Self-Competence

10 downloads 0 Views 138KB Size Report
Initially, self-esteem was conceived as an indivisible concept and within that framework the. Rosenberg Self-Esteem Scale (RSES; Rosenberg, 1965) was developed. This ten-item ..... EQS Structural Equations Program Manual. Encino, CA: ...
Personality and Individual Differences 42 (2007) 157–167 www.elsevier.com/locate/paid

Dutch translation of the Self-Liking/Self-Competence Scale – Revised: A confirmatory factor analysis of the two-factor structure Heleen Vandromme a

a,*

, Dirk Hermans a, Adriaan Spruyt

a,b

, Paul Eelen

a

Centre of Learning and Experimental Psychopathology, University of Leuven, Tiensestraat 102, B-3000 Leuven, Belgium b Postdoctoral researcher of the Research Fund KULeuven, University of Leuven, Tiensestraat 102, B-3000 Leuven, Belgium Received 30 January 2006; received in revised form 15 May 2006 Available online 21 August 2006

Abstract In the present study, the Self-Liking/Self-Competence Scale – Revised (Tafarodi & Swann, 2001) was translated into Dutch and psychometric properties of the questionnaire were assessed. According to Tafarodi and Swann (2001), self-esteem is composed of two dimensions, and the questionnaire was constructed specifically to tap these two dimensions of self-esteem, namely self-competence and self-liking. Based on a confirmatory factor analysis (CFA), results confirmed that a two-dimensional model with the factors selfcompetence and self-liking provided a superior fit to the data as compared to a unidimensional model of self-esteem. Reliability and validity of the questionnaire were appropriate. Ó 2006 Elsevier Ltd. All rights reserved. Keywords: Self-esteem; Self-competence; Self-liking; Psychometric properties; Reliability; Validity; CFA

*

Corresponding author. Tel.: +32 16 32 61 17; fax: +32 16 32 60 99. E-mail address: [email protected] (H. Vandromme).

0191-8869/$ - see front matter Ó 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.paid.2006.07.001

158

H. Vandromme et al. / Personality and Individual Differences 42 (2007) 157–167

1. Introduction For more than four decades, the concept of self-esteem has taken a central place in social psychology research. Nevertheless, there has been a great deal of controversy concerning the definition of this concept: Over a hundred different definitions of self-esteem have been proposed (Mruk, 1999). Generally, self-esteem refers to an individual’s overall evaluation of the self (Gecas, 1982). Initially, self-esteem was conceived as an indivisible concept and within that framework the Rosenberg Self-Esteem Scale (RSES; Rosenberg, 1965) was developed. This ten-item questionnaire is considered a measure of an individual’s global perception of his or her self-worth, and endorses the view of self-esteem being a unidimensional construct. However, self-esteem’s unidimensionality has been debated from the beginning. Convincing evidence for a dualistic approach of self-esteem was found by Tafarodi and his colleagues (Tafarodi & Milne, 2002; Tafarodi & Swann, 1995, 2001). Investigating the underlying factor structure of the RSES, Tafarodi and Swann (1995) demonstrated that a two-factor model provided a superior fit compared to a one-factor global self-esteem model. According to their model, self-esteem is comprised of the dimensions self-competence and self-liking. Self-competence (SC) refers to the generalized sense of one’s own efficacy or power. Generally, self-competence is considered to depend on the correspondence of goals or intentions with the outcomes of actions aimed at realizing those goals or intentions. The competence dimension of the Self-Liking/Self-Competence Scale reflects the degree to which people see themselves as capable and efficacious (efficacy-based self-esteem). Conversely, Self-liking (SL) refers to the generalized sense of one’s own worth as a social object and it is considered to depend more on internalized positive regards from others. The liking dimension of the SLCS-R is a reflection of the degree to which individuals feel they are persons of value (worth-based self-esteem). Based on their findings, Tafarodi and Swann (1995) developed a questionnaire specifically designed to measure both dimensions of self-esteem, namely the Self-Liking/Self-Competence Scale (SLCS). The authors suggested that the self-competence and self-liking dimension should be considered as two correlated but distinct factors of global self-esteem (Tafarodi & Swann, 2001). Based on a confirmatory factor analysis (CFA), the questionnaire was revised and reduced to a sixteen-item questionnaire, consisting of eight items for each of the two subscales (SLCS-Revised; Tafarodi & Swann, 2001). Several studies demonstrated that discerning self-liking from self-competence can be valuable. For example, Bardone, Perez, Abramson, and Joiner (2003) conducted a study in which students completed the RSES and the Eating Disorder Inventory (EDI; Garner, Olmstead, & Polivy, 1983) on two separate occasions. They demonstrated that self-competence related items of the RSES significantly predicted the amount of change in bulimic symptoms over time whereas this was not the case for the self-liking related items. Similarly, Sassaroli and Ruggiero (2005) studied the association between self-esteem (as measured by the SLCS) and eating disorder symptoms (as measured by the EDI). Results indicated that in a stressful situation both self-liking and self-competence predicted ‘Drive for thinness’, but that only self-liking was associated with the amount of ‘Body dissatisfaction’. Finally, Tafarodi and Vu (1997) presented their participants two sets of 20 anagrams, of which 10 were unsolvable. After receiving failure feedback, participants who had low self-liking – but not those who had low self-competence – persisted less in solving the puzzles.

