuals (62 females; f iveindiv iduals did not report their gender) , beingadul ts of mean age 27.17 years (SD=8,06), whocompleted the GEIS twice, with an interval of fourweeks between administrations. All participants inSample 4 were students.

Measures

Twelve different measures from the cognitiveas well as from the emotional-personality domainwere used in order to test the convergent anddiscr iminant val id i tv of the GEIS scales.Particularly:

Cognitive Ability MeasuresStandard Progressive Matrices (Raven, Coutt &

Raven, 1979). This is a 60-item test measuring fluidintell igence (Cattell, 1971). lt consists of f ive setsof 12 matrices and deals with sequences of relatedpatterns. The Raven Progressive Matrices test hasvery high internal consistency reliabil i ty (>0.90).

The AH4 (Heim, Watts & Simmonds, 1970).Th is grou p-ad ministered intell i gence test consistsof two sets of 65-items, yielding three scores:verbal intell igence, perceptual intell igence, and atotal intell igence score. According to Kline (1993),AH4's verbal and perceptual scores are typicalmeasures of crystall ized intell igence. All reliabil i t iesreported for the AH4 are higher than 0,80.

Measuring trait emotional intelligence ) 205

Personality MeasuresThe Traits Personality Questionnaire - TPQUe

(Tsaousis, 2002). This is a comprehensivemeasure of the f ive major dimensions ofpersonality (Extraversion, Neuroticism, Opennessto Exper ience, Agreeableness, and Conscien-tiousness) as well as of the most important traitsthat define each domain in the Greek language. ltconsists of 206 items and is based on Costa andMcCrae's (1992) def in i t ions of the mostacceptable factors in the five-factor theory. TPQuehas indicated acceptable rel iabi l i ty and val id i tydata (Tsaousis, 2002).

Locus of Control (Rotter, 1966). We usedRotter's l-E Scale to tap this variable in our study.Rotter 's or ig inal scale consisted of 23 forced-choice LOC i tems and six f i l ler i tems to obscurethe purpose of the test . The 23 i tems yielded aCronbach a=0.75

Emotional lntelligence Related MeasuresThe Schutfe Se/f-Report lnventory - SSR/

(Schutte et al., 1998). This self-report questionnairecomprises 33 self-referencing statements.According to the authors, a l l i tems loadsigni f icant ly on a s ingle factor (Schutte et a l . ,1998), which called Overall El. For the purposes ofthe present study, since our factor analytic resultsdid not replicate the four-factor solution suggestedby the previous researchers, we decided to useonly the Overal l El score. Alpha rel iabi l i ty for the

'

total scale was 0,85.Toronto Alexithymia Sca/e - IAS (Taylor, Ryan

& Bagby, 1985). This is a 26- i tem quest ionnaire,that measures a c l in ical syndrome known asalexi thymia (Si fneos, 1973), which is def ined asthe di f f icul ty of the indiv idual (a) to ident i fy anddescr ibe feel ings, (b) to communicate wi themotions, (c) to daydream, and (d) for externallyor iented thinking. The alpha coeff ic ient for TAS-20 total score was 0.92, whi le the alphas for thesub-scales ranged lrom 0.74 to 0.90.

The Trait Meta-Mood Sca/e - TMMS (Saloveyet al . , 1995). This instrument comprises 30 i tems.It contains three sub-scales: Attention (13 items),

Repa i r (6 i tems) , and C lar i t y (11 i tems) . l t a lsoprovides an overal l meta-mood score. Alpharel iabi l i t ies for each sub-scale were very highranging from 0.92 to 0.96. Alpha rel iabi l i ty for thetotal score was 0.97.

Social Skil/s Inventory - SS/ (Riggio, 1989).The SSI is a 9O-i tem quest ionnaire designed toassess bas ic communica t ion sk i l l s . Morespeci f ical ly, i t measures social ski l ls in s ixdomains (emot ional expressiv i ty, emot ionalsensitivity, emotional control, social expressivity,socialsensitivity, and social control) and providesa total score that reflect a global level of socialski l l development indicat ive of overal l social ski l lcompetence or social intell igence. Alpha reliabil i tyfor the total scale is 0,98. whi le alpha for the s ixscales ranged between 0,89 to 0,92,

The Emotional Empathy Scale - EES (Caruso& Mayer, 1997). This scale consists of 30 i temsmeasur ing the extent to which an indiv idual isable to feel what the other Derson feels. The EESprovides an overall score that represents the totaiempathy score. The alpha rel iabi l i ty index for th isscale is very high (alpha=0.97).

Well-being MeasuresThe Satisfaction With Life Sca/e - SVVLS

(Diener, Emmons, Larsen & Gri f f in, 1985). This isS-item questionnaire that is used to measure theparticipants' global, cognitive assessment of theirl i fe as a whole. The SWLS typically uses a 7-pointresponse format. We changed the responseformat to a S-point scale (1=strongly disagree, to5=strongly agree) because a 5-point responseformat was used for most of the ouestionnaires inthe survey. Diener et a l . (1985) have reportedevidence of d iscr iminant and convergent val id i tyfor the scale, whi le the alpha rel iabi l i ty was 0.72 inthis data set,

The Positive and Negative Affect Schedule -

PANAS (Watson, Clark & Tel legen, 1988). ThePANAS includes '10 posi t ive (happy, joyful .p leased, etc.) and 1 0 negat ive (depressed,frustrated, angry, etc.) emotion adjectives. Scoreson the ten oosi t ive emot ion i tems are summed tc

206 a loannis Tsaousis

indicate the participant's general level of PositiveAffect, while scores on the ten negative emotionitems are summed to indicate a participant'sgeneral level of Negative Affect. Alpha reliabilitiesfor both scales were high and acceptable (0.79and 0.75 for positive and negative scales,respectively).

