Confirmatory factor analysis of the Sensitivity to Punishment and Sensitivity to Reward Questionnaire

Personality and Individual Differences 37 (2004) 985–1002 www.elsevier.com/locate/paid

Confirmatory factor analysis of the Sensitivity to Punishment and Sensitivity to Reward Questionnaire Roisin M. OÕConnor, Craig R. Colder *, Larry W. Hawk, Jr. Department of Psychology, State University of New York at Buffalo, Park Hall, Box 604110, Buffalo, NY 14260-4110, USA Received 13 February 2003; received in revised form 24 October 2003; accepted 17 November 2003 Available online 17 January 2004

Abstract The Sensitivity to Punishment and Sensitivity to Reward Questionnaire (SPSRQ) was developed by Torrubia et al. (1995) as a self-report measure of the Behavioral Inhibition System (BIS) and the Behavioral Approach System (BAS), two motivational systems proposed by GrayÕs model (1975, 1982, 1987a, 1987b). Past research used exploratory factor analysis to examine the factor structure of the SPSRQ. Some support was found for 2 orthogonal factors––Sensitivity to Punishment and Sensitivity to Reward, which correspond to the BIS and BAS, respectively. The purpose of the present study was to further assess the underlying factor structure of the SPSRQ using Confirmatory Factor Analysis (CFA). Results did not support two-, three-, four-, or five-factor models. Based on factor loadings the two-factor CFA, items were removed systematically and a final model, with a reduced pool of items, was tested with the original data sample ðN ¼ 603Þ and two additional independent samples (N ¼ 104; N ¼ 163). Results based on the reduced pool of items suggested support for a two-factor model, but some problems were identified, thus further improvements in this self-report measure should be considered in future research. Ó 2003 Elsevier Ltd. All rights reserved. Keywords: Punishment; Reward; Approach; Inhibition; Factor analysis

1. Introduction GrayÕs model (Gray, 1981, 1982, 1987a, 1987b) proposes that personality is a function of three motivational systems. Two of the systems have been the focus of much research in a variety of fields. The Behavioral Inhibition System (BIS) is characterized by sensitivity to conditioned cues *

Corresponding author. Tel.: +1-716-645-3650x235; fax: +716-645-3801. E-mail address: [email protected]ffalo.edu (C.R. Colder).

0191-8869/$ - see front matter Ó 2003 Elsevier Ltd. All rights reserved. doi:10.1016/j.paid.2003.11.008

986

R.M. OÕConnor et al. / Personality and Individual Differences 37 (2004) 985–1002

for potential punishment and frustrating non-reward, and presentation of novel stimuli. Activation of this system is associated with the inhibition of behavior, such that those with a dominant BIS inhibit action to avoid potential aversive consequences. The BIS is the system that underlies anxiety. The Behavioral Approach System (BAS), which underlies elation/relief, is associated with activation or approach behaviors, and sensitivity to conditioned cues for reward or non-punishment, and is the basis for impulsivity, which is believed to be the core feature of disinhibition (Patterson & Newman, 1993). Presently, there are few good self-report measures that assess the BAS and BIS as theorized by  Gray. Torrubia, Avila, Molt o, and Grande (1995) developed a measure that holds great promise.  Based on a review of GrayÕs model, Torrubia, Avila, Molt o, and Caseras (2001) identified criteria that should be met by any measure intending to assess the BIS and BAS as theorized by Gray, including how the two dimensions relate to each other, and to neuroticism and extraversion. In addition, they argued that items should not describe overlearned situations, to which everyone will respond in the same way. An example of such an overlearned item is ‘‘Would you go to get the money if you won a prize in the lottery?’’ (Torrubia et al., 2001). The Sensitivity to Punishment and Sensitivity to Reward Questionnaire (SPSRQ, Torrubia et al., 1995), which is the focus of the present study, was developed according to these criteria. Where Sensitivity to Punishment (SP) items measure the BIS and Sensitivity to Reward (SR) items measure the BAS. SP items describe reactivity in situations with a predominance of punishment, while SR items describe reactivity in situations that are predominantly rewarding. SR items do not describe situations that assess active avoidance learning. This is important, as based on GrayÕs theoretical model, active avoidance learning may necessitate both activation of the BIS and BAS, and therefore, should not be used as a measure of the BAS only (Torrubia et al., 2001). In an earlier publication of the 48-item SPSRQ measure, Torrubia et al. (1995) conducted an exploratory factor analysis (EFA; principal components analysis) and reported that all items loaded adequately and as expected on two factors (24 items on each factor). The data supporting this conclusion (e.g., factor loadings), however, were not presented. Reliability was good for both scales (a reliability in the range of 0.76–0.84). Also, test–retest correlations over three-month and three-year periods were 0.89 and 0.57, respectively. Previous research also supports the validity of the SPSRQ. Torrubia et al. (1995) found that scores on the SP and SR were associated with performance during a computer task. High SP scores were associated with a low number of punishable errors and a low number of responses when the participant was unsure of a correct answer during a computer game. Furthermore, following termination of reward, those higher in sensitivity to punishment were quicker to extinguish approach behavior. While high scores on the SR scale were associated with a high  number of passive avoidance errors and Avila (2001) found a reduced ability to inhibit responses (passive avoidance) during a reaction time task by those who scored high on the SR. In addition, Brebner and Martin (1995) found a significant correlation between SR and Eysenck and EysenckÕs (1978) Impulsiveness scale ðr ¼ 0:43Þ. SP was also significantly negatively correlated with extraversion ðr ¼
0:50Þ and positively correlated with neuroticism ðr ¼ 0:54Þ. A more recent paper by  Caseras, Avila, and Torrubia (2003) similarly supports the convergent validity of the SP and SR scales with other questionnaire measures. Torrubia et al. (2001) again found support for reliability and validity of the SP and SR scales. The authors concluded that an exploratory factor analysis (EFA; principal factor analysis)

