You are on page 1of 16

European Journal of Personality, Eur. J. Pers.

26: 613628 (2012)


Published online 22 June 2012 in Wiley Online Library (wileyonlinelibrary.com) DOI: 10.1002/per.1861

An Inkblot for the Implicit Assessment of Personality: The Semantic Misattribution Procedure
FLORIN A. SAVA1*, LAURENIU P. MARICUOIU1, SILVIA RUSU1, IRINA MACSINGA1, DELIA VRG1, CLARA MICHELLE CHENG2 and B. KEITH PAYNE3
1 2

West University of Timisoara, Romania Carlow University, Pittsburgh, PA, USA 3 University of North Carolina at Chapel Hill, NC, USA

Abstract: Misattributions people make about their own reaction to ambiguous stimuli can be used to measure personality self-concepts implicitly. On the basis of a semantic misattribution priming paradigm [semantic misattribution procedure (SMP)], we assessed the implicit personality self-concept related to three dimensions included in the Big-Five model: conscientiousness, neuroticism, and extraversion. Across three studies (N1 = 98, N2 = 140, and N3 = 135), the SMP was robustly related, in the expected direction, to individual differences in self-reported personality questionnaires and managed to predict both self-reported and objectively measured behaviours. The main advantage of SMP over classical explicit measures of personality is its higher resistance to social desirability tendencies, although its psychometric properties are somewhat lower than those pertaining to explicit measures of personality. Finally, comparisons of our results with studies that used other implicit measures of personality selfconcept indicate that the SMP has higher criterion validity. Copyright 2012 John Wiley & Sons, Ltd. Key words: implicit measure; self-concept; personality; priming; affective misattribution procedure; semantic misattribution procedure; Big Five Personality assessment currently relies almost exclusively on explicit measures such as self-report questionnaires. However, two important limits have been consistently identied with regard to this approach. First is the inclination to provide socially desirable answers (e.g. Edwards, 1957; Furnham, 1997; Rothstein & Gofn, 2006). The second is the limit of introspection to accurately describe the self (e.g. Gosling, John, Craik, & Robins, 1998; Paulhus & Reid, 1991). Research on implicit social cognition from the last two decades has shown that implicit measures can provide an alternative way to assess many psychological constructs, without relying on self-reports (Gawronski & Payne, 2010). Until now, most research has focused on measuring implicit attitudes (Fazio, Sanbonmatsu, Powell, & Kardes, 1986; Greenwald & Banaji, 1995; Wittenbrink, Judd, & Park, 1997). Implicit measures have been used less in the area of personality assessment. Just as measuring implicit attitudes has produced major changes in the eld of social psychology and connected areas, we expect that new theoretical models will emerge from studying the implicit personality (see, for example, the Behavioral Process Model of Personality proposed by Back, Schmukle, & Egloff, 2009). From a pragmatic point of view, the resistance of implicit personality measures to social desirability is appealing for practitioners working in personnel selection (McDaniel, Beier, Perkins, Goggin, & Frankel, 2009; Schnabel, Banse, & Asendorpf, 2006). Thus, implicit measures of personality are promising for both theoretical and practical reasons. One of the main reasons for the rare use of implicit measures in the assessment of personality is the unacceptably low level of internal consistency found for many implicit measures (Gawronski, 2009). An exception is the use of Implicit Association Test (IAT) (Greenwald & Banaji, 1995) to assess what Asendorpf, Banse, and Mcke (2002) have dened as personality self-concept that part of ones self-concept that refers to descriptive attributes for personality (e.g. sociable, anxious, etc.). Several studies have suggested that the IAT could represent a valid implicit measure of personality (Perugini & Leone, 2009; Schmukle, Back, & Egloff, 2008; Schmukle & Egloff, 2005; Schnabel, Asendorpf, & Greenwald, 2008). The IAT possesses adequate psychometric properties in terms of its internal consistency (Gawronski, 2009) and predictive validity (Perugini, Richetin, & Zogmaister, 2010). However, such a measure may not be temporally stable across evaluations. Egloff, Schwerdtfeger, and Schmukle (2005) found a 1-year testretest reliability for IAT of .47, whereas a meta-analysis conducted by Hofmann, Gawronski, Gschwendner, Le, and Schmitt (2005) showed that the average testretest correlation for IAT was .51. Additional attempts to assess implicit personality self-concepts via different procedures were made by Schnabel et al. (2006), using the Implicit Association Procedure (IAP), by Teige, Schnabel, Banse, and Asendorpf (2004), using the Extrinsic Affective Simon Task (EAST), and by Boldero, Rawlings, Received 2 July 2011 Revised 20 April 2012, Accepted 26 April 2012

*Correspondence to: Florin Alin Sava, West University of Timisoara, Department of Psychology, 4 Vasile Parvan Blvd., Room 504, 300223 Timisoara, Romania. E-mail: afsava@socio.uvt.ro, afsava@gmail.com

Copyright 2012 John Wiley & Sons, Ltd.

614

F. A. Sava et al. we tested the SMP for measuring the personality selfconcept. In this direction, we used various personality descriptors (adjectives) as primes, whereas the ambiguous and neutral stimuli, consisting of Chinese ideographs, were used as targets. The task of a participant was to ignore the prime and to guess whether the immediately following Chinese ideograph ts him or her, or does not t him or her, in order to be printed on a personalized T-shirt. Because the target stimuli are ambiguous in their nature, we expect that a misattribution effect will occur, suggesting that a Chinese ideograph would be more likely judged as one that ts me if the associated prime is a word (e.g. shy) that is descriptive for the participants personality self-concept. We present three studies with the general aim to validate the SMP procedure as an effective implicit measure of personality. From a psychometric view, any successful implicit measure of personality, including the SMP, should meet several criteria: (i) reproducing the expected associations that have been previously demonstrated for a specic trait in terms of construct validity; (ii) providing an incremental predictive validity for behaviours over the amount explained by an explicit measure; (iii) manifesting more resistance than explicit measures towards socially desirable tendencies; (iv) being associated with stable, self-descriptive attributes (traits) and not with unstable, momentarily states; and (v) having adequate reliability in terms of internal consistency and testretest stability.

and Haslam (2007), using a Go/No-Go Association Task. These studies found these methods to suffer from a high amount of a specic method variance and to have inferior psychometric properties in comparison with explicit measures of personality. Therefore, the literature suggests that the IAT is the most practical implicit method currently in use for personality, although this measure also has limitations, especially stability over time, which is important for measuring traits. In this paper, we elected to study an alternative implicit method for measuring personality traits, derived from the affective misattribution procedure (AMP; Payne, Cheng, Govorun, & Steward, 2005). Unlike previous attempts to measure implicit personality, which were all based on similar mechanisms (e.g. task switching measures of the degree of association in memory between the self and specic personality traits; De Houwer, Teige-Mocigemba, Spruyt, & Moors, 2009), the AMP is based on a different mechanism, namely the misattribution of prime characteristics to neutral target stimuli (Gawronski, Cunningham, LeBel, & Deutsch, 2010; Payne et al., 2005). In a typical AMP task, participants are briey presented with a positive or a negative prime stimulus, which is followed by a neutral Chinese ideograph. They are asked to ignore the prime stimulus and to assess the neutral ideograph. However, the affective valence of the corresponding prime is not ignored by the participants, who are inclined to unintentionally evaluate the target as having the same valence as the prime. The mechanism behind such an effect is assumed to be the misattribution of a primes affective valence (which was supposed to be ignored) to its associated ambiguous and neutral target stimulus. This approach has excellent psychometric properties (e.g. internal consistency values over .80; Payne et al., 2005) and has been found to be less affected by motivations to present oneself favourably compared with explicit measures (Payne, Burkley, & Stokes, 2008). The robust properties of AMP were recognized by Gawronski (2009), who considered AMP as one of the only two implicit measures techniques (the other one being IAT) that consistently have shown reliable estimates that are acceptable from a psychometric point of view. Our main hypothesis, underlying the three subsequent studies presented in this article, is that the misattribution effect can be generalized from the affective aspects of a concept to its semantic connotation. In fact, recently, Deutsch and Gawronski (2009), as well as Imhoff, Schmidt, Bernhardt, Dierksmeier, and Banse (2011), found support for the extension of misattribution mechanism from affective reactions to semantic evaluations. These authors have shown that a misattribution effect could occur also for the semantic connotation of a neutral and ambiguous target stimulus. For instance, Deutsch and Gawronski (2009) found that participants tend to guess more frequently that a Chinese ideograph refers to an animate object if the primes were pictures of animals in comparison with cases where primes were pictures of inanimate things. The semantic misattribution procedure (SMP) is therefore a technique derived from AMP (Payne et al., 2005), on the basis of a misattribution of semantic meaning. In this article,
Copyright 2012 John Wiley & Sons, Ltd.

