Relations between implicit and explicit measures of personality - prospects of Implicit Association Test (IAT) in assessment of basic personality traits

Abstract

Introduction. The core idea behind the Implicit Association Test (IAT) is to obtain information that are not the result of controlled processes. Therefore, it is considered a technique with the potential to overcome some disadvantages of explicit measures, such as socially desirable responding. IAT consists of a task of double categorization whereby the respondents have to sort the stimuli belonging to superordinated categories. Study objective. The main objective of the study was to investigate the convergent and discriminative validity of IAT in the assessment of basic personality traits. Method. To investigate the relations between the implicit and explicit measures, Implicit Association Tests, several self-report measures, ratings by close others, and measures of cognitive abilities were administered to a sample of 224 university students at Belgrade University. Ratings by experts and linguistic measures were collected from a subsample of 99 subjects. The respondents completed the personality IAT, NEO-PIR and DELTA10 (S and R form), a short scale of attributes (DOCEAN), a short questionnaire for the assessment of relevant moderator variables, KOG9 battery test for the assessment of intellectual abilities, and self-assessment of intellectual abilities. In addition, they were observed by experts in structured interviews, in which linguistic parameters were collected through an automatic text analysis (LIWC). Results. The correlations between the self-reported and IAT Neuroticism (0.17) and Extraversion (0.16) were found to be small but significant (p<0.05). However, none of the paths from the personality IAT to the corresponding latent traits in SEM Correlated-Traits-Correlated-Uniqueness-MTMM-Model was significant. The model had the excellent fit: c2=338.18, p=0.00; RMSEA=0.042 (90% CI RMSEA 0.031- 0.053), SRMR=0.066 and CFI=0.96. The MTMM analyses on a subsample of 99 respondents, from which behavioural measures were collected, demonstrated relations iv between IAT and the parameters collected with automatic text analysis (LIWC). Subsequent analyses demonstrated a strong divergence between implicit and explicit method factors. Conclusion. Study has shown that the personality IAT measures have little in common with personality traits measured by the traditional self-report inventories or rating scales. However, MTMM analysis provides evidence about the predictive validity of IAT in spontaneous verbal behaviour.

Uvod. Ideja koja stoji u osnovi Testa Implicitnih Asocijacija (IAT) je prikupljanje informacija koje nisu pod uticajem kontrolisanih procesa, zbog čega se smatra da ova tehnika pruža mogućnost prevazilaženja nedostataka eksplicitnih mera, kao što je socijalno poželjno odgovaranje. IAT predstavlja zadatak dvostruke kategorizacije u kojem ispitanici svrstavaju stimuluse koji pripadaju nadreñenim kategorijama. Cilj. Glavni cilj istraživanja je bio da se utvrdi konvergentna i diskriminativna validnost IAT-a u proceni bazičnih dimenzija ličnosti. Metod. Kako bi se utvrdile relacije izmeñu implicitnih i eksplicitnih mera, na uzorku od 224 studenta beogradskog univerziteta, prikupljeni su podaci sa Testa Implicitnih Asocijacija, nekoliko mera samo-izveštaja, procene od strane bliskih drugih i mere kognitivnog funkcionisanja. Na poduzorku od 99 ispitanika prikupljene su i mere procene od strane eksperta, kao i lingvistički parametri. Primenjeni su sledeći instrumenti: IAT za procenu bazičnih dimenzija ličnosti, NEO PIR (S i R forma) i DELTA 10 (S i R forma), kratka skala atributa (DOCEAN), kratak upitnik za procenu relevantnih moderatorskih varijabli, KOG9 baterija za procenu intelektualnih sposobnosti i upitnik za samo-procenu intelektualnih sposobnosti. Dodatno, prikupljene su i ekspertske procene na osnovu posmatranja u strukturisanom intervjuu i lingvistički parametri na osnovu automatske analize teksta (LIWC). Rezultati. Korelacije izmeñu Neuroticizma merenim NEO PIR i IAT (0.17) i Ekstraverzije (0.16) su bile male ali značajne (p<0.05). Meñutim, nijedno od zasićenja sa IAT na korespodentnim latentnim faktorima u strukturnom MTMM modelu nije bilo značajno. Model je imao odlične indekse podesnosti: c2=338.18, p=0.00; RMSEA=0.042 (90% CIRMSEA 0.031-0.053), SRMR=0.066 i CFI=0.96. MTMM analiza na subuzorku od 99 ispitanika, na kojem su prikupljene i bihejvioralne mere, je kazala na relacije izmeñu IAT i parametara prikupljenih automatskom analizom teksta vi (LIWC). Naknadne analize su pokazale jako razmimoilaženje metodskih faktora izolovanih iz implicitnih i eksplicitnih mera. Zaključak. Istraživanje je pokazalo da mere prikupljene pomoću IAT imaju malo zajedničkog sa merama samoizveštaja ili procena od strane drugih. Ipak, MTMM analiza je ukazala na mogućnosti IAT u predviñanju spontanog verbalnog ponašanja.