H. Vandromme et al. / Personality and Individual Differences 42 (2007) 157–167

159

The aim of the present study was twofold. First, we developed a reliable and valid Dutch version of the SLCS-R (Tafarodi & Swann, 2001). The internal consistency of the questionnaire was examined. Next, convergent validity was assessed by correlating the self-competence and the selfliking subscale scores with scores on other well-established self-esteem questionnaires (RSES; SelfAttributes Questionnaire, SAQ). In addition, discriminant validity between self-competence and self-liking was addressed based on Steiger’s Z-tests, which examined the strength of the correlations between the scores on the SLCS-R subscales and the scores on the NEO-FFI (Costa & McCrae, 1992). Second, we assessed the factor structure of the SLCS-R. A confirmatory factor analysis was performed to test whether the underlying factor structure corresponds with the two-dimensional model proposed by Tafarodi and Swann (2001). It was predicted that the two-dimensional factor structure would provide a superior fit compared to a unidimensional structure. Aidman (1998) previously investigated the factor structure of this questionnaire. This study, however, made use of an exploratory, rather than a confirmatory factor analysis. Moreover, the twentyitem version of the SLCS was examined, instead of the revised sixteen-item version. Recently, an increasing number of studies use the SLCS(-R) as a measure of self-esteem, in experimental studies as well as clinical studies (e.g. predicting outcome in patients with a certain psychopathology). A reliable and valid Dutch version of the SLCS-R will undoubtedly be beneficial for future self-esteem research.

2. Method 2.1. Participants At the beginning of the academic year, 396 first-grade psychology students participated in a mass testing session for which they all received a course credit in exchange. Only the scores of those who fully completed the questionnaires were included in the analysis1. Three hundred forty one participants (274 women; mean age, 18.4 years) were retained for further analysis. To assess discriminant validity between self-competence and self-liking, 374 students completed a personality questionnaire in a collective testing session two weeks later. However, only the scores of those who also completed the first questionnaires correctly were retained, reducing the sample size to 307 students (250 women; mean age, 18.3 years). 2.2. Instruments 2.2.1. Self-Liking/Self-Competence Scale-Revised (SLCS-R) The SLSC-R is a self-report questionnaire (Tafarodi & Swann, 2001), which contains eight items for each of the two dimensions of self-worth, namely self-competence (e.g. ‘‘I am highly effective at the things I do’’) and self-liking (e.g. ‘‘I tend to devalue myself’’). Participants are 1

For 41 participants, the scores on the RSES were excluded from the sample because they failed to complete the questionnaire correctly. For similar reasons, scores of 18 participants on the SLCS-R as well as scores of 15 participants on the SAQ were excluded. By excluding all participants who failed to complete one or more questionnaires correctly, data of a total of 55 participants were lost.