STUDY 1 As Perugini and Banse (2007) argued, any new personality measurement, regardless of its theoretical background, has to demonstrate adequate psychometric properties in terms of its validity and reliability. In the rst study, we looked at the preliminary psychometric properties of the SMP for the assessment of neuroticism and conscientiousness, two specic dimensions included in the Five-Factor Personality Model (Costa & McCrae, 1992). They were selected because they were found in previous research to be the personality factors most strongly biased by socially desirable answers (Viswesvaran & Ones, 1999). A core set of 32 self-descriptive markers for neuroticism and conscientiousness, found in previous works (Hofstee, de Raad, & Goldberg, 1992; Peabody & Goldberg, 1989; Perugini, Gallucci, & Livi, 2000; Saucier, 1994; Sava, 2008), were included as primes in the SMP task, to cover both ends of a specic dimension (e.g. high and low levels of neuroticism). Several explicit measures of conscientiousness and neuroticism (low emotional stability) were used to address the relationship between implicit and explicit measurements of these traits. Similarly, the relationship between the SMP measures and social desirability (Crowne & Marlowe, 1964), as well as that between the SMP measures and indices of trait (stable self-descriptive adjectives) and state (momentarily self-descriptive adjectives), was assessed. Finally, the ability of the SMP measures to predict specic self-rated and objectively observed behaviours was evaluated. We based our choice of behaviours on previously
Eur. J. Pers. 26: 613628 (2012) DOI: 10.1002/per

Semantic misattribution procedure reported associations between specic behaviours and the dimensions included in the Five-Factor Personality Model (Back et al., 2009; Bogg & Brent, 2004; Paunonen, 2003). Method Participants The initial pool of participants consisted of 120 undergraduate students, who voluntarily agreed to participate in this research in exchange for course credit and who signed an informed consent form. However, either because of missing data (n = 5) or because of unrealistically fast reaction times on the SMP task (having RT less than 100 milliseconds in more than 10% of trails) (n = 17), the analyses were conducted on a nal sample that included 98 participants (85% females, Mage = 20.8 years, SDage = 3.8, age range: 1941). None of these participants had prior experience in working with implicit measures. Materials The SMP task was used to measure implicit levels of conscientiousness (C SMP) and neuroticism (N SMP). We used Direct RT v 2008 (Empirisoft, 2008) to implement the SMP. The entire task consisted of only 32 trials, divided evenly to obtain 16 trials for each trait (e.g. eight descriptors for the high level of conscientiousness and another eight descriptors for the low end of conscientiousness). The set of adjectives used as primes is presented in the Appendix. The 32 primes were presented in a random order, and they were randomly paired with a Chinese pictograph as target, randomly selected from a list of 32 characters. All Chinese pictographs were assessed in a pre-test (N = 120) to be neutral (M = 4.26, SD = 0.31), on a 7-point (1 to 7) Likert scale. In each trial, the prime duration was set at 200 milliseconds, followed by a white rectangular mask of 125 milliseconds, for a total stimulus-onset asynchrony (SOA) of 325 milliseconds. Primes were followed by the Chinese pictograph presented for 200 milliseconds and a grey rectangular noise pattern that took the place of the target until the participant responded. A novel aspect of the SMP is that self-relevance was measured by having participants judge whether they would like each Chinese character to be printed on a personalized T-shirt by responding Fits me or Does not t me. To score the SMP, we summed all trials in which participants considered the target as a good t following a high trait-descriptive prime, together with the trials in which participants considered the target as not a good t if the target was paired with primes descriptive for low levels of the trait. Therefore, higher scores for C SMP and N SMP were indicative of higher levels of implicit conscientiousness and neuroticism, respectively. An Adjective Checklist was used as a form of explicit measurement of descriptors included in the SMP task. Therefore, 32 adjectives were evaluated using a 7-point rating scale, asking participants to assess how accurately these adjectives described them (1 = very inaccurate to 7 = very accurate). This checklist was used to address the relationship between the implicit and explicit self-concepts. High scores indicate a higher level of neuroticism and conscientiousness in the participants explicit personality self-concept.
Copyright 2012 John Wiley & Sons, Ltd.

615

Neuroticism and Conscientiousness Self-Report (Explicit) Measures were collected using three independent instruments developed within the Big-Five (or Five-Factor Personality Model) framework: the NEO Five Factor Inventory (NEOFFI) (Costa & McCrae, 1992), the 50-item of International Personality Item Pool Big-Five Factor Markers (IPIP) (Goldberg et al., 2006), and the DECAS Personality Inventory (Sava, 2008). Where necessary, we reversed the score so that higher scores reect higher levels of neuroticism and conscientiousness. The validity of the (less-known) DECAS inventory was conrmed in a series of studies (i.e. conrmatory factor analysis) (Sava, 2008, 2009; Sava & Popa, 2011) that included, among others, the Big Five Questionnaire (BFQ) (Caprara, Barbaranelli, & Borgogni, 2008), Goldbergs IPIP items (Goldberg et al., 2006), and the NEO-FFI (Costa & McCrae, 1992). To summarize the main previous ndings concerning its concurrent validity, we found an uncorrected mean correlation with other Big-Five concurrent scales of .65 for conscientiousness and .66 for emotional stability. The MarloweCrowne 33-item scale was used for measuring social desirability tendencies among participants (Crowne & Marlowe, 1960). However, it should be noted that the MarloweCrowne scale loads highly on both facets of social desirability (Paulhus, 1984): self-deception (referring to inaccurate self-knowledge affecting explicit evaluations) and impression management (intention to present oneself to others in more favourable terms than it is really the case). Higher scores indicate a higher level of social desirability in the participants answers. The Positive and Negative Affect Schedule - Expanded Form (PANAS-X) questionnaire (Watson & Clark, 1994) was used in its version focusing on measuring the state affect (using instructions to rate how participants had felt in this moment), to measure positive and negative affects. This was particularly important to address the issue whether an SMP task (i.e. N SMP) is correlated either with chronic aspects of the self-concept (traits) or with contextual mood aspects (states). Higher scores on PANAS suggest a higher level of positive mood (for the PA scale) and a more intense negative mood (for the NA scale). Behaviours. Both self-reported (SR) and objective (O) behaviours were assessed for each trait. For conscientiousness, SR behaviours concerned the participants smoking, drinking, and church attendance behaviours. Objective observations included participants last semester academic performance (obtained from the University) and their punctuality in arriving to the laboratory. For neuroticism, we used one SR behaviour (ever being a client in counselling/psychotherapy) and an O measure (a memory test to measure bias in the direction of processing negative stimuli more intensely than the positive ones) (see Bar-Haim, Lamy, Pergamin, Bakermans-Kranenburg, & van IJzendoorn, 2007). For the latter task, we used a lexical decision task that included 40 primary target words (20 positive words and 20 negative words), which later were included in a recognition task together with other 40 different, but semantically related, words (20 positive and 20 negative). We asked participants to hit a key if they recognize a word that was also presented in the lexical decision task. Prior research
Eur. J. Pers. 26: 613628 (2012) DOI: 10.1002/per

616

F. A. Sava et al. of personality, behaviours related to the personality traits), we used a single factor score resulting from principal component analyses. For example, a single latent score for explicit C was obtained through the use of principal component analysis having as input the manifest scores from the three explicit scales reecting conscientiousness from NEO-FFI, IPIP, and DECAS. All scores representing these latent variables were z standardized scores; therefore, they have a mean of 0 and an SD of 1. Interested readers can nd in Table 1 a correlation matrix among all core variables included in this study along with their descriptive statistics and their reliability estimates. Additional correlations can be found in an online supplement le containing full correlation matrixes and details on hierarchical regressions associated with these variables. Furthermore, the distributions of all explicit and implicit measures were checked, but no variable required transformation to correct for excessive skewness. For the memory bias task, we employed a Signal Detection Theory strategy (Macmillan & Creelman, 2005), turning the hits and the false alarms for both positive and negative words into probabilities, and computing both sensitivity index (d0 ) and bias indicators (c). A higher value on sensitivity index shows a higher focus on the task (more words correctly recognized and fewer false alarms), whereas a higher bias indicates a tendency to commit more recognition errors adjusting for a persons sensitivity. Results Construct validity of the SMPs was assessed by examining the convergent and discriminant relationship among the SMPs and their respective explicit and behavioural measures. Associations with explicit personality measures As expected, the C SMP was positively correlated to explicit measures of C, and not correlated with N explicit measures, showing good discriminant validity (Table 1). This pattern of relationship was only partially reproduced for the N SMP, which was positively correlated with explicit N (r = .18, p < .05, one-tailed test), but also negatively correlated with explicit C. Of particular interest is the discriminant validity between the implicit and explicit forms of self-concept. Thus, the C SMP correlated with the explicit C descriptive adjectives but not with explicit N descriptive ones, whereas N SMP correlated with explicit N descriptive adjectives but not with the explicit C descriptive ones. All these divergent results are statistically signicant at ps < .05 (two-tailed), when z tests for establishing differences between paired correlation coefcients are employed. Moreover, neither C SMP nor N SMP correlated with measures of affective state, suggesting that the SMP task is not affected by the participants mood but reaches the chronic self-concept schemas. Another expected result was that implicit measures were less associated with social desirability than explicit measures were. Explicit measures of personality self-concept and the explicit personality trait measures were strongly associated with social desirability. In contrast, C SMP and N SMP were not signicantly correlated with social desirability using two-tailed tests.
Eur. J. Pers. 26: 613628 (2012) DOI: 10.1002/per

has indicated that threatening stimuli capture the attention of anxious persons, making them easier for them to encode (Bar-Haim et al., 2007). Procedure Participants were tested in a laboratory, in groups of 5 to 10 persons, for the explicit measures, where we used paper and pencil materials. They completed implicit measures in individual rooms seated in front of a computer. The order of presenting the instruments was not counterbalanced between packages (explicit versus implicit measurements), but only within packages. First, participants completed the implicit measures, followed by the explicit measures. Concerning the SMP task, the participants were seated in front of a computer and were informed that they will have to do performance tasks in which they have to take simple decisions very quickly, in an analogous manner with the instruction used by Payne et al. (2005). The verbatim instruction was the following: In the following task you will have to make a series of judgments as fast as you can. They are very simple; the difcult part is that you have to work very fast. You will see on the screen successive pairs of stimuli, the rst being an adjective and the latter being a Chinese character. The role of the rst stimulus in a pair of stimuli will be to draw your attention ready for processing the second stimulus. There is no semantic relationship between the two stimuli included in a pair. Your task will be to decide whether the Chinese character would be suitable for you to be printed on a personalized T-shirt, which should t you. To respond, you either press the C key (if the Chinese pictograph rather ts you) or press the N key (if the Chinese pictograph rather does not t you). To provide quick answers, please keep your two index ngers above the two assigned keys. Press the space bar for a practice block. After a set of eight trials consisting of self-descriptive adjectives (i.e. intelligent), but unrelated to C or N dimensions, a second block of instructions was provided. It is important to notice that some of the adjectives presented prior to the Chinese pictograph may affect your decision to consider the Chinese symbols as appropriate or inappropriate as logos for your personalized T-shirt. Therefore, please try your absolute best to not let those adjectives to bias your judgment whether the Chinese drawings are appropriate or not to be printed on your personalized T-shirt. The participants were then reminded the key to use for their responses and were asked to address the research assistant if they had questions regarding the SMP task. In the task, each pictograph was randomly paired with a prime adjective in a new random order generated by the computer program for each participant. Because the task consisted of only 32 trials, it took 1 to 2 minutes to complete the SMP. Data analysis For purposes of simplicity, when more than one measurement was used for a specic construct (i.e. explicit measures
Copyright 2012 John Wiley & Sons, Ltd.