R.M. OÕConnor et al. / Personality and Individual Differences 37 (2004) 985–1002

987

demonstrated ‘‘acceptable’’ support for two factors. However, a close look at the results suggests problems with the EFA. Results of the exploratory factor analysis, run separately for males ðN ¼ 470Þ and females ðN ¼ 1093Þ, indicated that seven items did not show substantial loadings on their respective factors (<0.32) for both sexes. Cross-loadings could not be detected as factor loadings were only presented for indicators of the hypothesized factor. Overall, the SPSRQ appears to be a promising measure for researchers interested in studying GrayÕs motivational systems. The validity data are impressive with the scales predicting other measures as expected, and performance on a variety of lab tasks. However, the hypothesized factor structure has not been satisfactorily examined. No studies have examined this measure using a confirmatory factor analysis (CFA). This is a notable omission given the a priori expectations of the factor structure. The goal of this study was to confirm the two-factor structure in several independent samples. We expected to support two-factors, but thought that some problematic items might be identified based on previous EFA results. Moreover, previous studies used factor analytic techniques that assume the observed variables are continuous and normally distributed. This is inappropriate given the binary response options of the SPSRQ. Accordingly, we estimated our CFAs using Weighted Least Squares (WLS) estimator, which is appropriate for binary observed data (Byrne, 2001; Kline, 1998).

2. Method 2.1. Procedure The Sensitivity to Punishment and Sensitivity to Reward questionnaire was administered to three independent samples of students taking the first year introductory psychology course at a large university. All students received 1 credit towards their experimental course requirement. In addition to the SPSRQ, two other questionnaires administered to one of the samples were included to assess validity (BIS/BAS; Carver & White, 1994; PANAS; Watson, Clark, & Tellegen, 1988). SPSRQ. An English translation of the SPSRQ (Torrubia et al., 1995, 2001) was administered to all three samples. Participants responded either ‘‘yes’’ or ‘‘no’’ to 24 items composing the SP scale and 24 items composing the SR scale. A response of ‘‘yes’’ was assigned a value of one and a ‘‘no’’ a zero, and in accordance with Torrubia et al. (1995, 2001), items were summed to form scale scores. As presented in detail in the introduction, there is adequate support for the reliability and validity of these scales. This is demonstrated by the satisfactory internal consistency of the scales and the test–retest correlations found by Torrubia et al. (1995, 2001). As well, the adequate validity of the SP and SR scales, as measures of the BIS and BAS, respectively, is demonstrated  using a laboratory task (Avila, 2001; Torrubia et al., 1995) and other self-report measures (Brebner & Martin, 1995; Caseras et al., 2003). BIS/BAS scale. Carver and White (1994) developed a self-report measure of the BIS and BAS that included one BIS scale and three BAS scales, which was administered to Sample 3. The BIS scale assesses sensitivity to potentially punishing stimuli (7 items e.g., ‘‘I feel worried when I think I have done poorly at something’’). The three aspects of BAS measured are reward responsiveness