160

H. Vandromme et al. / Personality and Individual Differences 42 (2007) 157–167

requested to rate the items on a 5-point Likert scale ranging from 1 (strongly disagree) to 5 (strongly agree). For each subscale separately, the item scores are then summed and combined into an overall subscale score. These can range from 8 to 40, with higher scores indicating higher selfcompetence or higher self-liking. In the present study, a Dutch version of the questionnaire was developed in which all 16 items of the SLCS-R (Tafarodi & Swann, 2001) were translated into Dutch. Next, all translated items were backtranslated by a native English speaker and these were submitted for revision to the two authors who originally developed the scale (personal communication with R. W. Tafarodi, February 18, 2004; personal communication with W. B. Swann, February 25, 2004). Based on their suggestions, a final Dutch version of the SLCS-R was obtained. 2.2.2. Rosenberg self-esteem scale (RSES) Rosenberg’s (1965) RSES is a 10-item questionnaire which measures feelings of global self-esteem. On a 4-point Likert scale, participants indicate how strongly they (dis)agree with statements such as ‘‘I feel that I have a number of good qualities’’. An overall self-esteem score is computed by summing the scores of all ten items. The Dutch version of the RSES, developed by Franck, De Raedt, Barbez, and Rosseel (submitted) was administered and the internal consistency of the questionnaire was good (a = .87). 2.2.3. Self-attributes questionnaire (SAQ) The SAQ (Pelham & Swann, 1989) is a self-report questionnaire that asks participants to compare themselves to other students of similar age and sex on eight different domains: intellectual or academic ability, social skills, musical ability, artistic ability, athletic ability, physical attractiveness, leadership ability and common sense. For each of these domains the participant is asked to indicate his or her relative position on a scale ranging from 5 (well below average) to 95 (well above average). Responses to this scale are then averaged to form a composite measure of the participants’ self-views. Internal consistency for this measure reached a value of .71. Similar results were reported by Bosson, Swann, and Pennebaker (2000), and by Brennan and Morris (1997), who found a Cronbach’s alpha of .64 and .62 respectively. 2.2.4. NEO five factor inventory (NEO-FFI) The NEO-FFI is a 60-item questionnaire designed to measure the five major personality domains: Neuroticism, Extraversion, Openness, Agreeableness, and Conscientiousness (Costa & McCrae, 1992). Participants were requested to indicate their level of agreement with each item on a 5-point Likert scale ranging from 1 (strongly disagree) to 5 (strongly agree). Internal consistency was good for Neuroticism and Extraversion, acceptable for Conscientiousness and marginally acceptable for Openness and Agreebleness (aN = .89, aE = .81, aO = .67, aA = .68, aC = .77). 2.3. Procedure All questionnaires were administered at the beginning of the semester during a mass testing session. Participants were requested to fill in their responses on a special form, which enabled computerized scoring of the responses. The self-esteem questionnaires were filled out two weeks before the NEO-FFI.