Semantic misattribution procedure


Table 1. A matrix of inter-correlations among core variables of interest in the study (Study 1, N = 98) Variables Implicit personality self-concept Conscientiousness SMP C (1) Neuroticism SMP N (2) Explicit personality traits Latent (explicit) C (3)a Latent (explicit) N (4)a Explicit personality self-concept Adjectives C (5) Adjectives N (6) Other relevant variables Social desirability (7) State positive affect (8) State negative affect (9) Behaviours C latent behavioural index (10)a Negative words sensitivity (11) Negative words bias (12) Mean Standard deviation 1 .46 .21 .31 .02 .31 .05 .20 .03 .14 .18 .19 .13 9.1 2.1 2 3 4 5 6 7 8 9 10 11

617

12

.53 .21 .18 .18 .28 .16 .00 .02 .14 .20 .19 7.8 2.4 .14 .82 .23 .54 .22 .09 .47 .16 .05 0.0 1.0 .09 .77 .29 .09 .33 .00 .04 .09 0.0 1.0 .86 .12 .47 .17 .06 .49 .01 .09 74.1 11.8 .82 .21 .01 .25 .06 .02 .08 56.7 10.6 .74 .10 .09 .39 .13 .02 17.4 4.5 .74 .08 .08 .06 .11 30.9 4.7

.85 .06 .25 .22 14.2 5.0 .03 .06 0.0 1.0 .49 .18 0.48 0.24

.52 0.47 0.18

Note: SMP, semantic misattribution procedure. Bold type correlations are signicant at ps < .05 (two-tailed test) (|r| .21, p < .05, two-tailed test; |r| .18, p < .05, one-tailed tests), and italic type for estimated internal consistencies. a It represents a composite score resulted from a principal component analysis that included the following: (a) the relevant scales from the three 5-factor personality instruments (NEO-FFI, IPIP, and DECAS) for the explicit measures of C and N, and (b) all self-reported and objective C-related measured behaviours.

Associations with behaviour measures The C SMP correlated in the expected direction with the C-related behavioural index (r = .18, p < .05, one-tailed test). Similar but larger correlations were found among the explicit measures of C and the C-related behavioural index. A positive correlation between N SMP, but not explicit N measures, and sensitivity to negative words was found, suggesting that people high on implicit N tend to process negative stimuli more deeply (r = .20, p < .05, one-tailed). Similarly, a negative correlation between the N SMP, but not explicit N measures, and memory bias was found, suggesting that people high on implicit N have less bias towards processing negative words (r = .19, p < .05, one-tailed). Furthermore, N SMP displayed good discriminant validity because neither the sensitivity for positive words nor the bias index for positive words was correlated with N SMP. Yet, both N SMP and C SMP offer rather small effect sizes in (rs less than .20) terms of their predictive power. Moreover, mixed results were obtained if we compare the criterion validity of implicit versus explicit measures of personality. Whereas explicit C correlated more strongly than implicit C with the C behavioural index (z = 2.60, p < .01, two-tailed), implicit N correlated more intense with negative words bias (N behaviour) than explicit N (z = 2.30, p < .05, two-tailed). Reliability Reliability estimates for C SMP and N SMP were obtained by correlating two subsets of SMP data (i.e. correlating the total score of two split-half versions of C SMP, containing eight trials each) and applying the SpearmanBrown prophecy formula. The obtained values indicated low levels of internal consistency, namely .46 for the C SMP and .53 for the N SMP. The relatively low reliability may reect the
Copyright 2012 John Wiley & Sons, Ltd.

small number of trials used. We addressed this limitation in our third study by increasing the number of trials. Although low reliability may have led to underestimated correlations, there was enough systematic variance to detect the expected relationships with explicit measures reported earlier. The incremental value of implicit measures As Perugini et al. (2010, p. 255) put it one of the hallmarks of a good measure is predictive validity. To examine predictive validity, we estimated the parameters of regression analyses predicting relevant behaviours from explicit traits, implicit traits, and their interaction (Table 2, upper panel). This analysis strategy allowed us to assess incremental validity (beyond explicit measures) in two ways. First and most obviously, we can assess the association between implicit measures and behaviours with the variance due to explicit measures removed. A second way that implicit measures might show incremental validity is in the interaction with explicit measures. Two meta-analyses of implicit measures indicate that implicit measures are more strongly predictive of behaviours when implicit and explicit measures are in agreement than when they conict (Cameron, BrownIannuzzi, & Payne, 2012; Greenwald, Poehlman, Uhlmann, & Banaji, 2009). Greenwald et al. (2009) suggested that conicting implicit and explicit scores reect a type of ambivalence that reduces the likelihood that attitudes or traits will be translated into behaviour. In contrast, consistent implicit and explicit associations may reinforce each other, increasing the predictive validity of each. One way to assess incremental validity, then, is to test the interaction between implicit and explicit measures. Both C SMP and N SMP were good predictors. Concerning C-related behaviours, we found an interaction between implicit and explicit measures (Table 2). Simple
Eur. J. Pers. 26: 613628 (2012) DOI: 10.1002/per

618

F. A. Sava et al.

Table 2. Standardized coefcients involving implicit and explicit C and N (Study 1, N = 98) Dependent variable C behaviour N behavioura Explicit C Explicit N Implicit C Implicit N Step 1 Explicit .46* .09 Implicit .31* .16 Explicit .31* .16 Explicit .46* .12 Implicit .21* .11 Explicit .29* .12 Step 2 Implicit .02 .20 Social desirability .45* .28* Social desirability .04 .12 Explicit .46* .13 Implicit .21* .13 Explicit .28* .11 Step 3 Implicit .01 .20 Social desirability .44* .31* Social desirability .03 .12 Interaction .18* .04 Interaction .02 .15 Interaction .06 .06 R1 .220* .009 .096* .025 .096* .025 R2 .000 .040 .216* .079* .002 .014 R3 .036* .002 .001 .023 .009 .003

Note: C, conscientiousness; N, neurotism. a In this case, N behaviour refers specically to the index of negative words bias because this indicator also includes in its formula the sensitivity for negative words. *ps < .05 (two-tailed); ps < .10 (two-tailed).

slope analyses indicated that persons with convergent C (high explicit C and high implicit C) behaved in a more conscientious manner (b = .64, p < .001) than did people with divergent C (high explicit C and low implicit C) (b = .27, p < .05). Consistent with the observation of Greenwald et al. (2009), implicit C provided additional information by reinforcing or opposing explicitly reported traits (Figure 1.) Concerning N-related memory bias, persons with high implicit N displayed fewer biases in recognizing negative words, which remained marginally signicant after controlling for explicit N. Similarly, the relationship between implicit N and N-related memory bias remained signicant even if we controlled for the participants performance on the entire recall performance (b = .20, p < .05, an additive effect of 4.0% of N-related memory bias variance over the one that is explained by the recall performance). Thus, implicit C facilitated prediction of behaviour in interaction with explicit C and implicit N facilitated prediction as an independent main effect. The implicitexplicit dynamic in relationship with social desirability tendencies Another robust and important relationship from a theoretical point of view is presented in the middle of Table 2, which displays regressions predicting explicit measures from their respective implicit measures and social desirability. We found that each of the explicit measures seem to be a mixture of the implicitly measured trait and social desirable tendencies. In contrast, the bottom of Table 2 displays regressions predicting the implicit measures from their explicit counterparts and social desirability. When explicit measures are in the equations, there was no association between implicit

measures and social desirability. No signicant interactions were found. As a further check, we tested whether social desirability was signicantly more associated with explicit than implicit measures. On average, we found signicantly higher correlations between explicit measures and social desirability, than between implicit measures and social desirability (z = 2.10, p < .05, two-tailed test). Discussion The aim of the rst study was to test whether an implicit measure of personality based on semantic misattributions effect can be used as a reliable and valid measure. The data suggest good convergent and discriminant validity for the C SMP and acceptable convergent validity for N SMP. Although both measures had a rather low internal consistency, they demonstrated they could serve as good predictors of various behaviours, in the expected direction. Importantly, the same data suggest that explicit measures of personality represent a composite score of the intended measured trait and social desirability, whereas the SMP scores were less strongly associated with social desirability tendencies. The role of social desirability in the explicit evaluation of a trait is consistent with Fazios MODE model (Fazio, 1990), suggesting that explicit measures of a psychological construct (e.g. questionnaires) reect deliberate self-presentation strategies, in addition to the evaluative associations in memory (e.g. the strength of link between self and a particular descriptor such as shy). We observed incremental validity beyond explicit measures for both C and N measures. For C, implicit and explicit measures interacted when predicting behaviours, suggesting

Figure 1. Interaction plots predicting correspondent behaviour between implicit and explicit measures of personality. C, conscientiousness; N, neuroticism; SMP, semantic misattribution procedure. Eur. J. Pers. 26: 613628 (2012) DOI: 10.1002/per

Copyright 2012 John Wiley & Sons, Ltd.