988

R.M. OÕConnor et al. / Personality and Individual Differences 37 (2004) 985–1002

(5 items, e.g., ‘‘When I get something I want, I feel excited and energized’’), drive (4 items, e.g., ‘‘If I see a chance to get something I want, I move on it right away’’), and fun seeking (4 items, e.g., ‘‘I crave excitement and new sensations’’). Participants were asked to rate their level of agreement with each item using a 4-point response scale (1 ¼ strong disagreement to 4 ¼ strong agreement). The items were averaged to form scale scores. There is adequate support for the reliability (e.g., Hawk & Kowmas, 2003; Jorm et al., 1999; Sutton & Davidson, 1997) and validity (Carver & White, 1994; Harmon-Jones & Allen, 1997; Heubeck, Wilkinson, & Cologon, 1998; Jorm et al., 1999; Sutton & Davidson, 1997) of these scales. PANAS. Watson et al. (1988) developed the Positive Affect (PA) and Negative Affect (NA) Schedule to assess trait affect, and this was administered to Sample 3. This measure consists of 20 adjectives, 10 with a negative (e.g., afraid, upset, nervous, ashamed) and 10 of a positive (e.g., excited, inspired, alert, active) valence. Participants used a 5-point scale (1 ¼ none of the time to 5 ¼ most of the time) to rate how often each adjective describes how they feel in ‘‘general’’. Items were averaged to form scale scores. Watson et al. (1988) reported internal consistency coefficients for the scales ranging from 0.89 to 0.90 for positive affect and from 0.84 to 0.87 for negative affect. Validity for this measure has also been supported (e.g., Clark & Watson, 1988, 1991; Watson, 1988a, 1988b). It was expected that SP would be related to NA and SR would be related to PA. 2.2. Participants Only data for those who completed all 48 items of the SPSRQ were included in the analyses. This excluded 45 (6.9%) people from Sample 1, 14 (11.9%) from Sample 2, and 9 (5.3%) from Sample 3. Within each sample people with missing data did not consistently miss the same items, and comparing those with and without missing data on demographics suggested no significant differences, which supported the conclusion that the data were missing at random. Overall, there were no systematic patterns of missingness, thereby suggesting that listwise deletion did not have a strong impact on our findings (Arbuckle, 1996). Sample 1. Sample 1 included a total of 603 students, 250 men (age M ¼ 19:0 years; SD ¼ 3.1; range: 17–59 years-old) and 353 women (age M ¼ 18:4 years; SD ¼ 2.4; range: 17–43 years-old). Of these, 67.2% ðN ¼ 405Þ described themselves as Caucasian, 4.8% ðN ¼ 29Þ as Hispanic/Latino, 9.6% ðN ¼ 58Þ African-American, 13.9% ðN ¼ 84Þ Asian or Asian-American, and 4.2% ðN ¼ 25Þ identified ‘‘Other not listed’’. Two people did not respond to the ethnicity item. This sample included 518 students in their freshman year, 57 sophomores, 16 juniors, 8 seniors, and 2 people who were in ‘‘5th year or more’’, and 2 people did not indicate year of education. Sample 2. Sample 2 included 104 students (49 men and 55 women). The average age of men was 19.8 years old (SD ¼ 1.1) and 20.2 years old for women (SD ¼ 1.0). Of these, 58.6% ðN ¼ 61Þ described themselves as Caucasian, 2.9% ðN ¼ 3Þ as Hispanic/Latino, 7.7% ðN ¼ 8Þ AfricanAmerican, 26.9% ðN ¼ 28Þ Asian or Asian-American, and 2.9% ðN ¼ 3Þ identified ‘‘Other not listed’’. One person did not indicate a response for ethnicity. This sample included 12 students in their freshman year, 33 sophmores, 26 juniors, 25 seniors, and 8 people in ‘‘5th year or more’’. Sample 3. Sample 3 included 160 students; 102 men (age M ¼ 19:1; SD ¼ 0.8) and 58 women (age M ¼ 19:0; SD ¼ 0.9). Of these, 66.9% ðN ¼ 107Þ described themselves as Caucasian, 5.0% ðN ¼ 8Þ as Hispanic/Latino, 11.2% ðN ¼ 18Þ African-American, 10.0% ðN ¼ 16Þ Asian or Asian-