Work Stress MeasureIhe ASSEf (Cartwright & Cooper, 2002). The

ASSET is an effective tool in diagnosingoccupational stress, combining both the sourcesand the effects of stress. ASSET conceptualizesoccupational stress as influenced by a variety ofsources, such as work relat ionships, work- l i febalance, over load, job secur i ty, etc. l t a lsoprovides scores for organizational commitment aswell as for physical health and psychological well-being, s ince these measures, according to themodel, are recognized to be affected byoccupat ional stress. Al l but one ASSET sub-scales (Work-Life Balance, which was excludedfrom the analysis) demonstrated satisfactori lyinternal consistency reliabil i t ies ranging between0.64 to 0.83.

Procedure

In th is sect ion only the procedure that wasfollowed during the validation phase is presented,since i t was the most compl icated. Due to thelarge number of measures used during this phase( i .e , th i r teen) , par t i c ipants f rom sample 3(N=699), were div ided into s ix di f ferent groups,each of which completed a l imi ted number ofmeasures. The f i rst group, apart f rom the theGEIS was asked to complete addi t ional ly aquest ionnaire booklet containing four measures(TAS, TMMS, SSI and EES). The second group,apart from the GEIS was asked to completeadditionally the two measures of cognitive abil ity(Raven and AH4) as well as the personalitymeasure (TPQue). The third group, apart from theGEIS was asked to complete additionally only theASSET work stress inventory. The fourth group,

apart from the GEIS was asked to completeaddi t ional ly the two measures of Wel l -being(SWLS and PANAS). The fifth group, apart fromthe GEIS was asked to complete additionally onlythe SSRI, and finally, the sixth group, apart fromthe GEIS was asked to complete additionally theLocus of Control questionnaire.

3. Results

Development of the ltems

This stage focused on determining the basicdimensions of the Concept Model and on writ ingappropriate items to measure them. Theconceptual model adopted in this study is the onesuggested by Mayer and Salovey (1997). Basedon this model, 250 i tems were generated, whichformed the initial item pool from which items for thefour scales were developed. From them, the best180 items were selected in order to form the initialversion of GEIS. The first step in item selection wasthe development oI a marker set of items for eachscale. Markers form a core cluster of items that isclosely related to other items of the scale, but notclosely related to items of other scales. Theadvantage of using a marker set in this init ial stageof item selection is that overlap between items fromvarious scales is contrclled. Two stages can bedist inguished in the development of each set ofsca/e markers. First, all the items relating to eachscale were collected and factor analyzed usingPrincipal Component Analysis (unrotated solution).Second, four items (two with the highest positiveand two with the highest negative loadings, tocontrol for the acquiescence effect) from thegenerated factor analytic solutions were chosen,and a marker set of items for each of the fourscales was composed. In the final step, each itemwas correlated with every scale. ltems were onlyselected if they were highly correlated with thescale under construction, and of low correlationwith the other scales. At the end of this phase, thetotal number of items composing this first versionof GEIS was 82.

Measuring trait emotional intelligence ) 207

1 0Scree Plot

49

28 34 40 46

Component Number

Figure 1Scree plot of the GEIS (N=511)

(l)f(E

c(l)

.9UJ

373 1251 91 3

1 61 0

43

5222

Factor Structure

To define the factor structure of the init ialversion of the questionnaire, the data weresubjected to a principal components analysis withoblique direct oblimin rotation; an oblique rotationwas selected since all four sub-scales werepositively and significantly intercorrelated. ltemswere retained only if they had a factor loading of0.40 or higher on a factor and if they had not a highsecondary loading on another factor (<0.40). Afterrotating the solution and eliminating any items thatmet the above criteria, 52 items were retained,accounting tor 40"/o of the total variance.

Each factor is described below. Factor 1 wasnamed Use of Emotions for Facilitating Thinking

(UF). The fifteen items that compose this factor arerelated to the ability of the individualto harness theirown emotions in order to solve problems viaoptimism and self-assurance, two emotional statesthat facilitate inductive reasoning and creativity. Thecontent of the items in this factor resembles Mayerand Salovey's (1997) "Emotional Facil itation ofThinking" scale, Factor 2 was termed Caring andEmpathy (CEmp); this factor consists of fifteen itemsthat are related to the willingness of the individual tohelp other people and his/her ability to comprehendanother's feelings and to re-experience them. Thisfactor taps characteristics that are similar to thoseincluded into Mayer and Salovey's "Understanding

and Analyzing Emotions" scale. Factor 3 wastermed Control of Emotions (CE); it consists of

208 a loannis Tsaousis

twelve items that are related to the ability of theindividual to control and regulate emotions inthemselves and others, and seems to correspond toMayer and Salovey's "Reflective Regulation ofEmotions" scale. Factor 4 was named Expressionand Recognition of Emofrons (ER); the ten items thatcompose this factor are related to the ability of theindividualto express and recognize accurately theirown emotional reactions, and taps characteristicsthat correspond to Mayer and Salovey's"Perception, Appraisal and Expression of Emotion"scale. A sample of items in each sub-scale appearsin the Appendix.