Semantic misattribution procedure that consistency between implicit and explicit traits may increase the ability to predict behaviour. For N, the implicit measure showed an additive effect when predicting behaviours. Both of these patterns of incremental validity have been demonstrated in studies of implicit attitudes (see Perugini, 2005, for a review). Perugini et al. (2010) has explicated the various possible patterns of interactions between implicit and explicit measures, and theoretical implications of each. However, research on implicit personality measures is still a new enterprise. Current theoretical models are not able to make a priori predictions about exactly when each pattern of ndings should be found. Our goal in this study was to examine whether implicit measures provided information beyond the knowledge gained with explicit measures. Indeed, our results suggest that some information would have been missed if we had examined only explicit measures. An important task for future research will be to develop comprehensive theories that predict when each pattern of relationships (e.g. additive or interactive) will emerge. To reach that point, basic research on the utility of implicit measures as reported here is a necessary rst step. Regarding the main shortcomings of the results presented in the rst study, we underline the low level of internal consistencies for the two SMP measures. Although they are generally higher than those presented in sequential priming (Gawronski, 2009), they are lower than expected if we compare them with a typical internal consistency found in studies using the AMP technique (Payne et al., 2005). Yet, these internal consistency values found for C SMP and N SMP can be seen as acceptable, particularly if we look at the context in which they have been obtained. Unlike other implicit procedures, which require hundreds of trials, the SMP employed only 16 trials with dichotomous responses (Fits me versus Does not t me). Therefore, a value around .50 for internal consistency may be acceptable for an implicit measure containing only 16 dichotomous items per dimension. In Study 3, we examined the effect of increasing the number of trials, nding substantially higher reliability. A second area of concern regarding the results from the rst study involves some issues regarding the discriminant validity of N SMP. Although the N SMP proved discriminant validity in relationship with the explicit self-concept, its association with the explicit inventory-type measures of trait N was either not discriminatory (small but signicant correlation with explicit N and C) or absent (the association becoming insignicant if controlled for social desirability). However, the absence of a signicant correlation between explicit and implicit measures of the same concept is not unusual in the literature (Hofmann et al., 2005), as the implicit and explicit measures of a particular construct may rely on different mechanisms (e.g. propositional for the explicit measures and associational for the implicit measures). In fact, this result replicates the results found by Schmukle et al. (2008), who found signicant correlations between explicit and implicit personality measures only for C and E (extraversion), but not for neuroticism. A second possible interpretation for the divergent results between N SMP and explicit descriptors of N versus explicit
Copyright 2012 John Wiley & Sons, Ltd.

619

inventory-type measures of N could be found in the theoretical differences between the personality self-concept and the personality trait. The latter usually includes the rst one, but it cannot be reduced to it. For instance, a high level of neuroticism could be predicted by a high level of selfconcept descriptors referring to N (i.e. I am helpless), but it can be also predicted from a persons beliefs and his or her way of interpreting the world (Ellis, 1994). In some cases, a persons beliefs reect their personality traits (i.e. it is unacceptable to have a mess in your room) even if they are not self-referential and therefore may not be included in the personality self-concept. In order to overcome some of the aforementioned limitations and to replicate the main results found in Study 1, we conducted a second study, using a similar research strategy.

STUDY 2 The focus of Study 2 was on replicating the results from Study 1 while expanding the evidence for predictive validity by examining both SR and O behaviours. In the case of C SMP, which proved to have good convergent and divergent validity, we introduced more SR behaviours to be predicted, taking into account different areas connected to conscientious behaviour such as health-related behaviours and norm-keeping behaviours. In addition, we introduced an other-reported (OR) behaviour consisting of an index of school attendance for subjects where their attendance was optional. We also paid particular attention to a specic O behaviour that can tap impulsive processes consisting of a laboratory task to measure task persistence. A replication of the results presented in Study 1 for C, in addition to further evidence of the C SMP incremental predictive power, will consolidate the position of C SMP as a valid measure of implicit conscientiousness. In addition, from a theoretical perspective, we were interested to see whether an additive pattern or an interaction pattern between implicit and explicit C is better supported by our data). As for the N SMP measure, due to aforementioned limits concerning its discriminant validity, we introduced an explicit measure of self-esteem as an appropriate indicator of self-concept-related aspects closed to emotional stability (low neuroticism), and we compared its predictive pattern with the one presented for classical explicit inventory-type measures of N.

Method Participants The initial pool of participants consisted of 156 undergraduate students, who voluntarily agreed to participate in this research in exchange for course credit and who signed an informed consent form. Sixteen cases were dropped from the analysis (six due to missing data and 10 due to higher than 10% of trials with RT under 100 milliseconds); therefore, 141 participants (M = 22.6; SD = 4.9, 67% females) were included in the data analysis.
Eur. J. Pers. 26: 613628 (2012) DOI: 10.1002/per

620

F. A. Sava et al. the beginning telling the participants that the task was completed. Almost all participants stopped doing the arithmetic exercises before completing the task (only 4.6% nished it), solving on average 72 additions (SD = 39). Data analysis As in Study 1, we computed several latent variables as global measures for those constructs where more than one manifest variable was included. Thus, we had specic indicators for explicit C and explicit N, and we computed an overall score for all SR and OR C-related behaviours. We also look at O C-related behaviours, using the participants performance at the persistency task. All scores for the latent variable were standardized in z scores; therefore, they have a mean of 0 and an SD of 1. Results As displayed in Table 3, the results replicated the main ndings of Study 1 for C SMP, both in terms of its convergent validity and divergent validity. Thus, the C SMP was positively associated with explicit C but not to explicit N. The C SMP measurement correlated in the expected direction with all types of behaviours regardless of whether they were self-reported, other-reported or recorded in the laboratory. Self-esteem was signicantly associated with less implicit N, but the association with explicit N failed to reach signicance (p = .13, in a two-tailed test). Likewise, as shown in Table 4, the explicit C measure includes a mixture of the implicitly measured trait and socially desirable tendencies. The implicit measure was less dependent on the level of social desirability. Similarly, explicit N was associated with social desirability, whereas implicit N was not. Likewise, we found signicantly higher correlations when associating explicit measures with social desirability, than when associating implicit measures with social desirability (z = 2.42, p < .05, two-tailed test). Regarding the power of C SMP to predict behaviours, we can note that both C SMP and explicit C predict C-related behaviours in the expected direction, suggesting an additive pattern of prediction, although marginally, both for SR behaviours and for objective performance in the persistency task (Figure 2). Finally, reliability estimates for C SMP and N SMP were obtained using the split-half procedure corrected with the SpearmanBrown prophecy formula. The obtained values indicated replicated the results from Study 1, internal consistency values being .55 for C SMP and .57 for N SMP. Discussion The results of Study 2 supported our prediction that the SMP could be a valid indirect measure of personality self-concept. Although we focused more on validating the C SMP, we also obtained evidence of discriminant validity for the N SMP. Each implicit measure was more closely associated with other measures of the same trait than other traits. Furthermore, the implicit and explicit C measures were found to additively predict various SR and O behaviours.
Eur. J. Pers. 26: 613628 (2012) DOI: 10.1002/per

Materials The SMP task used to measure implicit levels of conscientiousness and neuroticism via Direct RT v 2008 (Empirisoft, 2008) was the same as the one described in Study 1. Neuroticism and Conscientiousness Self-Report Measures were collected using the 50-item IPIP (Goldberg et al., 2006) and the DECAS Personality Inventory (Sava, 2008), two of the explicit instruments used in the rst study. The same MarloweCrowne 33-item scale was used for measuring social desirability tendencies among participants (Crowne & Marlowe, 1960). The well-known 10-item Rosenberg Self-Esteem Scale (Rosenberg, 1965) was used for assessing the explicit selfesteem (explicit SE). Higher scores are indicative for a higher level of explicit SE. Self-reported and other-reported behaviours. A set of six particular behaviours was included in a Behavioral Report Form. This referred to behaviours related to health (smoking, drinking, and the body weight index) and to norm-keeping behaviours (church attendance, involvement in voluntary activities, and absence of indelity in the last two years). Two latent scores were computed on the basis of factor analysis procedure (all behaviours having a loading of at least .30 on their intended factor). Higher scores reect a persons orientation towards more conscientious behaviours (i.e. healthy behaviours and norm-keeping). In addition, we asked for their permission to collect data from the classes they had in that particular semesters, where the class policy allowed for noncompulsory attendance. Attendance data were provided by the instructors. Higher scores indicate better class attendance. Task persistence was purposefully designed to measure an O behaviour. It was implemented on computer using the Media Lab software (Empirisoft, 2008). The task consisted in a set of 140 simple three digits arithmetical additions (e.g. 245 + 132) that were presented one at a time, while participants were asked to do all computations mentally, without any additional support such as paper and pencil. Additional details are presented in the next section. Procedure In its major parts, the procedure used for this study was similar to the one presented in the rst study. First, participants completed the SMP task individually on computer, and then they lled in the ve explicit questionnaires (IPIP, DECAS, MarloweCrowne social desirability, Rosenberg SE, and the Behavioral Report Form), which were presented in a counterbalanced order. Their last task in the laboratory session was the persistence task, which was completed in front of a computer, which recorded the time spent on the task (time), the number of additions performed (speed), and the number of correct responses offered (correct answers). In the instruction for this, it was explicitly specied that this was the last task in the session, that there is no time limit for this task, and that they can work as much as they want being free to leave at any time, without completing the entire task and without any penalties. However, we programmed the computer to generate a message after 20 minutes from
Copyright 2012 John Wiley & Sons, Ltd.