R.M. OÕConnor et al. / Personality and Individual Differences 37 (2004) 985–1002

989

American, and 6.9% ðN ¼ 11Þ identified ‘‘Other not listed’’. This sample included 88 freshman, 45 sophomores, 21 juniors, and 6 seniors. 2.3. Analyses Running a confirmatory factor analysis with ML estimation with dichotomous data as indicators of underlying continuous latent factors can be potentially problematic. As discussed in Kline (1998) and Byrne (2001), the parameter estimates derived using the observed correlations or covariances of these non-continuous indicators might not accurately reflect the real correlations/ covariances between the constructs. This is because correlations between non-continuous observed measures tend to be smaller than what would be expected with the underlying continuous latent variables. Byrne (2001) notes that this problem of inaccurate parameter estimates, due to attenuated correlations is of greatest concern when there are fewer than five categories for a variable, as is seen with the present data (two-category responses). Accordingly, WLS estimation in Mplus version 2.12 (Muthen & Muthen, 2002) was used to accommodate the binary nature of the observed variables. This estimation method adjusts the covariance matrix so that results represent what is expected if the measures had been continuous. 3. Results 3.1. Correlation analysis In general, convergent and divergent validity is supported when items within each scale have higher correlations than items across scales. The mean inter-item correlation for the SP scale was 0.15 (Mdn ¼ 0.14; range ¼ )0.07 to 0.48), and 0.11 (Mdn ¼ 0.10; range ¼ )0.06 to 0.45) for the SR scale, while the mean correlation across scales was 0.03 (Mdn ¼ 0.03; range ¼ )0.30 to 0.24). Overall, the correlations did appear to be slightly lower across than within scales, however, the correlations within scales did not provide strong support for convergent validity. 3.2. Confirmatory factor analysis A CFA was first run with the Sample 1 data so as to evaluate a two-factor structure of the SPSRQ. Full measurement model. A two-factor model was specified in which the 24 SP items were hypothesized as indicators of one factor and the 24 SR items were hypothesized as indicators of another factor. A covariance was estimated between the latent factors. A weak relationship between SP and SR was expected, given the hypothesized orthogonality of these constructs. In evaluating our CFAs, we examined several fit indices. Based on Hu and BentlerÕs (1999) criteria for good fit, the Comparative Fit Index (CFI) and Tucker-Lewis Index (TLI) should be greater than 0.95, and the Root Mean Square Error of Approximation (RMSEA) less than 0.06. Caution should be taken, however, in the interpretation of fit indices when a large pool of observed items is being analyzed, as in this case many parameter estimates will be constrained to zero when simple factor structure is hypothesized. With a large number of constraints, fit indices (e.g., CFI; TLI) are more likely to reflect a poor fit, which can be attributed to a large number of trivial

990

R.M. OÕConnor et al. / Personality and Individual Differences 37 (2004) 985–1002

discrepancies between the observed and model implied covariance matrices. The v2 =df, which adjusts for the sample size, is believed to be a better indicator of the model fit in this situation (Kline, 1998). Generally a v2 =df less than 3.0 is considered good. Model fit should not be evaluated based on a single criterion; rather, fit indices, v2 =df, and size of factor loadings should be weighed equally when drawing conclusions regarding model fit (Byrne, 2001; Kline, 1998). The criteria used in this study to distinguish between substantial and non-substantial factor loadings are based on those by Comrey and Lee (1992). Comrey and Lee (1992) suggest that loadings greater than 0.71 should be considered excellent, 0.63 are very good, 0.55 are good, 0.45 are fair, and 0.32 are poor, where items with factor loadings of less than 0.32 should not be interpreted. The less than 0.32 criterion for weak loading factors is used in the interpretations of the results here. Consistent with the correlations, the two-factor measurement model did not fit the data well 2 (v ð295; N ¼ 602Þ ¼ 968:90, p < 0:001; v2 =df ¼ 3:28; CFI ¼ 0.73; TLI ¼ 0.77; RMSEA ¼ 0.06). In addition to the fit indices, which suggest a poor fit of this model, the parameter estimates also confirmed that the model was not fitting the data well. The standardized factor loadings, critical values, and residual variances are presented in Table 1. Several factor loadings were weak and the percent of variance accounted for in the observed measures, by the latent factors, was sometimes very low (R2 s range from 0.02 to 0.75). There was a significant covariance between the factors (sxy ¼ 0:02, p < 0:05); however, the correlation was small ðr ¼ 0:23Þ, suggesting a weak relationship between the factors. The significance of the factor covariance is hypothesized as attributable to the large sample size. The alpha reliability coefficients for each of the scales (SP: a ¼ 0:81; SR: a ¼ 0:74) was good. In conclusion, there was not strong support for the two-factor model of the SPSRQ items, as proposed by Torrubia et al. (1995, 2001). These findings suggested that it was necessary to examine alternative factor solutions. Accordingly, an EFA was used to identify alternative factor structures that might provide a better fit to the data. 3.3. Exploratory factor analysis