H. Vandromme et al. / Personality and Individual Differences 42 (2007) 157–167

161

3. Results 3.1. Descriptive statistics Results about means, standard deviations and range of the observed scores of the self-esteem questionnaires are reported in Table 1. Descriptive statistics of the NEO-FFI subscales are reported in Table 2. 3.2. Internal consistency Internal consistency was assessed separately for the self-competence and the self-liking subscale of the SLCS-R by calculating both coefficient alpha and split-half reliability. According to the reliability standards of Barker, Pistrang, and Elliott (2002), Cronbach’s alpha can be considered acceptable for the self-competence subscale (a = .78), and good for the self-liking subscale (a = .89). Split-half reliabilities were also estimated, based on a first half versus second half split of all questionnaire items2. Again, higher reliabilities were found for the self-liking subscale (r = .92) compared to those found for the self-competence subscale (r = .75). Reliabilities can be considered good for self-liking, and acceptable for self-competence (Barker et al., 2002). 3.3. Convergent validity To assess convergent validity, scores on both subscales of the SLCS-R were compared to scores on the RSES and the SAQ, and zero-order correlations were performed using Pearson correlations (Table 1). A correlation of .53 was obtained between the scores on the self-competence scale and the scores on the RSES, indicating good convergent validity (Barker et al., 2002). Similarly, a correlation of .36 was obtained between the scores on the self-competence subscale and the ones on the SAQ, which demonstrates acceptable validity. Next, a correlation of .74 was found for the self-liking subscale of the SLCS-R with the RSES, and a correlation of .40 for self-liking with the SAQ. All correlations were highly significant (p’s < .0001), suggesting strong convergent validity for both self-competence and self-liking. 3.4. Discriminant validity In Table 2, the scores on both SLCS-R subscales were correlated with the scores on each of the five personality dimensions as measured by the NEO-FFI (Costa & McCrae, 1992). To demonstrate discriminant validity between self-competence and self-liking, we compared the strength of the correlation between a personality dimension, e.g. Neuroticism and the self-competence subscale with the correlation of Neuroticism and the self-liking subscale using Steiger’s Z-test (Steiger, 1980). This test enables us to see whether self-competence and self-liking are equally strongly 2

Reliabilities reported here are Spearman-Brown split-half reliabilities. By creating two subscales with only half as many items as the full scale, a systematic underestimation of true reliabilities arises. A correction upward by the Spearman-Brown formula provides a more accurate estimate of the final split-half correlation (see Feldt & Charter, 2003).

162

H. Vandromme et al. / Personality and Individual Differences 42 (2007) 157–167

Table 1 Descriptive statistics and correlations among SC, SL, RSES, and SAQ composite self-views Mean (SD) SC SL RSES SAQ

24.2 25.4 28.9 50.8

(4.1) (5.9) (4.7) (11.2)

Range 10–35 10–39 13–40 20.0–75.6

SC

SL



.47 –

RSES **

**

.53 .74** –

SAQ .36** .40** .37** –

Note. N = 341. SC = Self-competence; SL = Self-liking; RSES = Rosenberg Self-esteem Scale; SAQ = Self-attributes Questionnaire. ** p < .0001.

Table 2 Descriptive statistics of the NEO-FFI and discriminant validity N E O A C

Mean (SD)

Range

36.1 43.0 41.2 42.8 40.5

12–55 22–58 27–54 29–56 26–56

(8.3) (6.1) (5.3) (5.0) (5.6)

SC .41** .26** .07** .08** .51**

SL .68** .40** .14* .04** .17**

Z 4.98** 1.90* 0.92 0.50 4.75**

Note. N = 307. NEO-FFI = NEO five factor inventory; N = Neuroticism; E = Extraversion; O = Openness; A = Agreeableness; C = Conscientiousness; SC = Self-competence; SL = Self-liking; Z = Steiger’s Z. * p < .05. ** p < .01.

associated with Neuroticism. In this test, Fisher’s r correlations are transformed into Z-scores and the difference between a pair of correlations rxy and rxz is calculated, with x the score on Neuroticism, y the score on the self-competence subscale of the SLCS-R and z the score of the self-liking subscale of the SLCS-R. The results indicated that self-liking is more strongly correlated to Neuroticism, compared to self-competence, Z = 4.99, p < .0001. Similar Z-tests were performed for all personality dimensions of the NEO-FFI. Also Extraversion was associated more strongly with self-liking as with self-competence, Z = 1.90, p < .05. However, the degree of Conscientiousness was more strongly correlated to self-competence than to self-liking, Z = 4.75, p < .0001. For Openness and Agreeableness, the corresponding Z-value did not reach significance, respectively Z = 0.92, p > .10 and Z = 0.50, p > .10, suggesting that both self-competence and self-liking are equally strongly associated with these dimensions of personality. 3.5. Confirmatory factor analysis A first-order CFA was performed to test the assumed underlying two-factor structure of the Dutch SLCS-R. To assess the claim of Tafarodi and his colleagues that the SLCS-R can be best conceptualized by the dimensions SL and SC (Tafarodi & Milne, 2002; Tafarodi & Swann, 2001), two measurement models were compared. In the unidimensional model, all 16 items were specified as indicators of one global self-esteem factor. The second model demonstrated the a priori struc-