Verification of the factor structure

To explore the factor structure of the f inalversion of GEIS i tems, a pr incipal componentanalysis wi th obl ique direct obl imin rotat ion wasperformed in a new sample (Sample 2).According to the resul ts, a four- factor solut ionemerged (see Figure 1). The first factor explained17.83% of the total variance; the second factorexplained 10.76%, and the thi rd and fourthfactors, an addi t ional 7.53% and 5.58%,respectively (a total of 42% of the explainedvar iance). The corresponding resul ts arepresented in Table 1. Final ly, separate factoranalyses were conducted for males and femalesin order to invest igate the stabi l i ty of the factorstructure across gender. The resul ts indicatedthat, wi th very few except ions, a l l i tems wereloaded on the same factors in both cases.

Reliabil ity Analysis

lnternal Consistency. The coefficient alphareliabil ity index for each scale of the final versionof the GEIS was as fo l lows: 0.80 for ER scale,0.83 for CE scale, 0.92 for UF scale, and 0.83 forCEmo scale. These results indicate that GEIS is arel iable test , s ince al l scales meet the minimumcr i te r ion o f >0 .70 (Nunna l ly & Berns te in , 1994) .The corresponding rel iabi l i ty index for the totaltest was also high (0.89).

fest-retest reliability. The Pearson product-moment correlation coefficients testing the test-retest re l iabi l i ty of the four scales of the f inalversion of the GEIS were also above the minimumrequirement value of 0.70. Particularly, it was 0.78for ER scale, 0.83 for CE scale, 0,92 for UF scale,and 0.76 for CEmp scale. The test-retest reliabil i tycoefficient for the total test was also high (0.90).

Convergent and Discriminant Validity

Table 2 presents the correlat ions betweenthe GEIS scales and al l cogni t ive and emot ionalvar iables used in th is study, In the top sect ion ofthe table are the intercorrelations among the fourscales of the GEIS. As can be seen, all scares areintercorrelated to each other, a resul t whichsuggests that the conceptual model comprisesconstructs which are related to each other. Thisresul t a lso explains why the percentage ofvar iance explained by the four factors ismoderate (42%).

The next sect ion of the table presents thecorrelat ions between GEIS scales and the BigFive dimensions of personal i ty (TPQue) Al lcorrelat ions were consistent wi th thoseant ic ipated according to the resul ts f rom otherstudies. More speci f ical ly, Neurot ic ism shares aconsiderable amount of variance with most of theGEIS scores, particularly with the CE scale (-0.60)and the U F sca le ( -0 ,58) . Fur thermore ,Extraversion is correlated posi t ivelv wi th the UFscale (0.39) and the CEmp scale (0.34). Bothpersonal i ty dimensions are also correlated withGEIS overal l score (-0.54 and 0.38, respect ively) .Final ly, Agreeableness is correlated posi t ivelywith CEmp scale (0.35) as wel l as wi th the GEISoveral l score (0.24). The remaining twopersonal i ty dimensions (Conscient iousness andOpenness to Exper ience) demonstrated ei therlow or no signi f icant correlat ion wi th GEIS scales.

In terms of cogni t ive abi l i ty measures, theresul ts in Table 2 ( top) indicate that e i ther f lu idintel l igence (as measured by Raven ProgressiveMatrices) or crystall ized intell igence (as measured

----4

Table 1Factor structure of the GEIS Scales

Item No Use of Emotion Caring andEmpathy

Control ofEmotion

Expression andRecognit ion

2239I

401 2324845443

2729421 6254 124342 1461 01 9375

28505 1331 4361 5357

3 1't7

64721

385249201 31 1231 8434

269

30

o.820.810.800.790.730.720.710.690.670.650.640.640.600.530.500 . 1 20 . 1 00.080 . 1 40.090 . 1 10 .010.050.090 , 1 90.05

-0.03-0.030 0 30.24n 1 q

0 .210.240.220 .210.290.330 . 2 10.070 . 1 70.000.300 . 1 20 . 1 1

-0.09-0 .160.060.200.06

-0.030.25

-0 03

17.83

0.080.040.030.050.060.000 . 1 40.050.060.030 . 1 10.29

-0.010 . 1 00.090.710.710.680.670.640.620.610.590.550.550.510.500.460.450.400 . 1 20 .210 0 40.290.080.08

-0.050.020.020 . 0 10 . 1 B

-0.210.240 . 1 80 . 1 20 . 1 5

-0.020 . 1 90.050 3 0

-0.070 . 1 3

0.200 . 1 80.250 . 1 60. ' t 10 . 1 90 .210.230.280 3 40.200 1 90 . 1 90 . 1 70 .210 . 1 40.070 .000 . 1 00.05

0.050.030 .000 . 0 10.04

-0,03-0.050.020.250.25

-0.02-0.030 1 90 .01

-0.200.090 .040 . 1 50 1 00 .210 . 1 00 0 80.220.220 . 1 20 . 1 0

-0.120.05v . z o

0 . 1 40.230 1 80 . 1 4

-0 030 .210 . 1 70.020.200.07

-0 080 .210 . 1 00.720.660.640.620.s90.580,550.530.490.45

5.58

0 . 1 40 . 1 70 1 30 . 1 60 3 00.06

-0.04-0 .130 . 1 4

-0.040.750.690.690.630.630.610.590.580.580.520.430.400 .210.29

-0.04-0 .1 10.200.330 .210 . 1 00 . 1 40 , 0 1

% Variance 10.76 7.53

Note: values in boldface indicate the items that load on the corresponding factor

210 a loannis Tsaousts

Table 2Correlation coefficients of the GEIS scales with various criterion scales

Criterion Scales ER CE UF CEmp Overall El

The GEIS Scales (N=1210)Expression & Recognition of EmotionsControl of EmotionsUse of Emotions for Facil itating ThinkingCaring & Empathy