1; 2

The Kaiser–Guttman Rule (Tabachnick & Fidell, 2001) suggested a 14-factor solution. However, after three factors, each additional factor accounted for only 2–5% of the variance, suggesting a two or three factor solution. Both models were examined, however, the three-factor solution was not theoretically interpretable (Note. In the three-factor model 12% of the variance was accounted for by the second factor and only 5% by the third factor). Moreover, the threefactor solution yielded several items that did not load substantially on any factors and several 1

We performed an EFA assuming continuous variables so that we could replicate the EFA presented by Torrubia et al. (2001). In general, the items that loaded weakly on factors in Torrubia et al. (2001) also loaded weakly in our sample. 2 Torrubia et al. (2001) examined the factor structure for men and women and found few gender differences on factor loadings. We performed a multiple group analysis to determine whether there was invariance of the factor structure across sex. CFAs (Maximum Likelihood) with continuous data were run with the unconstrained solution and with the solution where factor loadings were constrained across sex. Nested v2 test supported invariance across sex as the constrained model did not result in a decrement in model fit (v2diffð46Þ ¼ 60:10; p > 0:05). Based on these findings, all further analyses were run with the data collapsed across sex.

R.M. OÕConnor et al. / Personality and Individual Differences 37 (2004) 985–1002

991

Table 1 Standardized factor loadings and standardized residuals for the full measurement model in Sample 1 N ¼ 603 Indicators of Sensitivity to Punishment Latent Factor 1. Do you often refrain from doing something because you are afraid of it being illegal? 3. Do you prefer not to ask for something when you are not sure you will obtain it? 5. Are you often afraid of new or unexpected situations? 7. Is it difficult for you to telephone someone you do not know? 9. Do you often renounce your rights when you know you can avoid a quarrel with a person or an organization? 11. As a child, were you troubled by punishments at home or in school? 13. In tasks that you are not prepared for, do you attach great importance to the possibility of failure? 15. Are you easily discouraged in difficult situations? 17. Are you a shy person? 19. Whenever possible, do you avoid demonstrating your skills for fear of being embarrassed? 21. When you are with a group, do you have difficulties selecting a good topic to talk about? 23. Is it often difficult for you to fall asleep when you think about things you have done or must do? 25. Do you think a lot before complaining in a restaurant if your meal is not well prepared? 27. Would you be bothered if you had to return to a store when you noticed you were given the wrong change? 29. Whenever you can, do you avoid going to unknown places? 31. Are you often worried by things that you said or did? 33. Would it be difficult for you to ask your boss for a raise (salary increase)? 35. Do you generally try to avoid speaking in public? 37. Do you, on a regular basis, think that you could do more things if it was not for your insecurity or fear? 39. Comparing yourself to people you know, are you afraid of many things? 41. Do you often find yourself worrying about things to the extent that performance in intellectual abilities is impaired? 43. Do you often refrain from doing something you like in order not to be rejected or disapproved of by others? 45. Generally, do you pay more attention to threats than to pleasant events? 47. Do you often refrain from doing something because of your fear of being embarrassed? Indicators of Sensitivity to Reward Latent Factor 2. Does the good prospect of obtaining money motivate you strongly to do some things? 4. Are you frequently encouraged to act by the possibility of being valued in your work, in your studies, with your friends or with your family? 6. Do you often meet people that you find physically attractive? 8. Do you like to take some drugs because of the pleasure you get from them? 10. Do you often do things to be praised? 12. Do you like being the center of attention at a party or a social meeting? 14. Do you spend a lot of your time on obtaining a good image?