H. Vandromme et al. / Personality and Individual Differences 42 (2007) 157–167

163

Table 3 Factor loadings for all measurement models

Item Item Item Item Item Item Item Item Item Item Item Item Item Item Item Item

2 4 8 10 12 13 14 16 1 3 5 6 7 9 11 15

Unidimensional model

Competence-liking A

Competence-liking B

SC

SC

.41 .36 .37 .26 .45 .29 .39 .44 .71 .78 .82 .50 .77 .76 .62 .74

.57 .56 .54 .54 .65 .59 .47 .58

SL

SL

.58 .56 .52 .47 .68 .54 .49 .58 .73 .78 .84 .50 .79 .75 .62 .75

.76 .81 .83 .49 .78 .76 .61 .74

Note. SC = Self-competence; SL = Self-liking.

ture of the questionnaire and will be referred to as the Competence-Liking A model. It distinguished two latent factors, namely Self-competence and Self-liking, with the eight items questioning self-competence specified as indicators of the SC factor and the eight items questioning self-liking as indicators of the factor SL. For both models, error covariances were constrained to zero. Model testing was performed using LISREL 8.50 (Jo¨reskog & So¨rbom, 2001). Standardized factor loadings found in the Unidimensional model ranged from .26 to .82, all p’s < .0001; in the Competence-Liking A model, the factor loadings ranged from .47 to .65 for SC and from .50 to .84 for SL; all p’s < .0001 (Table 3). Modification indices in the latter model indicated a considerable improvement of the model fit, when measurement errors would be allowed to vary for certain items. Similar to Pullman and Allik (2000), and Franck et al. (submitted), a maximum of two correlated errors were allowed within a third measurement model3. Within this modified model, which will be referred to as the Competence-Liking B model, factor loadings ranged from .47 to .68 for SC and from .49 to .83 for SL; all ps < .0001. In both two-dimensional models, covariance between the factors SC and SL was allowed and freely estimated. The standardized model revealed a correlation of .55 between both factors for the Competence-Liking A model. A correlation of .56 was found for the Competence-Liking B model. Goodness-of-fit indices for all models were examined (Table 4). Apart from the traditional Chi-square statistic, for which a higher value indicates a worse fit, we also consider the Comparative Fit Index (CFI; Bentler, 1995), the Non-Normed Fit Index (NNFI; Bentler & Bonnet, 1980), the Normed Fit Index (NFI; Bentler & Bonnet, 1980), the Root Mean Square Error of 3 Within the Competence-Liking B model, error covariance was allowed between item1 and item3, and within item10 and item13, with r, respectively r = .16 and r = .24. In both cases, the items were indicators of the same latent factor.

164

H. Vandromme et al. / Personality and Individual Differences 42 (2007) 157–167

Table 4 Goodness-of-fit for Self-Liking/Self-Competence Scale – Revised version (SLCS-R) measurement models Model Null Unidimensional Competence-Liking A Competence-Liking B

df 120 104 103 101

v2 **

4912 836** 383** 300**

CFI

NNFI

NFI

RMSEA

SRMR

CAIC

– .89 .95 .96

– .87 .94 .96

– .87 .93 .94

– .144 .089 .076

– .100 .063 .057

5021 1055 608 539

Note. CFI = Comparative Fit Index; NNFI = Non-Normed Fit Index; NFI = Normed Fit Index; RMSEA = RootMean Squared Error Approximation; SRMR = Standardized Root Mean Square Residual; CAIC = Consistent Version of Akaike’s Information Criterion. ** p < .0001.