0.1 6** 0 . 1 5 * *0.39**

0.24**0 .1 4" *0.22**

0.53**0.68**0.78**0.56**

The TPQUe (N=180)NeuroticismExtraversionAgreeablenessConscientiousnessOpenness to Experience

-0.01

0.1 6*0.06

-0 .1 4-0.09

-0.60**

0.040.030 . 2 1 * *

-0.04

-0.58**

0.39**0 . 1 6 *0 . 1 B *0.08

-0 .01

0.34**0,35**

-0 .01

0 . 1 2

-0.54**

0.38**0.24**0 . 1 30.04

Raven Progressive Matrices (N=70)Total Score -0 .1 1 -0.02 -0.1 0 -0.17 -0 .1 4

AHa (N=105)Verbal ReasoningPerceptual ReasoningTotal lQ Score

-0.17-0 .1 5-0.17

0 . 0 10.090.05

0 .010.090.05

-0.02-0 .1 1-0.07

-0,06-0 .01-0.03

The SSRI (N=152)Overall El 0.30** 0 . 1 5 * * 0.57** 0.49** 0 . 6 1 * *

The TAS (N=236)ldentify & Describe FeelingsCommunicate with EmotionsLimited DaydreamingDifficulty for Externally ThinkingOverall TAS Score

0.29**0.28**0.63"*0.68**0.52**

0 . 2 1 * *0.39**0.43**0.79**0"43"*

0.29**0.34**0.46**0,86**0.47**

0.56**0.60**0.70**0.73**0 .7 4**

0 . 4 1 * *0.49**0.65**0 . 9 1 * *0.64**

The TMMS (N=236)AttentionRepairClarityOverall TMMS Score

0.75**0.75**0.57**0.78**

0.59**0.70**0.58**0.69**

0,62**0.74**0.86**0.78**

0 .86**0.80**0.73**0.89**

0.83**0.89**0.83**0.94**

The SSI (lt=236)Emotional ExpressivityEmotional Sensitivity

0.77**0.62**

0.40**0.55**

0.59**0.67**

0.78**0.85**

0 .75**0 . 8 1 * *

Measuring trait emotional intelligence ) 217

Criterion Scales ER CE UF CEmp Overall El

Emotional ControlSocial ExpressivitySocial SensitivitySocial ControlOverall SSI Score

0.44**0,68**0.60**0 .71**0.74**

0.70**0.46**0.42**0.67**0 .61* *

0.65**0.63**0.45**0.75**0 . 7 1 * *

0,69**0 . 7 1 * *0.82**0.79**0.89**

0 .7 4**0.74**0,68**0.87**0.88**

The EES (N=236)Overall Empathy Score 0.70** 0.59** 0.63** 0.90** 0.84**

The SWLS (N=2261 0 . 1 4 * 0 . 1 0 0.33** 0 .09 0,29**

The PANAS (N=226)Positive AffectNegative Affect

0 . 1 9 * **0. 1 B**

0 .02-0.47**

0 . 5 1 * *-0.30**

0.27**-0.24**

0.44**-0.48**

Locus of Control (tV=213)LOC Total -0.12 -0,035** *0 .41** 0.00 -0,39**

The ASSET (N-2121Work RelationshipsOverloadControlRessources & CommunicationPay & BenefitsYour JobCommitment of the Employee to the

OrganisationCommitment of the Organisation to the

EmployeePhysical well-beingPsychological Well-being

-0 .21**-0.22**-0.23**-0.27**-0.33**-0 .1 1

0 .04

-0 .1 0-0.08-0.09

-0,36**-0.45**-0.38**-0.37**-0.26*-0.28**

0.32**

0.29**-0.43**-0.57r"k

-0.48**-0.40**-0 .51**-0.32**-0 .1 0-0.37**

0.56**

0.52**-0.44*x-0.63**

- 0 . 1 0-0 .1 5*-0 .1 5*-0 .1 0-0.28**

0.04

0.21**

0.23**0 .02

-0 .1 3

-0,48**-0.49**-0.53**-0.42**-0.34**-0.34**

0.49**

0.42**-0 .41**-0 .61**

Table 2 Continued

NqIe. ER= Expression & Recognit ion, CE= Control of Emotions, UF= Use of Emotion for Faci l i tat ing Thinking,CEmp- Caring & Empathy, TPQue= Traits Personality Questionnaire, SSRI = Schutte Self-Report Inventory, TAS=Toronto Alexithymia Scale, TMMS= Traits Meta-Mood Sca.le, SSI= Social Skills Inventory, EES= Emotion EmpathyScale, SWLS= Satisfaction with Life Scale, PANAS= Positive and Negative Affect Schedule* p < 0 . 0 5 , * * p < 0 . 0 1

212 a Ioannis Tsaousis

by AHa) showed no correlat ion wi th the GEISscales, suggesting that El, at least as measuredby the GEIS, is independent of the standardcognitive abil ity construct.

With respect to the emot ional intel l igenceconstructs, the correlations were in the expecteddirect ion and signi f icant, Part icular ly, a l l GEISscales exhibited positive correlations with the TASsub-scales (mean r=0.54), the TMMS sub-scales(mean r=0.76), and the Schutte Sel f -Reportlnventory Overal l scale (mean r=0.42).Furthermore, they exhibi ted strong posi t ivecorrelat ions wi th the SSI sub-scales (mean r=0,68), and the EES scale (correlat ions rangedfrom 0.59 to 0.90).