Semantic misattribution procedure


Table 3. A matrix of inter-correlations among studys variables (Study 2, N = 140) Variables Implicit personality self-concept Conscientiousness SMP C (1) Neuroticism SMP N (2) Explicit personality traits Latent (explicit) C (3) Latent (explicit) N (4) Other relevant variables Social desirability (5) Explicit self-esteem (6) Behaviours C latent behavioural index (7) O task persistence (8) Mean Standard deviation 1 .55 .35 .39 .05 .23 .15 .41 .22 9.1 2.5 2 3 4 5 6 7

621

.57 .15 .13 .15 .23 .14 .06 8.9 2.4 .14 .45 .27 .22 .23 0.0 1.0 .39 .53 .00 .11 0.0 1.0 .74 .39 .05 .16 17.9 4.9 .81 .13 .03 32.4 4.4 .23 0.0 1.0 13.5 5.5

Note: O, objective; SMP, semantic misattribution procedure. Bold type for correlations signicant at ps < .05 (two-tailed test) (|r| .17, p < .05, two-tailed tests; |r| .15, p < .05, one-tailed tests), and italic type for estimated internal consistencies.

Table 4. Standardized coefcients involving implicit and explicit C and N (Study 2, N = 140) Step 1 Criterion C behaviour Tasks persist Explicit C Explicit N Explicit SE Implicit C Implicit N Explicit SE Explicit .22* .23* Implicit .39* .12 .22* Explicit .39* .12 .22* Explicit .07 .17 Implicit .31* .05 .16* Explicit .37* .06 .19* Step 2 Implicit .41* .17 Social desirability .38* .38* .37* Social desirability .05 .13 .07 Explicit .06 .16 Implicit .30* .05 .16* Explicit .37* .06 .19* Step 3 Implicit .43* .17 Social desirability .39* .38* .37* Social desirability .05 .13 .08 Interaction .13 .05 Interaction .05 .06 .03 Interaction .03 .08 .09 R1 .046* .053* .151* .011 .047* .151* .011 .047* R2 .134* .021 .140* .142* .133* .001 .024 .005 R3 .018 .003 .002 .004 .001 .001 .031 .008

Note: C, conscientiousness; N, neurotism; SE, self-esteem. *ps < .05 (two-tailed); ps < .10 (two-tailed).

Figure 2. Interaction plots predicting C-related behaviour between implicit and explicit measures of C. C, conscientiousness; SMP, semantic misattribution procedure.

This additive pattern is consistent with previous ndings (e.g. Asendorpf et al., 2002; Schnabel et al., 2006), suggesting that implicit measures can have an incremental predictive validity over explicit measures. Concerning social desirability, the data converged with the results from Study 1. Results suggest that explicit measures, unlike the implicit ones, represent a mixture of the measured trait and social desirability. Study 3 built on these results in four ways. First, we generalized the SMP to a third personality trait (extraversion). Second, we examined whether an increase in the number of trials for the SMP task would lead to an increase in internal
Copyright 2012 John Wiley & Sons, Ltd.

consistency as expected from classical test theory. Third, we compared the SMP with the more commonly used IAT. Finally, we examined the testretest stability of the SMP over six months.

STUDY 3 In this study, we introduced a new personality self-concept measure based on SMP approach to assess the level of implicit extraversion (E SMP). To check for its construct
Eur. J. Pers. 26: 613628 (2012) DOI: 10.1002/per

622

F. A. Sava et al. adjectives for the E trait. However, for the C trait, we used the adjectives used by Steffens and Schulze-Knig (2006). Extraversion and Conscientiousness Self-Report Measures were collected using the 50-item IPIP (Goldberg et al., 2006), the NEO-FFI (Costa & McCrae, 1992), and the DECAS Personality Inventory (Sava, 2008), the same instruments as presented in Study 1. The uncorrected mean between E DECAS and E scales from other Big-Five instruments was .67. The same MarloweCrowne 33-item scale was used for measuring social desirability tendencies among participants (Crowne & Marlowe, 1960). Self-reported behaviour measures. A set of four particular SR behaviours were included in a Behavioral Report Form. They referred to both C-related behaviours (smoking, drinking, and church attendance) and E-related behaviours (attendance at parties). Task-related objective behaviour. We recorded the time needed by each participant to complete the set of questionnaires because extraverts are thought to have a faster pace of work than introverts (Eysenck, 1967). The mirror task. A complementary behavioural measure related to both E and C traits was the mirror task, which was inspired by Gormly (1984). In this task, the participants had to write down a short message Michelangelo. The difcult part was that they were not allowed to see the message directly or to look straight at the paper on which they were supposed to write. Therefore, the mirror was the only way they could monitor their performance while writing the message. Furthermore, they had to write down the message so that it will appear as correctly written in the mirror (e.g. starting with the last letter written upside down). The task ended when participants successfully managed to correctly write down the message or, in a few cases, if they give up this task. We recorded two indicators based on this task. One of them was theoretically related to C: the number of errors made (Steffens & Schulze-Knig, 2006), expecting that low C people will act more impulsively and will make more errors and therefore will need more time to nish the task. The other indicator was related to E, consisting of self-reported likeability of the task. On the basis of previous research, we expected that high E individuals will nd this novel task more pleasant (de Jonge & Slaets, 2005; Paunonen, 2003). Procedure The procedure used for this study was similar to the one presented in the other studies. Firstly, participants completed the SMP task individually on computer, and then they lled out explicit questionnaires (IPIP, NEO-FFI, DECAS, MarloweCrowne social desirability, and the Behavioral Report Form), which were presented in a counterbalanced order. Their last task of the session was the mirror task, which was completed individually, in a separated room. Data analysis As in previous studies, we computed several latent variables as global measures for those constructs where more than one manifest variable was included. Thus, we used latent indicators for explicit C and explicit N, as well as for C-related
Eur. J. Pers. 26: 613628 (2012) DOI: 10.1002/per

validity, we also measured the level of explicit extraversion, and we introduced new SR and O behaviours to see how well E SMP can (incrementally) explain the variance found in these behaviours (Back et al., 2009; Paunonen, 2003). Furthermore, to assess divergent validity for the E SMP, we compared its pattern of correlation with another implicit measure C SMP, which consistently demonstrated good convergent and divergent validity. In addition, unlike studies one and two, where the SMP task was based on recording the participants responses at 16 trials per dimension, in the third study, we doubled the number of trials to increase internal consistency. Similarly, to look for more indicators of reliability than internal consistency, we assessed 6-month testretest stability for the C SMP and E SMP measures. For this, we took the advantage of using a small proportion of the participants used in this study in a subsequent unrelated study, at a 6-month time distance between the two studies. Finally, in order to gather more data about the construct validity of the two SMP measures, we also included a different but more commonly used kind of implicit measures, namely the IAT (Greenwald, McGhee & Schwartz, 1998). The IAT has been validated as an implicit measure of personality (Schmukle et al., 2008; Steffens & Schulze-Knig, 2006). We created IAT measures of C and E.

Method Participants A total of 145 psychology undergraduates participated in a study on innovative ways to assess personality for extra course credit. One hundred and thirty-ve of the participants (67% females; age M = 20.2, age SD = 2.1) met the eligibility criteria to be included in data analysis (less than 10% of trials having RT under 100 milliseconds). Data from 37 participants were also collected six months later to look for testretest stability of the SMP measures. Materials The SMP task used to measure implicit levels of conscientiousness and extraversion via Direct RT v 2008 (Empirisoft, 2008) was the same as the one described in Study 1. The only difference was that participants completed 32 trials per dimension instead of 16 trials. We used the same adjectives presented in the Appendix, each descriptor being used two times as primes in this longer version of SMP task. The IAT (Greenwald et al., 1998) is a popular implicit method used for assessing the strength of semantic associations between a target concept (the self, in our case) and various relevant adjectives (in our case for E or C). In a typical IAT, participants have to categorize relevant words into categories related to self or non-self, and high or low trait categories. The assumption behind IAT is that (for example) highly extraverted individuals will sort faster when self and high extraversion categories share the same button, than when different buttons are assigned to these categories. We employed a similar methodology for the IAT task to the one used by Schmukle et al. (2008), using the same relevant
Copyright 2012 John Wiley & Sons, Ltd.

Semantic misattribution procedure behaviours and for E-related behaviours. All items included in a specic latent variable loaded at least .30 in their specic latent variable. All scores for the latent variable were z standardized; therefore, they have a mean of 0 and an SD of 1 (Table 5 and 6). Because of highly positively skewed distributions for both C SMP and E SMP (e.g. ratios above 3 when dividing skewness to its standard error), we transformed the data using natural logarithm function to obtain a normal distribution for both implicit measures.

623

explicit measures of C z = 2.09, p < .05, two-tailed test. Similarly, E SMP partially demonstrated construct validity by correlating with explicit E and by predicting E-related behaviours, but not C-related behaviour. However, E SMP failed to show good discriminant validity because it also correlated signicantly both with explicit C and E. These correlations were not signicantly different, z = 0.20, p > .05. Associations with behaviour Both SMP measures were found to be signicant predictors for behaviours, strengthening their predictive validity. Thus, C SMP, but not explicit C, managed to predict the number of errors in the mirror task (r = .21, p < .05, two-tailed test, and r = .00, p > .05, respectively). The E SMP, but not explicit E, predicted the time needed by participants to complete the set of questionnaires they received (r = .22, p < .05, two-tailed test, and r = .14, p = .12, respectively) (Figure 3). Interesting results were found with regard to the relationship among implicit and explicit measures of personality and social desirability. The same pattern of relationship was found

Results Implicitexplicit associations The results concerning the C SMP robustly replicate the pattern of correlation found in the rst two studies. Thus, C SMP correlated moderately with explicit C and managed to predict C-related behaviours, but not E-related behaviours. Moreover, although implicit C also correlated with explicit E, the magnitude of this correlation was signicantly lower than the magnitude of correlation between implicit and

Table 5. A matrix of inter-correlations among study variables (Study 3, N = 135) Variables Implicit personality self-concept Conscientiousness SMP C (1) Extraversion SMP E (2) Conscientiousness IAT C (3) Extraversion IAT E (4) Explicit personality traits Latent (explicit) C (5) Latent (explicit) E (6) Other relevant variables Social desirability (7) Behaviours C latent behavioural index (8) E latent behavioural index (9) Mean Standard deviation 1 .61 .44 .11 .06 .41 .20 .15 .25 .13 2.9 0.2 2 3 4 5 6 7 8 9

.73 .03 .16 .32 .34 .14 .03 .21 2.9 0.3

.78 .24 .02 .07 .12 .02 .03 .39 .33

.81 .06 .06 .04 .13 .04 .25 .38 .21 .47 .21 .00 0.0 1.0 .14 .04 .36 0.0 1.0 .74 .15 .06 17.5 4.6 .23 0.0 1.0 0.0 1.0

Note: SMP, semantic misattribution procedure. Bold type for correlations signicant at ps < .05 (two-tailed test) (|r| .17, p < .05, two-tailed tests; |r| .15, p < .05, one-tailed tests), and italic type for estimated internal consistencies. None of these correlations has changed their status (signicant versus not signicant) if raw data for SMP C and SMP E were used instead of ln SMP C and ln SMP E. These transformed scores were used to correct for the positive skewed distributions of the initial raw values.