Approximation (RMSEA; Browne & Cudeck, 1993), the Standardized Root Mean Square Residual (SRMR; see e.g. Hu & Bentler, 1999) and the consistent version of Akaike’s Information Criterion (CAIC; Bozdogan, 1987). Generally, a good-fitting model reaches values of at least .90 for the CFI, the NNFI and the NFI, indicating that only a small amount of the observed variance is left unexplained by the proposed model. According to this criterion, the Unidimensional model does not provide a good fit to the data, with none of the fit indices exceeding .90. On the other hand, all three fit indices of the Competence-Liking models attained the required .90 value, though the values for the Competence-Liking B model were consistently higher than those for the Competence-Liking A model. Next, the RMSEA index was considered. Based on MacCallum, Browne, and Sugawara (1996), only a poor fit (RMSEA > .10) was found for the Unidimensional model, whereas a mediocre fit was found for the Competence-Liking A model (RMSEA = .09). The Competence-Liking B model provided an acceptable fit to the data (RMSEA = .08). Next, the SRMR was considered, with smaller values indicating a better model fit. Values for both Competence-Liking models are very satisfactory, however the SRMR for the Unidimensional model indicates a bad model fit. Finally, the issue of parsimony was considered by the CAIC, with smaller values representing a better fit of the hypothesized model (Hu & Bentler, 1995). Both Competence-Liking models, containing two latent factors, appeared to be less parsimonious than the Unidimensional model. In its turn, the Competence-Liking B model appeared to be slightly less parsimonious. Next, we tried to identify the best among the three models. Since the models are nested, a Chi-square difference test can be used to compare the goodness-of-fit between two models directly. A superior fit was found for the Competence-Liking A model compared to the Unidimensional model, as well as a superior fit for the Competence-Liking B model compared to the Unidimensional model, respectively v2(1) = 454, p < .0001; v2(3) = 536, p < .0001. Finally, both two-dimensional models were compared, indicating a superior fit in favour of the CompetenceLiking B model, v2(2) = 82, p < .0001.

4. Discussion The primary aim of this study was to develop a reliable and valid Dutch version of a frequently used self-esteem questionnaire, namely the Self-Liking/Self-Competence Scale – Revised. A CFA was also performed to check the a priori factor structure (Tafarodi & Swann, 2001).

H. Vandromme et al. / Personality and Individual Differences 42 (2007) 157–167

165

According to Tafarodi and his co-authors (Tafarodi & Milne, 2002; Tafarodi & Swann, 1995, 2001), self-esteem comprises two dimensions, namely self-competence and self-liking. They developed the SLCS to question an individual’s generalized sense of efficacy and power (self-competence) and one’s generalized sense of worth as a social object (self-liking) (Tafarodi & Swann, 1995). Reliability and validity measures of both the self-competence and the self-liking subscale were good, though the reliabilities for self-liking were consistently higher compared to those for self-competence. The present findings correspond with those of Brown and Zeigler-Hill (2004), who also reported higher reliabilities for self-liking than for self-competence (aSC = .81; aSL = .92). Evidence of discriminant validity between self-competence and self-liking was also provided based on the correlations between the SLCS-R and the NEO-FFI subscales. Conscientiousness on the one hand, measuring the degree to which someone describes himself or herself as organized, planful and thorough, is associated more strongly with self-competence. Neuroticism and extraversion on the other hand are more strongly associated with self-liking. Neuroticism reflects one’s degree of anxiousness and vulnerability whereas extraversion reflects one’s degree of self-consciousness, assertiveness and personally perceived activity level. These relationships are in line with what could be expected based on the content of the personality dimensions. Subsequently, the underlying factor structure of the questionnaire was examined. A CFA was performed to test whether the two-dimensional structure, as proposed by Tafarodi and Swann (2001), could be supported in our Dutch version of the SLCS-R. A model with the two latent factors self-competence and self-liking was compared to a unidimensional model in which all items loaded on a single global self-esteem factor. Based on the goodness-of-fit indices, the two-dimensional model clearly provided a superior fit to the data, indicating that the SLCS-R could be conceptualized better by the dimensions self-competence and self-liking. Moreover, allowing two measurement errors to covary led to an increase in the fit indices, indicating that the Competence-Liking B model provides an even better fit to the data. However, several authors have raised concerns about the validity and the use of such approximate fit indices (e.g. Beauducel & Wittmann, 2005). Based on a simulation study, Beauducel and Wittmann (2005) showed that small distortions in the simple structure can lead to serious misfit of the approximate fit indices, whereas misfit for the SRMR and the RMSEA does not occur. They propose a two-index-strategy, in which the SRMR is combined with the RMSEA to detect misspecification in both the structural and the measurement model parameters. Considering these two indices, it becomes clear that we need to be cautious in claiming that the Competence-Liking A model is an adequate model. According to the criteria developed by MacCallum et al. (1996), the model only obtained a mediocre fit, and some residual error still exists within the model. On the other hand, the Competence-Liking B model appears to reach an acceptable model fit. However, allowing (a maximum of two) measurement errors to covary can lead to a loss of meaning of the model (Gerbing & Anderson, 1984). Generally, correlated errors are due to an unknown common source, such as the wording of an item, or a socially desirable response tendency. For example, when a respondent gives a socially desirable response to one item, the probability that a socially desirable response will be given to another item, can be assumed to increase. In that case measurement errors can become correlated. By allowing two errors to covary, the model fit of the Competence-Liking B model was slightly improved. However, the overall pattern of factor loadings remained entirely similar. Hence, we believe that the possible loss of meaning of the model remained very limited.