Regarding the wel l -being measures used inthis study (SWLS, PANAS, and two measuresirom the ASSET test), the results (Table 2,bottom) showed correlations moderate inmagnitude and in the predicted direct ion. Forexample, it was found that SLWS was positivelycorrelated with UF (0.33) and Overal l El score(0.29). Moreover, GEIS scales were posi t ivelycorrelated with PANAS Positive Affect scale(mean r=0.35), and negat ively wi th PANASNegative Affect scale (mean r=0.33). Finally, theASSET physical and psychological wel l -beingscales were negatively correlated with CE (-0.43and -0.57, respect ively) , UF (-0.44 and -0.63,

respec t ive ly ) , and Overa l l E l score ( -0 .41 and-0.61, respectively). Additionally, it was found thatLocus of Control scale was negatively correlatedwith both CE (-0.35) and UF (-0.41) scales, as wellas with El overall score (-0.39). There was nocorrelation between locus of control and either ERor CEmp scales.

Final ly, the correlat ion coeff ic ients betweenGEIS scales and ASSET's occupat ional stressindicators were almost all negative and significant(only one was positive but it was not significant).Simi lar ly, in terms of ASSET's two JobCommitment scales, s igni f icant posi t ivecorrelations ranging from 0.21 to 0.56 with four ofthe f ive GEIS scales were found, the except ionbeing the ER scale (0.04 and -0.11, both ns.) .

4. Discussion

The purpose of this study was to (a) developa rel iable and val id sel f - report measure of theconstruct of emot ional intel l igence, and (b) toprovide val id i ty evidence which just i fy that thenewly developed instrument is a measure of thetrai t emot ional intel l igence, a theoret icalperspective which argues that El can be viewedwithin the context of the personality domain.

In terms of the f i rst goal , the GEISdemonstrated acceptable psychometric properties,which justify its use as a reliable and valid measureof El. More specifically, the factor analytic datasuggest a four-factor solution, which bears a closeresemblance to Mayer and Salovey's (1997)theoreticalframework. lt should be reminded at thispoint, that this theoretical model is based on Mayerand Salovey's (1997) early work, where El is nottreated exclusively as an abil ity model, as it wassuggested in their later work (Mayer et al., 1999). Allscales demonstrated high internal consistency,indicating that they are homogeneous in theirmeasurements. Furthermore, test-retest datacovering a four-week period indicates the temporalreliabil i ty of the GEIS.

One o f the main goa ls dur ing thedevelopment of th is instrument was thedemonstrat ion of the convergent as wel l as thediscr iminant val idat ion of the GEIS scales. Thedata f rom the studies reoorted herein orovidesupport for good convergent and discriminantvalidity of the GEIS scales, suggesting that the testtaps a fairly broad range of related emotionalconstructs. On the one hand, al l the GEIS scalesdemonstrated moderate to high positive correlationcoeff ic ients wi th constructs such as empathy,social ski l ls (social intel l igence), emot ionalexpressiveness, and wel l -being, On the otherhand, the GEIS scales were correlated negativelywith constructs such as locus of control, negativeaffect, low physical and psychological well-beingand work stress. Moreover, the GEIS scalesevidenced moderate to high posi t ive correlat ioncoeff ic ients wi th two instruments, which direct lv

Measuring trait emotional intelligence ) 213

or indirectly are used as measures of El: the TraitMeta-Mood Scale (Salovey et a l . , 1995) and theSchutte Sel f -Report Inventory (Schutte et a l , ,1998). These resul ts just i fy the concurrentvalidation of the newly developed instrument.

To invest igate the second goal , namelywhether the GEIS is a measure based on the traitemot ional intel l igence tradi t ion, the GEIS scaleswere correlated with a personal i ty measure aswel l as wi th both types of intel l igence ( f lu id andcrystal l ized). The resul ts f rom the analysisshowed that GEIS scales were correlatednegat ively wi th Neurot ic ism and posi t ively wi thExtraversion. Low but s igni f icant correlat ionswere also reported with Agreeableness andConscient iousness dimensions. The order ofmagnitude of these correlations was comparableto that found previously in the l i terature (e.9.Davies, Stankov & Roberts, 1998, Dawda & Hart,2000. Fr iedman et a l . , 1980. Newsome, Day &Catano, 2000. Roger & Naj iar ian, 1989. Van DerZee, Thijs & Schakel, 2002). The only dimensionnot correlated with GEIS scales was theOpenness to Exper ience, a scale which, in anycase, has previously been related to cogni t iveabil it ies (Costa & McCrae, 1992).

Conversely, GEIS scores were unrelated toboth f lu id intel l igence (measured by RavenProgressive Matrices) and crystall ized intell igence(measured by AH4 test) , a resul t which isconsistent wi th theoret ical considerat ionsaccording to which trait El is related to personalitybut not to indicators of lQ (Carrol l , 1993.Newsome et al . , 2000. Petr ides, P6rez-P6rez-GonzAlez & Furnham ,2007. Saklofske, Austin &Minski , 2003), whi le the opposi te has been foundwith abil ity El (e.9., for a meta-analytic review ofpersonal i ty and abi l i ty correlates of El , see VanRooy, Viswesvaran & Pluta, 2005). Once again,such results stress the necessity of changing traitEl 's label in order that i t does not contain thenot ion of " intel l igence" anymore (Mikolajczak etal . , in press).