Factor loadings

Residual variance

0.24 0.48 0.68 0.56 0.24

0.94 0.77 0.53 0.68 0.94

0.16 0.46

0.98 0.79

0.60 0.55 0.67

0.64 0.70 0.56

0.56

0.69

0.27

0.93

0.36

0.87

0.30

0.91

0.39 0.66 0.52 0.55 0.76

0.85 0.57 0.73 0.70 0.42

0.59 0.59

0.65 0.66

0.62

0.61

0.41 0.87

0.83 0.25

0.43 0.33

0.81 0.89

0.21 0.27 0.52 0.42 0.48

0.96 0.93 0.73 0.83 0.77

(continued on next page)

992

R.M. OÕConnor et al. / Personality and Individual Differences 37 (2004) 985–1002

Table 1 (continued) N ¼ 603

Factor loadings

Residual variance

16. Do you need people to show their affection for you all the time? 18. When you are in a group, do you try to make your opinions the most intelligent or the funniest? 20. Do you often take the opportunity to pick up people you find attractive? 22. As a child, did you do a lot of things to get peopleÕs approval? 24. Does the possibility of social advancement move you to action, even if this involves not playing fair? 26. Do you generally give preference to those activities that imply an immediate gain? 28. Do you often have trouble resisting the temptation of doing forbidden things? 30. Do you like to compete and do everything you can to win? 32. Is it easy for you to associate tastes and smells to very pleasant events? 34. Are there a large number of objects or sensations that remind you of pleasant events? 36. When you start to play with a slot machine, is it often difficult for you to stop? 38. Do you sometimes do things for quick gains? 40. Does your attention easily stray from your work in the presence of an attractive stranger? 42. Are you interested in money to the point of being able to do risky jobs? 44. Do you like to put competitive ingredients in all of your activities? 46. Would you like to be a socially powerful person? 48. Do you like displaying your physical abilities even though this may involve danger?

0.52 0.51

0.73 0.74

0.35 0.53 0.60

0.88 0.72 0.64

0.53 0.41 0.46 0.25 0.17 0.22 0.64 0.48

0.72 0.83 0.79 0.94 0.97 0.95 0.58 0.77

0.60 0.47 0.51 0.45

0.64 0.78 0.74 0.80

Note. All factor loadings significant at p < 0:05.

items that loaded on multiple factors. Thus, the more parsimonious two-factor solution was preferred. The standardized factor loadings for the two-factor solution, using the varimax rotation, are reported in Table 2. Use of the orthogonal rotation was based on the hypothesis that the constructs of BIS and BAS are independent systems (Gray, 1981, 1987a, 1987b). Furthermore, the results of the promax rotation showed that the two-factors were not strongly correlated ðr ¼ 0:18Þ. It was evident from the standardized factor loadings that eleven of the items did not load substantially on either factor (Table 2, items 1, 4, 6, 8, 9, 11, 23, 27, 32, 34, and 36) and two items loaded substantially on both factors (Table 2, items 16 and 45). The remaining items loaded substantially on only the expected factor. In order to improve the model, we trimmed items systematically. First, items that did not load substantially on either factor were trimmed from the model, and then items that loaded substantially on both factors were trimmed. This was done iteratively. In total, 13 items, six from the SP scales and seven from the SR scale, were trimmed. Items that were trimmed are indicated by an asterisk in Table 2. 3.4. The final measurement model 3.4.1. CFA with Sample 1. Finally, we tested the trimmed measurement model using a CFA. The final measurement model included 18 indicators of the SP factor and 17 indicators of the SR factor. All items loaded substantially and significantly ðp < 0:05Þ on their respective factors, thus