Table 6. Standardized coefcients involving implicit and explicit C and N (Study 3, N = 135) Step 1 Criterion C behaviour E behaviour Explicit C Explicit E Implicit C Implicit E Explicit .20* .36* Implicit .42* .34* Explicit .42* .34* Explicit .12 .33* Implicit .35* .32* Explicit .44* .32* Step 2 Implicit .20* .10 Social desirability .41* .09 Social desirability .05 .10 Explicit .12 .28* Implicit .37* .32* Explicit .42* .33* Step 3 Implicit .20* .11 Social desirability .39* .10 Social desirability .03 .09 Interaction .01 .16* Interaction .12 .04 Interaction .09 .07 R1 .043* .131* .173* .114* .173* .114* R2 .034* .009 .166* .009 .002 .010 R3 .000 .033* .012 .001 .008 .005

Note: C, conscientiousness; E, extraversion. *ps < .05 (two-tailed); ps < .10 (two-tailed). Copyright 2012 John Wiley & Sons, Ltd. Eur. J. Pers. 26: 613628 (2012) DOI: 10.1002/per

624

F. A. Sava et al.

Figure 3. Interaction plots predicting correspondent behaviour between implicit and explicit measures of personality. C, conscientiousness; E, extraversion; SMP, semantic misattribution procedure.

for C SMP, suggesting that explicit C is a mixture of the intended trait and social desirability. Likewise, a signicantly higher correlation was found when associating explicit C with social desirability, then when associating implicit SMP C with social desirability (z = 3.81, p < .01, two-tailed test). The pattern was different for extraversion. Neither E SMP, nor explicit E was associated with social desirability level. Reliability Internal consistency estimates for C SMP and E SMP were .61 and .73, respectively, by using the split-half procedure corrected with the SpearmanBrown prophecy formula. The increase in the number of trials had a positive impact on the SMP reliability. Reliability estimates were slightly higher for the IATs (.78 for C and .81 for E). Comparison with IAT Despite good internal consistency, we found little evidence of predictive validity for the IAT measures. IAT measures were not associated with SMP measures or explicit measures of the same trait. Moreover, IATs were not associated with any of the behavioural measures. The only signicant correlation for the IAT measures was between the two IATs themselves. Because they were intended to measure different traits, this likely reects shared method variance. Of course, these null effects should be interpreted with caution because there are a variety of reasons a study might fail to detect true effects in the population. A direct test of correlational differences obtained in this study between the average criterion validity of SMP (r = .23) and the average criterion validity of IAT (r = .03) reveals a marginally signicant result, favouring the SMP ability to predict behaviours z = 1.69, p < .10, two-tailed test. Temporal stability The SMP demonstrated good testretest stability. Thus, 37 persons of the participants included in this study had the opportunity to be part of a convenience sample in a subsequent but unrelated study. We found high testretest correlations for both C SMP (r = .50, p < .01) and E SMP (r = .75, p < .01). These values were slightly lower than the one found for the explicit measures (DECAS Personality Inventory), where we obtained a 6-month testrest correlation of .72 for explicit C and of .86 for explicit E. In this latter sample, we also replicated the good internal consistency values found in Study 3 by using the larger number of trials (.67 for C SMP and of .80 for E SMP).
Copyright 2012 John Wiley & Sons, Ltd.

Discussion In the third study, we replicated the good psychometric properties of the C SMP, and we introduced a measurement of E SMP, which demonstrated reasonably good validity and excellent reliability for an implicit measure. Both C SMP and E SMP managed to predict behaviours in a discriminative manner, in the expected direction. Moreover, both performed well in comparison to the IAT, which is the most widely used implicit measure of personality. Data from this correlational study suggest that social desirability may create a bias in explicit personality measures. In fact, such a result is also supported by a meta-analysis carried out by Birkeland, Manson, Kisamore, Brannick, and Smith (2006), who reported that job applicants score higher than non-applicants in four out the ve traits included in the Big-Five model (namely openness, extraversion, conscientiousness, and emotional stability). The difference between the two contexts is not related to self-deception but to impression management. Our studies suggest that implicit measures of personality such as the SMP can help avoid social desirability bias. Yet, one important limitation of Study 3 with regard to the issue of social desirability is the absence of an experimental design in which participants are asked to fake their responses both to explicit and implicit measures of personality. Future research could examine this issue experimentally. Finally, a possible explanation for the lack of convergent validity between IAT and SMP is that they rely on different processes. For instance, De Houwer and Moors (2010) distinguish between implicit measures that are symbolic (i.e. based on the meaning of responses) versus non-symbolic (i.e. based on speed or some other aspect of responses rather than the meaning of the response). The SMP is a symbolic measure, whereas the IAT is non-symbolic (the estimate is derived from the speed rather than the meaning of the responses; De Houwer and Moors, 2010). Such an interpretation is consistent with the pattern of correlation between implicit and explicit measures because only SMP scores correlated with explicit (symbolic) measures. However, a more denitive conclusion about the differences between the SMP and IAT must await further research. GENERAL DISCUSSION Summary The aims of this paper were to introduce and to validate a new implicit method for assessing personality self-concept,
Eur. J. Pers. 26: 613628 (2012) DOI: 10.1002/per

Semantic misattribution procedure derived from an AMP (Payne et al., 2005), but also extending these effects to semantic misattributions (Deutsch & Gawronski, 2009). Unlike other previous studies in the area of implicit personality self-concept, which either used IAT (Greenwald & Banaji, 1995) or used derived-IAT measures from the same family (EAST, De Houwer, 2003; GNAT, Nosek & Banaji, 2001; IAP, Schnabel et al., 2006), the SMP task is based on misattributing the activated semantic content from the prime to an ambiguous target. Data supported the validity of the SMP, and when the number of trials was 32 per task, we found good reliability, both in terms of internal consistency and 6-month testretest stability. Theoretical implications From a theoretical point of view, this paper is consistent with dual process models of implicit attitudes (i.e. MODE, Olson & Fazio, 2009 and APE, Gawronski & Bodenhausen, 2006), although the present data do not distinguish between models. Thus, looking at the relationship between social desirability and participants responses to a personality questionnaire, we found that explicit measures of more socially desirable traits such as emotional stability and conscientiousness were associated with social desirability, whereas the implicit measures were not. From this perspective, the scores of an implicit personality measure can be seen as a result of automatic activation of implicit self-concept, whereas the scores found in an explicit personality questionnaire tend to reect both automatically activated traits and adjusting for social desirability in a deliberative process. Moreover, looking at the way all three SMP tasks managed to predict both SR and O behaviours, as well as both long-term and momentarily behaviours, we found support in favour of dual process models. These ndings suggest that there is a qualitative distinction between implicit and explicit personality self-concepts (i.e. a person may have a high level of explicit C but a low level of implicit C). Thus, in each of the three studies, implicit measures of personality self-concept were able to predict less deliberative (spontaneous) behaviours. For instance, people high in implicit N were more attentive towards negative words stimuli and therefore committed fewer false positive errors even if controlled for their recall performance. Similarly, people with a high level of implicit C marginally predicted the participants performance in a laboratory task to measure task persistence. Likewise, people with high implicit C committed fewer errors in the mirror task, thus showing less impulsivity in writing the message. Nonetheless, people with a high level of implicit E required less time needed to ll in the questionnaires than those with low implicit E. Explicit measures were signicant predictors in only one of the aforementioned tasks: in the persistence task, people with high explicit C demonstrated a more persistent behaviour. The pattern of relationships for the implicit and explicit personality self-concept measures with SR behaviours found heterogeneous results. In the rst two studies, we found an interaction and an additive effect, respectively, as concerning the ability of both implicit and explicit Cs to predict C-related behaviours. In Study 3, we found an overlap case
Copyright 2012 John Wiley & Sons, Ltd.