This c lose relat ionship between trai t El andpersonal i ty has brought up the issue of

distinctiveness between the two constructs. Onthe one hand, some researchers argue that trait Elis nothing more than a blend of wel l -establ ishedpersonal i ty t ra i ts, and as a consequence, i tsmeasurement does not offer something new to thestudy of individual differences (Matthews, Zeidner& Roberts, 2002. Schutte et al., 1998). On the otherhand, there are studies which support theincremental validity of the trait El over personalityin the predict ion of var ious l i fe outcomes. Forexample, Palmer, Donaldson, and Stough (2002)have shown that t ra i t El explains a considerableamount of variance of l i fe satisfaction even aftercontrol l ing for personal i ty var iance. Addi t ional ly,Saklofske, Austin, and Minski (2003) have foundthat trait El explains life satisfaction (positively) anddepression-proneness (negatively) over and abovethe basic personality dimensions. Finally, Petrides,P6rez-P6rez-Gonz6lez & Furnham (2007) havereportod that trait El was incrementally associatedwith ruminat ion, l i fe sat isfact ion, depression,dysfunctional attitudes, and coping after the effectsof personality have been controlled for. The resultsfrom this study provide further supporting to theargument that t ra i t El is mainly related tocharacteristics i n the affective/personal ity do mai n,and less wi th ski l ls in the cogni t ive domain.

A possible l imi tat ion of th is study could bethat convergent and discr iminant val id i ty resul tsare based on cross-sect ional sel f - reports,resulting in possible contamination from commonmethod var iance. In th is case, one could arguethat the correlation between the measures wil l beh igher than i t idea l l y shou ld be becausepart ic ipants wi l l apply the same biases to eachtask. Simi lar ly, i t cannot be excluded that theabsence of re lat ionship between trai t El andintel l igence tests was simply the product ofdivergent measurement methods (sel f - reportversus performance), just l ike the quasi nul lrelationship between abil ity and trait El, which areuncorrelated although their sampling domains areclosely related. For that reason, future researchshould be focused on the val idat ion of the GEISscales v ia exoer imental rather correlat ional

,---4

2l4 a loannis Tsaousis

studies. For example, i t could be examinedwhether there is any correspondence betweenpeople's self-perceptions of their abil ity torecognize, process, and uti l ize emotion-ladeninformation and their specific actual abil ity toidentify facial expressions.

To sum up, the results from this study justifythe GEIS as a reliable and valid measure of traitemotional intell igence. Furthermore, they provideevidence which support a basic premise of trait Eltheory, that seltreport questionnaires of Eloperationalise a construct that is unrelated tocapabi l i t ies, competencies, and ski l ls . Rather,these questionnaires can be used as themeasurement vehic le of a constel lat ion ofemotion-related self-perceptions and dispositionsthat is located at the lower levels of personalityhierarchies.

References

Aust in, E. J. , Saklofske, D. H., Huang, S. H. S. &McKenney, D. (2004). Measurement ofemotional intell igence: testing and cross-validating a modified version of Schutte et al. 's(1998) measure. Personality and lndividualDifferences, 36, 555-562.

Bar-On, R. (1996). The era of the EQ: Defining andassessing emotional intelligence. Paperpresented at the 100th annual convention of theAmerican Psychological Association, Toronto.

Bar-On, R. (1997). The Emotional Quotient lnventory(EQ-i): Technical manual. Toronto, Canada:Multi-Health Systems, Inc.

Bar-On, R., Brown, J. M., Kirkcaldy, B. D. & Thome, E.P. (2000). Emotional expression and implicationsfor occupational stress; an application of theEmotional Quotient Inventory (EO-i). Personalityand lndividual Differences, 28, 1 107-1 1 18.

Bar-On, R. & Parker, J. D. A. (Eds). (2000). Ihehandbook of emotional intelligence: Theory,development, and application at home, school,and in the workplace. San Francisco: Jossey-Bass.

Benet-Martinez, V. (2006). Cross-Cultural PersonalityResearch: Conceptual and Methodologicallssues. In R. W. Robins, C. Fra ley & R. F.Krueger (Eds), Handbook of Research Methodsin Personality Psychology. London: GilfordPress.

Boyatzis, R. E., Goleman, D. & Rhee, K. (2000).

Clustering competence in emotional intelligence:Insights from the Emotional CompetenceInventory. In R. Bar-On & J. D. A. Parker (Eds),Handbook of emotional intelligence (pp. 343-362,). San Francisco: Jossey-Bass.

Carrol l , J. B. (1993). Human cognit ive abi l i t ies: Asurvey of factor-analytic studies. New York:Cambridge University Press.

Cartwright, S. & Cooper, C. L. (2002). ASSEL Anorganisational sfress screening fool. London:Robertson Cooper Limited & Cubiks.

Caruso, D. R. & Mayer, J. D. (1999) . A quick scale forempathy. Manuscript for preparation.

Cattel l , R. B. (1966). The Scree test for the numberof factors. Multivariate Behavioural Research. 37.209-224.

Cattell, R. B. (1971) . Abilities: Their structure, growth

and action. New York: Houghton Miff l in.Ciarrochi, J., Deane, F. & Anderson, S. (2002).

Emotional intelligence moderates the relationshipbetween stress and mental health. Personali tyand lndividual Differences, 32, 197-209.

Ciar roch i , J . , Forgas, J . P. & Mayer J . D. (Eds) .(2001). Emotional lntelligence in every day life: Ascientific inquiry. Philadelphia, PA: PsychologyPress.

Costa, P. T. & McCrae, R. R. (1992). Revised NEOPersonality lnventory WEO-PIR) and NEO FiveFactor lnventory professional manual. Odessa,FL: Psychological Assessment Resources.