R.M. OÕConnor et al. / Personality and Individual Differences 37 (2004) 985–1002

993

Table 2 Exploratory factor analysis with categorical data: standardized factor loadings for the two-factor solution using varimax rotation in Sample 1 N ¼ 603 Items from the Sensitivity to Punishment Scale 1. Do you often refrain from doing something because you are afraid of it being illegal?* 3. Do you prefer not to ask for something when you are not sure you will obtain it? 5. Are you often afraid of new or unexpected situations? 7. Is it difficult for you to telephone someone you do not know? 9. Do you often renounce your rights when you know you can avoid a quarrel with a person or an organization?* 11. As a child, were you troubled by punishments at home or in school?* 13. In tasks that you are not prepared for, do you attach great importance to the possibility of failure? 15. Are you easily discouraged in difficult situations? 17. Are you a shy person? 19. Whenever possible, do you avoid demonstrating your skills for fear of being embarrassed? 21. When you are with a group, do you have difficulties selecting a good topic to talk about? 23. Is it often difficult for you to fall asleep when you think about things you have done or must do?* 25. Do you think a lot before complaining in a restaurant if your meal is not well prepared? 27. Would you be bothered if you had to return to a store when you noticed you were given the wrong change?* 29. Whenever you can, do you avoid going to unknown places? 31. Are you often worried by things that you said or did? 33. Would it be difficult for you to ask your boss for a raise (salary increase)? 35. Do you generally try to avoid speaking in public? 37. Do you, on a regular basis, think that you could do more things if it was not for your insecurity or fear? 39. Comparing yourself to people you know, are you afraid of many things? 41. Do you often find yourself worrying about things to the extent that performance in intellectual abilities is impaired? 43. Do you often refrain from doing something you like in order not to be rejected or disapproved of by others? 45. Generally, do you pay more attention to threats than to pleasant events?* 47. Do you often refrain from doing something because of your fear of being embarrassed? Items from the Sensitivity to Reward Scale 2. Does the good prospect of obtaining money motivate you strongly to do some things? 4. Are you frequently encouraged to act by the possibility of being valued in your work, in your studies, with your friends or with your family?*

Factor 1 (SP)

Factor 2 (SR)

h2

0.27

)0.12

0.09

0.46

0.06

0.21

0.69 0.58 0.23

0.02 )0.11 0.11

0.48 0.34 0.06

0.11 0.44

0.26 0.13

0.08 0.21

0.60 0.57 0.64

0.06 )0.27 0.06

0.36 0.40 0.41

0.57

)0.01

0.32

0.25

0.11

0.07

0.35

0.05

0.12

0.28

0.14

0.10

0.40 0.62 0.53

)0.09 0.24 )0.02

0.17 0.44 0.28

0.56 0.76

)0.18 )0.01

0.35 0.58

0.60

)0.02

0.36

0.56

0.15

0.34

0.60

0.19

0.40

0.34

0.34

0.23

0.86

0.02

0.74

0.04

0.43

0.19

0.15

0.26

0.09

(continued on next page)

994

R.M. OÕConnor et al. / Personality and Individual Differences 37 (2004) 985–1002

Table 2 (continued) N ¼ 603

Factor 1 (SP)

Factor 2 (SR) h2

6. Do you often meet people that you find physically attractive?* 8. Do you like to take some drugs because of the pleasure you get from them?* 10. Do you often do things to be praised? 12. Do you like being the center of attention at a party or a social meeting? 14. Do you spend a lot of your time on obtaining a good image? 16. Do you need people to show their affection for you all the time?* 18. When you are in a group, do you try to make your opinions the most intelligent or the funniest? 20. Do you often take the opportunity to pick up people you find attractive? 22. As a child, did you do a lot of things to get peopleÕs approval? 24. Does the possibility of social advancement move you to action, even if this involves not playing fair? 26. Do you generally give preference to those activities that imply an immediate gain? 28. Do you often have trouble resisting the temptation of doing forbidden things? 30. Do you like to compete and do everything you can to win? 32. Is it easy for you to associate tastes and smells to very pleasant events?* 34. Are there a large number of objects or sensations that remind you of pleasant events?* 36. When you start to play with a slot machine, is it often difficult for you to stop?* 38. Do you sometimes do things for quick gains? 40. Does your attention easily stray from your work in the presence of an attractive stranger? 42. Are you interested in money to the point of being able to do risky jobs? 44. Do you like to put competitive ingredients in all of your activities? 46. Would you like to be a socially powerful person? 48. Do you like displaying your physical abilities even though this may involve danger?