625

so that if introduced together as predictors, only the implicit C predicted the overall behavioural index. Likewise, only explicit E predicted SR E-related behaviours, if both implicit and explicit Es are included in the same regression equation. Such a mixture of results is not uncommon for studies focusing on implicit personality self-concept measures. Whereas some studies suggest that a dual process model is more appropriate to describe the relationship between implicit and explicit measures of personality (Asendorpf et al., 2002, for shyness; Back et al., 2009 for neuroticism and extraversion), others found divergent results in the direction that only explicit measures were able to predict behaviours (Back et al., 2009, for openness and conscientiousness; Schnabel et al., 2006, for angriness). The question of whether implicit and explicit measures should predict behaviours in an additive way, an interactive way, or should each predict different behaviours is a subject of ongoing debate, and no consensus has yet emerged (Perugini, 2005; Perugini et al., 2010). Across studies, we found all three patterns. Although future theoretical development is needed to identify when each pattern is most likely, our results clearly demonstrate that information would have been lost if we had used only explicit measures of personality. In this respect, this study adds signicantly to the evidence found by Back et al. (2009), in favour of the utility of using implicit measurements of personality in addition to explicit ones. Overall, the most important contribution of this paper is that the SMP task provides initial evidence for convergent validity (having consistent and signicant correlations with the same-trait explicit measures, particularly for C). The study also establishes preliminary evidence for criteria validity, as the SMP predicted both SR behaviours and O situational behaviours in the expected direction. Furthermore, if we take into consideration the reliability data presented in Study 3 for both the initial and follow-up samples, we can consider the SMP to have similar psychometric properties to the IAT task (Back et al., 2009) and even higher testretest correlation for a 6-month follow-up of .50 for C SMP and .75 for E SMP, as compared with the average value of .51 for the IAT task (Hofmann et al., 2005). Limitations and future directions A limitation of this study was that we did not specically control for the valence of traits in this study (Schnabel & Asendorpf, 2010). However, we replicated for the three traits that we measured the results obtained by Back et al. (2009), nding negative correlation between a negative valence trait (N) and a positive valence trait (C), and positive correlation between the two measured positive traits E and C. In this sense, it may be the case that the misattribution effect occurred not only because of a semantic processing of the stimuli but also because of an affective misattribution effect determined by the difference in social desirability of various descriptors used as primes. In fact, a mixed effect of both semantic and affective misattribution might be involved because we found signicant correlations between the implicit traits (average value of .33) that can be explained
Eur. J. Pers. 26: 613628 (2012) DOI: 10.1002/per

626

F. A. Sava et al. REFERENCES


Asendorpf, J. B., Banse, R., & Mcke, D. (2002). Double dissociation between implicit and explicit personality self-concept: The case of shy behavior. Journal of Personality and Social Psychology, 83, 380393. DOI: 10.1037//00223514.83.2.380 Back, M. D., Schmukle, S. C., & Egloff, B. (2009). Predicting actual behavior from the explicit and implicit self-concept of personality. Journal of Personality and Social Psychology, 97, 533548. DOI: 10.1037/a0016229 Bar-Haim, Y., Lamy, D., Pergamin, L., Bakermans-Kranenburg, M. J., & van IJzendoorn, M. H. (2007). Threat-related attentional bias in anxious and nonanxious individuals: A meta-analytic study. Psychological Bulletin, 133, 124. DOI: 10.1037/ 00332909.133.1.1 Birkeland, S. A., Manson, T. M., Kisamore, J. L., Brannick, M. T., & Smith, M. A. (2006). A meta-analytic investigation of job applicant faking on personality measures. International Journal of Selection and Assessment, 14(4), 317335. Bogg, T., & Brent, W. R. (2004). Conscientiousness and healthrelated behaviors: A meta-analysis of the leading behavioral contributors to mortality. Psychological Bulletin, 130, 887919. DOI: 10.1037/00332909.130.6.887 Boldero, J. M., Rawlings, D., & Haslam, N. (2007). Convergence between GNAT-assessed implicit and explicit personality. European Journal of Personality, 21, 341358. DOI: 10.1002/ per.622 Cameron, C. D., Brown-Iannuzzi, J. L., & Payne, B. K. (2012). Sequential priming measures of implicit social cognition: A meta-analysis of associations with behavior and explicit attitudes. Personality and Social Psychology Review. DOI: 10.1177/ 1088868312440047 Caprara, G. V., Barbaranelli, C., & Borgogni, L. (2008). Manualul de utilizare a BFQ-2 [The users handbook for Big Five Questionnaire 2]. Cluj-Napoca, Romania: Sinapsis. Costa, P. T., & McCrae, R. R. (1992/2008). Manual tehnic. Inventarul de personalitate NEO FFI [The technical manual of NEO FFI Personality Inventory]. Cluj-Napoca, Romania: Sinapsis. Crowne, D. P., & Marlowe, D. (1960). A new scale of social desirability independent of psychopathology. Journal of Consulting Psychology, 24(4), 349354. Crowne, D. P., & Marlowe, D. (1964) The approval motive. New York: Wiley. De Houwer, J. (2003). The Extrinsic Affective Simon Task. Experimental Psychology, 50(2), 7785. DOI: 10.1027//1618 3169.50.2.77 De Houwer, J., & Moors, A. (2010). Implicit measures: Similarities and differences. In B. Gawronski, & B. K. Payne (Eds.), Handbook of implicit social cognition: Measurement, theory, and applications. New York, NY: Guilford Press. De Houwer, J., Teige-Mocigemba, S., Spruyt, A., & Moors, A. (2009). Implicit measures: A normative analysis and review. Psychological Bulletin, 135, 347368. DOI: 10.1037/a0014211 Deutsch, R., & Gawronski, B. (2009). When the method makes a difference: Antagonistic effects on automatic evaluations as a function of task characteristics of the measure. Journal of Experimental Social Psychology, 45, 101114. DOI: 10.1016/j.jesp.2008.09.001 Edwards, A. L. (1957). The social desirability variable in personality assessment and research. New York: Dryden. Egloff, B., Schwerdtfeger, A., & Schmukle, S. C. (2005). Temporal stability of the Implicit Association TestAnxiety. Journal of Personality Assessment, 84, 8288. DOI: 10.1207/ s15327752jpa8401_14 Ellis, A. (1994). Reason and emotion in psychotherapy: Comprehensive method of treating human disturbances: Revised and updated. New York, NY: Citadel Press. Empirisoft Company. (2008). DirectRT and MediaLab software, v 2008. Retrieved from: www.empirisoft.com Eysenck, H. J. (1967). The biological basis of personality. Brunswick, NJ: Charles C. Thomas Publisher.
Eur. J. Pers. 26: 613628 (2012) DOI: 10.1002/per

by the joint effect of both a common method bias and a similar valence bias effect. Yet, we have solid evidence for the presence of a semantic misattribution effect provided both by divergent validity of SMP measures in relationship with specic self-concept descriptors and by divergent validity of SMP measures in predicting specic behaviours that have been previously related to particular traits. In this direction, future studies employing a mathematical model (i.e. Payne, Hall, Cameron, & Bishara, 2010) to identify and quantify the processes that account for the results on the SMP measures will be very helpful for dissociating between various sources that inuence a participants response to the SMP task. Another limitation consists in the absence of an experimental design, to demonstrate clearly whether the SMP measures are more resistant than explicit measures to faking attempts. Because the AMP and SMP ask participants to explicitly rate the target stimuli, it is possible that some respondents fail to follow the instruction and to rate self-relevant adjectives instead of pictographs (Nosek, Hawkins, & Frazier, 2011). In such cases, the assessment will no longer be an indirect one but an explicit evaluation. Yet, the average correlation between various SMP measures and their related explicit measures was .23, a value that is rather similar with the one found in a meta-analysis that found an average correlation of .21 between the explicit and IAT implicit measures of selfconcept (Hofmann et al., 2005). If participants failed to follow the instruction and made direct evaluation, we would expect higher correlations between the SMP results and the explicit personality measures, not only due to the direct evaluation but also due to common method feature (for instance, both DECAS Personality Inventory and SMP tasks used a Yes/No answer format).

Conclusion In sum, the present study suggests that the SMP is an effective alternative for measuring implicit personality selfconcept. While possessing similar construct and reliability features as the IAT, it has better criterion validity, managing to predict both SR and O behaviours. Likewise, the SMP has better psychometric characteristics than other currently used implicit measures of personality. Future research should further examine divergent validity and address effects of common method variance that leads to moderate correlations between scales that are unsound from a theoretical perspective before the test is ready for usage in an applied context such as personnel selection.

ACKNOWLEDGEMENT This research was supported by CNCSISUEFISCSU: project number PNII-IDEI 1076/2009 and project number PN-II-RU-TE-2011-3-0230. This organization had no role in the design and implementation of the study.
Copyright 2012 John Wiley & Sons, Ltd.

Semantic misattribution procedure


Fazio, R. H. (1990). Multiple processes by which attitudes guide behaviour: The MODE model as an integrative framework. In M. P. Zanna (Ed.), Advances in experimental social psychology (Vol. 23, pp. 75109). San Diego, CA: Academic Press. Fazio, R. H., Sanbonmatsu, D. M., Powell, M. C., & Kardes, F. R. (1986). On the automatic activation of attitudes. Journal of Personality and Social Psychology, 50, 229238. Furnham, A. F. (1997). Knowing and faking ones Five-Factor personality score. Journal of Personality Assessment, 69(1), 229243. Gawronski, B. (2009). Ten frequently asked questions about implicit measures and their frequently supposed, but not entirely correct answers. Canadian Psychology, 50(3), 141150. DOI: 10.1037/a0013848 Gawronski, B., & Bodenhausen, G. V. (2006). Associative and propositional processes in evaluation: An integrative review of implicit and explicit attitude change. Psychological Bulletin, 132, 692731. DOI: 10.1037/00332909.132.5.692 Gawronski, B., & Payne, B. K. (Eds.) (2010). Handbook of implicit social cognition. New York: Guilford. Gawronski, B., Cunningham, W. A., LeBel, E. P., & Deutsch, R. (2010). Attentional inuences on affective priming: Does categorisation inuence spontaneous evaluations of multiply categorisable objects? Cognition and Emotion, 24, 10081025. DOI: 10.1080/02699930903112712 Goldberg, L. R., Johnson, J. A., Eber, H. W., Hogan, R., Ashton, M. C., Cloninger, R., Gough, H. G. (2006). The international personality item pool and the future of public-domain personality measures. Journal of Research in Personality, 40, 8496. DOI: 10.1016/j.jrp.2005.08.007 Gormly, J. (1984). Correspondence between personality trait ratings and behavioral events. Journal of Personality, 52(3), 220232. Gosling, S. D., John, O. P., Craik, K. H., & Robins, R. W. (1998). De people know how they behave? Self-reported act frequencies compared with on-line codings by observers. Journal of Personality and Social Psychology, 74, 13371349. Greenwald, A. G., & Banaji, M. R. (1995). Implicit social cognition: Attitudes, self-esteem, and stereotypes. Psychological Review, 102, 427. Greenwald, A. G., McGhee, D. E., & Schwartz, J. K. L. (1998). Measuring individual differences in implicit cognition: The Implicit Association Test. Journal of Personality and Social Psychology, 74, 14641480. Greenwald, A. G., Poehlman, T. A., Uhlmann, E. L., & Banaji, M. R. (2009). Understanding and using the implicit association test III. Meta-analysis of predictive validity. Journal of Personality and Social Psychology, 97, 1741. DOI: 10.1037/a0015575 Hofmann, W., Gawronski, B., Gschwendner, T., Le, H., & Schmitt, M. (2005). A meta-analysis on the correlation between the Implicit Association Test and explicit self-report measure. Personality and Social Psychology Bulletin, 31, 13691385. DOI: 10.1177/0146167205275613 Hofstee, W. K. B, de Raad, B., & Goldberg, L. R. (1992). Integration of the Big-Five and circumplex approaches to trait structure. Journal of Personality and Social Psychology, 63(1), 146163. Imhoff, R., Schmidt, A. F., Bernhardt, J., Dierksmeier, A., & Banse, R. (2011). An inkblot for sexual preference: A semantic variant of the Affect Misattribution Procedure. Cognition and Emotion, 25(4), 676690. DOI: 10.1080/02699931.2010.508260 de Jonge, P., & Slaets, J. P. J. (2005). Response sets in self-report data and their associations with personality traits. The European Journal of Psychiatry, 19(4), 209214. MacMillan, N. A., & Creelman, C. D. (2005). Detection theory: A users guide. Mahwah, N.J.: Lawrence Erlbaum. McDaniel, M. J., Beier, M. E., Perkins, A. W., Goggin, S., & Frankel, B. (2009). An assessment of the fakeability of self-report and implicit personality measures. Journal of Research in Personality, 43, 682685. DOI: 10.1016/j.jrp.2009.01.011 Nosek, B. A., & Banaji, M. R. (2001). The Go/No-Go association task. Social Cognition, 19(6), 625664. DOI: 10.1521/ soco.19.6.625.20886
Copyright 2012 John Wiley & Sons, Ltd.