Conte, J . M. (2005) . A rev iew and cr i t ique ofemotional intel l igence measures. Journal ofOrganizational Behavior, 26, 433-440.

Davies, M. , Stankov, L . & Rober ts , R. D. (1998) .

Emot ional In te l l igence: In search of an e lus iveconstruct. Journal of Personality and Socia/Psychology, 25, 989-1 0 1 5.

Dawda, D. & Hart, S. D. (2000). Assessing emotional

' .

Measuring trait emotional intelligence ) 215

intel l igence: rel iabi l i ty and val idity of the Bar-On

Emotional Quotient Inventory (EO-i) in university

students. Personality and lndividual Differences,

28,797-812.Diener, E., Emmons, R. A., Larsen, R. J. & Grif f in, S.

(1985). The Satisfaction With Life Scale. Journal

of Personality Assessm ent, 49, 7 1 -75.

Dulewicz, V. & Higgs, M. (2000). ElQ-Managerial

User Guide. Windsor: NFER-Nelson.

Dulewicz, V. , H iggs, M. & Slask i , M. (2003) .

Measuring emotional intel l igence: content,

construct, and criterion-related validity. Journal

of Managerial Psychology, 1 8, 405-420.

Fr iedman, H. S. , Pr ince, L . M. , R igg io , R. E. &

DiMat teo, M. R. (1980) . Unders tanding and

assessing nonverbal expressiveness: The

Affective Communication Test, Journal of

Personalig and Social Psychology, 39, 333-351.

Geher , G. , Warner , R. M. & Brown, A. S. (2001) .

Predict ive val idity of the emotional accuracy

research scale. lntelligence, 29, 373-388.

Goleman, D. (1998a). Working with Emotional

lntelligence. New York: Bantam.

Goleman, D. (1998b). What makes a leader? Harvard

Buslness Review, 76, 93-102.

Heim, A. W. , Wat ts , K. P. & Simmonds, V. (1970) .

AH4, AHs and AHO fests. Windsor: NFER.

Kline, P. (1993). The handbook of psychological

{esfrng. London : Routledge.

Lane, R. D., Quinlan, D., SchwarIz, G., Walker, P. &

Zei t l in , S. (1990) . The leve ls o f emot iona l

awareness scale: A cognit ive-developmental

measure of emotion. Journal of Personality

Assessment, 55, 124-134.

Law, K. S. , Wong, C. S. & Song, L . J . (2004) . The

construct and cri terion val idity of emotional

intel l igence ,and i ts potential ut i l i ty for

management studies. Journal of Applied

Psychology, 89 (3), 483-496.

Matthews, G., Zeidner, M. & Roberts, R. D. (2002).

Emotional intelligence: Science and myth.

Cambridge, MA: MIT Press.

Mayer , J . D. & Geher , G. (1996) . Emot ional

intel l igence and the identi f icat ion of emotion.

lntelligence, 22, 89-1 13.

Mayer, J. D. & Salovey, P. (1997). What is emotionalintell igence? ln P. Salovey & D. Sluyter (Eds),Emotional development and emotionali nte I I i g e n c e : Ed u c ati o nal i m p I i c ati ons. N ew Yo rk :Basic Books.

Mayer, J. D., Caruso, D. R. & Salovey, P. (1999).Emotional Intell igence meets traditionalstandards for an Intell igence. lntell igence, 27,267-298.

Mayer, J. D., Salovey, P. & Caruso, D. R. (2002).Technical manualfor the MSCEIT V.2.0. Toronto,Canada: MHS.

Mayer, J. D., Salovey, P., Caruso, D. R. & Sitarenios,G. (2003). Measuring emotional intell igence withthe MSCEIT Y2.0. Emotion, 3, 97-105.

Mikolajczak, M., Luminet, O., Leroy, C. & Roy, E.(2007). Psychometric properties of the TraitEmot ional Intel l igence Quest ionnaire (TElQue;Petrides & Furnham, 2003). Journal ofPersonalityAssessmenf, 28, 338-353.

Nay, W, R. (1979). Multimethod clinicalassessment.New York: Gardner Press.

Newsome, S., Day, A. L. & Catano, V. M. (2000).Assessing the predictive validity of emotionalintelligence. Personality and individualDifferences, 29, 1005-1 0 1 6.

Nikolaou, l. & Tsaousis, l. (2002). EmotionalIntel l igence in the Workplace: Explor ing i tseffects on Occupational Stress andOrganisational Commitment. The lnternationalJournal of Organizational Analysis, 10 (4),327-342.

Nunnally, J. C. & Berstein, l. H. (1994). Psychometrictheory (3rd ed.). New York: McGraw-Hill.

Palmer, B. R., Donaldson, C. & Stough, C. (2002).Emotional intel l igence and l i fe sat isfact ion.Personality and lndividual Differences, 33, 1091-1 1 0 0 .

P6rez, J. C., Petr ides, K. V. & Furnham, A. (2005).Measur ing t ra i t emot ional intel l igence. In R.Schulze & R. D. Roberts (Eds), lnternationalHandbook of Emotional lntelligence. Cambridge,MA: Hogrefe & Huber.

Petrides, K. V. & Furnham, A. (2006). The role of traitemotronal intel l igence in a gender-speci f ic

L-

216 a loannis Tsaousis

model of organizational variables. Journal ofApplied Social Psychology, 46, 552-569.