166

H. Vandromme et al. / Personality and Individual Differences 42 (2007) 157–167

As was to be expected, a strong correlation between both latent factors was found for both twodimensional models, respectively r = .55; r = .56. An even stronger correlation was obtained by Tafarodi and Swann (2001), who found a correlation of .78. Defenders of a unidimensional view of self-esteem sometimes consider the strong correlation between both subscales to be problematic, but one can easily assume that self-competence and self-liking mutually influence each other. For example, when someone experiences himself or herself capable of performing certain tasks or achieving certain goals, it is very likely that positive feelings will arise from this experience. This will lead that person to perceive himself or herself as (more) valuable. In this way, self-competence can influence self-liking. On the other hand, when someone feels good about himself or herself, it is very likely that he or she will also feel confident to rise to certain challenges. This will enable that person to experience success and to achieve the belief of self-competence. In turn, self-liking can also influence self-competence. It can be concluded that the Dutch version of the SLCS-R has good psychometric properties. The proposed two-dimensional factor structure of the SLCS-R, discerning SC from SL, clearly provides a better fit than a unidimensional model.

References Aidman, E. V. (1998). Analysing global dimensions of self-esteem: Factorial structure and reliability of the Self-Liking/ Self-Competence Scale. Personality and Individual Differences, 24, 735–737. Bardone, A. M., Perez, M., Abramson, L. Y., & Joiner, T. E. (2003). Self-competence and self-liking in the prediction of change in bulimic symptoms. International Journal of Eating Disorders, 34, 361–369. Barker, C., Pistrang, N., & Elliott, R. (2002). Research Methods in Clinical Psychology (second ed.). Chichester: Wiley. Beauducel, A., & Wittmann, W. (2005). Simulation study on fit indices in confirmatory factor analysis based on data with slightly distorted simple structure. Structural Equation Modeling, 12, 41–75. Bentler, P. M. (1995). EQS Structural Equations Program Manual. Encino, CA: Multivariate Software, Inc. Bentler, P. M., & Bonnet, D. G. (1980). Significance tests and goodness-of-fit in the analysis of covariance structures. Psychological Bulletin, 88, 588–606. Bosson, J. K., Swann, W. B., & Pennebaker, J. W. (2000). Stalking the perfect measure of implicit self-esteem: The blind men and the elephant revisited? Journal of Personality and Social Psychology, 79, 631–643. Bozdogan, H. (1987). Model selection and Akaike’s information criterion (AIC): The general theory and its analytical extensions. Psychometrika, 52, 345–370. Brennan, K. A., & Morris, K. A. (1997). Attachment styles, self-esteem, and patterns of seeking feedback from romantic partners. Personality and Social Psychology Bulletin, 23, 23–31. Brown, R. P., & Zeigler-Hill, V. (2004). Narcissism and the non-equivalence of self-esteem measures: A matter of dominance? Journal of Research in Personality, 38, 585–592. Browne, M. W., & Cudeck, R. (1993). Alternative ways of assessing model fit. In K. Bollen & S. Long (Eds.), Testing Structural Equation Models (pp. 136–162). Newbury Park, NJ: Sage. Costa, P. T., Jr., & McCrae, R. R. (1992). Revised NEO Personality Inventory (NEO-PI-R) and NEO Five-Factor Inventory (NEO-FFI): professional manual. Odessa, FL: Psychological Assessment Resources. Feldt, L. S., & Charter, R. A. (2003). Estimating the reliability of a test split into two parts of equal or unequal length. Psychological Methods, 8, 102–109. Franck, E., De Raedt, R., Barbez, C., & Rosseel, Y. Psychometric properties of the Dutch Rosenberg self-esteem scale. European Journal of Psychological Assessment, submitted for publication. Garner, D. M., Olmstead, M. P., & Polivy, J. (1983). The eating disorder inventory: A measure of cognitive-behavioral dimensions of anorexia nervosa and bulimia. In P. L. Darby, P. E. Garfinkel, D. M. Garner, & D. V. Coscina (Eds.), Anorexia Nervosa: Recent Developments in Research (pp. 173–184). New York: Liss.