627

Nosek, B. A., Hawkins, C. B., & Frazier, R. S. (2011). Implicit social cognition: From measures to mechanisms. Trends in Cognitive Sciences, 15(4), 152159. DOI: 10.1016/j.tics.2011.01.005 Olson, M. A., & Fazio, R. H. (2009). Implicit and explicit measures of attitudes: The perspective of the MODE model. In R. E. Petty, R. H. Fazio, & P. Briol (Eds.), Attitudes. Insights from the new implicit measures (pp. 1963). New York: Psychology Press. Paulhus, D. L. (1984). Two-component models of socially desirable responding. Journal of Personality and Social Psychology, 46, 598609. DOI: 10.1037/0022-3514.46.3.598 Paulhus, D. L., & Reid, D. B. (1991). Enhancement and denial in socially desirable responding. Journal of Personality and Social Psychology, 60(2), 307317. Paunonen, S. V. (2003). Big Five factors of personality and replicated predictions of behavior. Journal of Personality and Social Psychology, 84(2), 411424. DOI: 10.1037/0022-3514. 84.2.411 Payne, B. K., Burkley, M., & Stokes, M. B. (2008). Why do implicit and explicit attitude tests diverge? The role of structural t. Journal of Personality and Social Psychology, 94, 1631. Payne, B. K., Cheng, C. M., Govorun, O., & Steward, B. (2005). An inkblot for attitudes: Affect misattribution as implicit measurement. Journal of Personality and Social Psychology, 89 (3), 277293. DOI: 10.1037/00223514.89.3.277 Payne, B. K., Hall, D. L., Cameron, C. D., & Bishara, A. J. (2010). A process model of affect misattribution. Personality and Social Psychology Bulletin, 36(10), 13971408. DOI: 10.1177/ 0146167210383440 Peabody, D., & Goldberg, L. R. (1989). Some determinants of factor structures from personality-trait descriptors. Journal of Personality and Social Psychology, 57, 552567. Perugini, M. (2005). Predictive models of implicit and explicit attitudes. British Journal of Social Psychology, 44, 2945. Perugini, M., & Banse, R. (2007). Editorial. Personality, implicit self-concept and automaticity. European Journal of Personality, 21, 257261. DOI: 10.1002/per.637 Perugini, M., & Leone, L. (2009). Implicit self-concept and moral action. Journal of Research in Personality, 43, 747754. DOI: 10.1016/j.jrp.2009.03.015 Perugini, M., Gallucci, M., & Livi, S. (2000). Looking for a simple Big Five factorial structure in the domain of adjectives. European Journal of Psychological Assessment, 16(2), 8797. DOI: 10.1027//10155759.16.2.87 Perugini, M., Richetin, J., & Zogmaister, C. (2010). Prediction of behavior. In B. Gawronski, & B. K. Payne (Eds.), Handbook of implicit social cognition: Measurement, theory, and applications (pp. 255277). New York: Guilford Press. Rosenberg, M. (1965). Society and the adolescent self-image. Princeton, NJ: Princeton University Press. Rothstein, M. G., & Gofn, R. D. (2006). The use of personality measures in personnel selection: What does current research support? Human Resource Management Review, 16, 155180. DOI: 10.1016/j.hrmr.2006.03.004 Saucier, G. (1994). Mini-markers: A brief version of Goldbergs unipolar Big-Five markers. Journal of Personality Assessment, 63(3), 506516. Sava, F. A. (2008). Inventarul de personalitate DECAS. Manualul utilizatorului [The DECAS Personality Inventory. The users handbook]. Timisoara: ArtPress. Sava, F. A. (2009). Maladaptive schemas, irrational beliefs and their relationship with Five-Factor personality model. Journal of Cognitive and Behavioral Psychotherapies, 9(2), 135147. Sava, F. A., & Popa, R. I. (2011). Personality types based on the big ve model. A cluster analysis over the Romanian population. Cognition, Brain, Behavior: An interdisciplinary journal, 15(3), 359384. Schmukle, S. C., & Egloff, B. (2005). A latent state-trait analysis of implicit and explicit personality measures. European Journal of Psychological Assessment, 21, 100107. DOI: 10.1027/ 10155759.21.2.100
Eur. J. Pers. 26: 613628 (2012) DOI: 10.1002/per

628

F. A. Sava et al.
Steffens, M. C., & Schulze-Knig, S. (2006). Predicting spontaneous Big Five behavior with Implicit Association Tests. European Journal of Psychological Assessment, 22, 1320. DOI: 10.1027/10155759.22.1.13 Teige, S., Schnabel, K., Banse, R., & Asendorpf, J. B. (2004). Assessment of multiple implicit self-concept dimensions using the Extrinsic Affective Simon Task. European Journal of Personality, 18, 495520. DOI: 10.1002/per.531 Viswesvaran, C., & Ones, D. S. (1999). Meta-analyses of fakability estimates: Implications for personality measurement. Educational and Psychological Measurement, 59, 197210. DOI: 10.1177/ 00131649921969802 Watson, D., & Clark, L. A. (1994). The PANAS-X. Manual for the Positive and Negative Affect Schedule Expanded form. Retrieved from http://www.psychology.uiowa.edu/faculty/clark/panas-x.pdf Wittenbrink, B., Judd, C. M., & Park, B. (1997). Evidence for racial prejudice at the implicit level and its relationship with questionnaire measures. Journal of Personality and Social Psychology, 72, 262274. DOI: 10.1037/00223514.72.2.262

Schmukle, S. C., Back, M. D., & Egloff, B. (2008). Validity of the ve-factor model for the implicit self-concept of personality. European Journal of Psychological Assessment, 24, 263272. DOI: 10.1027/10155759.24.4.263 Schnabel, K., & Asendorpf, J. B. (2010). The self-concept. New insights from the implicit measurement procedures. In B. Gawronski, & B. K. Payne (Eds.), Handbook of implicit social cognition: Measurement, theory, and applications (pp. 408 425). New York: Guilford Press. Schnabel, K., Asendorpf, J. B., & Greenwald, A. G. (2008). Understanding and using the Implicit Association Test: V. Measuring semantic aspects of trait self-concepts. European Journal of Personality, 22, 695706. DOI: 10.1002/ per.697 Schnabel, K., Banse, R., & Asendorpf, J. B. (2006). Employing automatic approach and avoidance tendencies for the assessment of implicit personality self-concept: The Implicit Association Procedure (IAP). Experimental Psychology, 53, 6976. DOI: 10.1027/16183169.53.1.69

APPENDIX SELF-CONCEPT-RELATED ADJECTIVES INCLUDED IN THE SMP TASK


SMP High N Low N High C Low C High E Low E Adjectives (Romanian translation between brackets) anxious, emotive, helpless, fearful, strained, sensitive, dreamer, gloomy (anxios, emotiv, neajutorat, temtor, tensionat, sensibil, vistor, posomort) calm, well-balanced, relaxed, condent, strong, emotionally stable, serene, rm (calm, echilibrat, relaxat, stpn pe sine, puternic, stabil emoional, degajat, hotrt) serious, organized, thorough, diligent, persevering, bookish, conscientious, cautious (serios, organizat, meticulos, srguincios, perseverent, tocilar, contiincios, prudent) scatter-brained, disorderly, unpunctual, lazy, careless, easy-going, rebellious, inconsistent (zpcit, dezordonat, nepunctual, lene , nepstor, comod, rebel, inconsecvent) sociable, active, talkative, jokester, energetic, cheerful, extraverted, bold (sociabil, activ, vorbre, glume, energic, voios, extrovertit, ndrzne) shy, lonely, passive, quiet, reserved, taciturn, sober, introverted (timid, retras, pasiv, linitit, rezervat, tcut, sobru, introvertit)

Copyright 2012 John Wiley & Sons, Ltd.

Eur. J. Pers. 26: 613628 (2012) DOI: 10.1002/per

You might also like