Petr ides, K. V. & Furnham, A. (2003). Trait emotionalin te l l igence: Behav ioura l va l idat ion in twostudies of emotion recognit ion and reactivi ty tomood induction. European Journal ofPersonality, 17, 39-57 .

Petr ides, K. V. & Furnham, A. (2001). Trait emotionalin te l l igence: Psychometr ic invest igat ion wi threference to establ ished trait taxonomies.European Journal of Personality, 15,425-448.

Pet r ides, K. V. & Furnham, A. (2000) . On thedimensional structure of emotional intel l igence.Personality and lndividual Differences, 29, 313-320.

Petrides, K. V. P6rez-P6rez-GonzAlez, J. C. &Furnham, A. (2007). On the cri terion andincrementa l va l id i ty o f t ra i t emot iona lintelligence. Cognition and Emotion, 21, 26-55.

Raven, J. C., Court, J. H. & Raven, J. (1979). Manualfor Raven's Progressive Matrices and VocabularySca/es. London: H. K. Lewis.

Riggio, R. E. (1989). Social Ski i ls tnventory. TheManual. New York: Consult ing PsychologistsPress.

Roger , D. & Najar ian, B. (1989) . The const ruct ionand val idation of a new scale for measurementemotion control. Personality and lndividualDifferences, 70, 845-853.

Rot ter , J . B. (1966) . Genera l ized expectanc ies forinternal versus external control of reinforcement.Psychological Monographs: General andApplied, 80,1-28.

Sak lo fske, D. H. , Aust in , E. J . & Minsk i , P. S. (2003) .Factor structure and validity of a trait emotionalintelligence measure. Personality and tndividualDifferences, 34, 707 -721 .

Salovey, P. & Mayer . J . D. (1990) . Emot ionallntel l igence. lmagination, Cognit ion andPersonality, 9, 185-21 1.

Salovey, P., Mayer. J. D., Goldman, S. L., Turvery, C.& Palfai, T. P. (1995). Emotionalattention. clari ty,and repai r : exp lor ing emot iona l in te l l igenceus ing the Tra i t Meta-Mood Scale . In J . W.Pennebaker (Ed.), Emotion, disclosure, and

health. (pp. 125-154). Washington, DC:American Psychological Associat ion.

Schut te , N. S. , Malouf f , J . M. , Hal l , L . E. , Hagger ty ,D. J . , Cooper , J . T . , Golden, C. J . & Dornheim, L .(1998) . Development and va l idat ion o f ameasure of emotional intel l igen ce. Personali tyand lndividual Differences, 25, 167-177.

Sifneos, P. E. (1973). The prevalence of alexithymiccharacterist ics in psychosomatic patients.Psychotherapy and psychosomatics, 22, 2SS-262.

Slaski, M. & Cartwright, S. (2002). Health,per formance and emot ional in te l l igence: anexploratory study of retail managers. Stress andHealth,78. 63-68.

Taylor, G. J., Ryan, D. & Bagby, R. M. ( '1985).Towardthe development of a new self-report Alexitymiascale. Psychotherapy and Psychosomatics, 44,1 9 1 - 1 9 9 .

Tet t , R. P. , Fox, K. E. & Wang, A. (2005) .Development and val ldation of a self-reportmeasure o f emot iona l in te l l igence as amult idimensional trait domain. Personali ty andSocral Psychology Bulletin, 37, 859-888.

Tsaousis, l . (2002). The Traits Personali tyQuest ionnai re (TPQue) . In B. De Raad & M.Perugrni (Eds), Big Five Assessmenf. Gott ingen:Hogrefe & Huber.

Van der Zee,K. , Th i js , M. & Schakel , L . (2002) . There la t ionsh ip o f emot iona l in te l l igence wi thacademic in te l l igence and the Big F ive.European Journal of Personali ty, I 6, '103-125.

Van Rooy, D. L., Viswesvaran, C. & Pluta, P. (2005).An Evaluation of Construct Val idity: What ls ThisThing Cal led Emot ional In te l l igence? HumanPerformance 1 8, 445-462.

Watson, D. , C lark , L . A. & Te l legen, A. (1988) .Development and val idation of brief measures ofposit ive and negative affect: The PANAS scales.Journal of Personality and Socia/ Psychology, 54,1 063- 1 070.

Wang, C. S. & Law, K. S. (2002). The effects ofleader and fo l lower emot iona l in te l l igence onperformance and att i tude: an exploratory study.The Leadership Quarterly, 1 3, 243-292.

,----4

Measuri.ng trait emotional intelligence ) 2I7

Appendix

Examples of items for GEIQ Scales

Scales Items

Expression & Recognition of Emotions Most people find it difficult to understand what I really feel (R)I find it difficult to express my feelings to the others (R)I rarely analyze my feelings (R)I tend to disregard my feelings.People can usually understand how I feel by just looking my face

Control of Emotions I get mad easi ly but this does not last long.I usual ly control my angerWhen I am in an emergency situation, I usual ly lose seltcontrol (R)When I experience unpleasant emotions, I usually react intensely (R)Before important events, I usually feel tense (R)

Use of Emotion for Faci l i tat ing Thinking l t is very dif f icult for me to be optimist ic (R)

I tend to focus on the negative side of a situation (R)I easi ly f ind alternatives when things are gett ing badMost of the time my problems do not affect my pedormanceI feel confident before important life events

Caring & Empathy I believe that I am a person who cares and helps othersI l ike to talk with others for their problemsI respect other people's feel ingsI do not care about other people's problems (R)I am interested in other people's motives

Note: items marked with (R) are reverse scored