H. Vandromme et al. / Personality and Individual Differences 42 (2007) 157–167

167

Gecas, V. (1982). The self-concept. Annual Review of Sociology, 8, 1–33. Gerbing, D. W., & Anderson, J. C. (1984). On the meaning of within-factor correlated measurement errors. Journal of Consumer Research, 11, 572–580. Hu, L., & Bentler, P. M. (1995). Evaluating model fit. In R. H. Hoyle (Ed.), Structural Equation Modeling: Concepts Issues and Applications (pp. 76–99). London: Sage. Hu, L., & Bentler, P. M. (1999). Cut-off criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling, 6, 1–55. Jo¨reskog, K. G., & So¨rbom, D. (2001). LISREL 8.50. Chicago: Scientific Software International. MacCallum, R. C., Browne, M. W., & Sugawara, H. M. (1996). Power analysis and determination of sample size for covariance structure modeling. Psychological Methods, 1, 130–149. Mruk, C. (1999). Self-Esteem: Research (second ed.). New York: Springer. Pelham, B. W., & Swann, W. B. (1989). From self-conceptions to self-worth: On the sources and structure of global selfesteem. Journal of Personality and Social Psychology, 57, 672–680. Pullman, H., & Allik, J. (2000). The Rosenberg self-esteem scale: its dimensionality, stability, and personality correlates in Estonian. Personality and Individual Differences, 28, 701–715. Rosenberg, M. (1965). Society and the Adolescent Self-Image. Princeton, NJ: Princeton University Press. Sassaroli, S., & Ruggiero, G. M. (2005). The role of stress in the association between low self-esteem, perfectionism, and worry, and eating disorders. International Journal of Eating Disorders, 37, 135–141. Steiger, J. H. (1980). Tests for comparing elements of a correlation matrix. Psychological Bulletin, 87, 245–251. Tafarodi, R. W., & Milne, A. B. (2002). Decomposing global self-esteem. Journal of Personality, 70, 443–483. Tafarodi, R. W., & Swann, W. B. Jr., (1995). Self-liking and self-competence as dimensions of global self-esteem: Initial validation of a measure. Journal of Personality Assessment, 65, 322–342. Tafarodi, R. W., & Swann, W. B. Jr., (2001). Two-dimensional self-esteem: Theory and measurement. Personality and Individual Differences, 31, 653–673. Tafarodi, R. W., & Vu, C. (1997). Two-dimensional self-esteem and reactions to success and failure. Personality and Social Psychology Bulletin, 23, 626–635.