)0.12 )0.02

0.28 0.30

0.09 0.09

0.22 )0.12

0.44 0.51

0.24 0.27

0.28 0.38 0.02

0.37 0.37 0.55

0.22 0.28 0.30

)0.17

0.47

0.25

0.25 0.17

0.45 0.56

0.26 0.34

0.13

0.46

0.23

0.09

0.42

0.18

)0.16 0.05

0.50 0.21

0.28 0.05

)0.02

0.18

0.03

0.17

0.16

0.05

0.08 0.28

0.60 0.38

0.37 0.22

)0.05

0.64

0.41

)0.09 0.10 )0.20

0.49 0.51 0.54

0.25 0.27 0.33

Note. * Items deleted from final model; h2 ¼ Communalities.

supporting the two-factor model (Note. Lowest standardized factor loading is 0.34 for item 25) (see Table 3). The fit indices (v2 ð204; N ¼ 603Þ ¼ 744:06, p < 0:001; v2 =df ¼ 3:65; CFI ¼ 0.80; TLI ¼ 0.83; RMSEA ¼ 0.07), however, suggested a poor fit of the model to the data. This may be attributable to the large number of items and possibly the large sample size. The v2 =df, which adjusts for sample size, however, is also indicative of poor fit. The percent of variance accounted for in the observed measures by the latent factors was generally higher than in previous models (R2 s range from 0.12 to 0.75). Although the factor covariance was significant, again the correlation was small (sxy ¼ 0:03, r ¼ 0:13; p < 0:05). The alpha reliability coefficient for both scales were acceptable and similar to those found with the full model (SP: a ¼ 0:83; SR: a ¼ 0:74).

R.M. OÕConnor et al. / Personality and Individual Differences 37 (2004) 985–1002

995

Table 3 Standardized factor loadings for the final measurement model (Samples 1–3)

Indicators of Sensitivity to Punishment 3. Do you prefer not to ask for something when you are not sure you will obtain it? 5. Are you often afraid of new or unexpected situations? 7. Is it difficult for you to telephone someone you do not know? 13. In tasks that you are not prepared for, do you attach great importance to the possibility of failure? 15. Are you easily discouraged in difficult situations? 17. Are you a shy person? 19. Whenever possible, do you avoid demonstrating your skills for fear of being embarrassed? 21. When you are with a group, do you have difficulties selecting a good topic to talk about? 25. Do you think a lot before complaining in a restaurant if your meal is not well prepared? 29. Whenever you can, do you avoid going to unknown places? 31. Are you often worried by things that you said or did? 33. Would it be difficult for you to ask your boss for a raise (salary increase)? 35. Do you generally try to avoid speaking in public? 37. Do you, on a regular basis, think that you could do more things if it was not for your insecurity or fear? 39. Comparing yourself to people you know, are you afraid of many things? 41. Do you often find yourself worrying about things to the extent that performance in intellectual abilities is impaired? 43. Do you often refrain from doing something you like in order not to be rejected or disapproved of by others?

Sample 1 ðN ¼ 603Þ

Sample 2 ðN ¼ 104Þ

Sample 3 ðN ¼ 160Þ

Factor loadings

Factor loadings

Residual variance

Factor loadings

Residual variance

Latent Factor 0.48 0.77

0.52

0.73

0.45

0.80

0.68

0.53

0.74

0.46

0.61

0.63

0.58

0.66

0.80

0.36

0.68

0.54

0.45

0.80

0.66

0.56

0.45

0.80

0.61

0.63

0.68

0.53

0.67

0.55

0.59 0.67

0.65 0.55

0.69 0.64

0.53 0.59

0.58 0.75

0.66 0.44

0.57

0.67

0.50

0.75

0.66

0.56

0.34

0.88

0.48

0.77

0.09*

0.99

0.40

0.84

0.62

0.61

0.44

0.80

0.63

0.60

0.67

0.55

0.60

0.64

0.53

0.72

0.60

0.64

0.60

0.64

0.58

0.66

0.69

0.52

0.54

0.71

0.77

0.40

0.82

0.33

0.73

0.47

0.58

0.66

0.62

0.62

0.65

0.58

0.56

0.69

0.64

0.59

0.59

0.65

0.60

0.64

0.68

0.54

0.77

0.41

Residual variance

(continued on next page)

996

R.M. OÕConnor et al. / Personality and Individual Differences 37 (2004) 985–1002

Table 3 (continued)

47. Do you often refrain from doing something because of your fear of being embarrassed?

Sample 1 ðN ¼ 603Þ

Sample 2 ðN ¼ 104Þ

Sample 3 ðN ¼ 160Þ

Factor loadings

Residual variance

Factor loadings

Residual variance

Factor loadings

Residual variance

0.87

0.25

0.81

0.34

0.89

0.20

0.82

0.80

0.36

0.52

0.73

0.76

0.48

0.77

0.50

0.75

0.79

0.50

0.75

0.52

0.73

0.82

0.35

0.88

0.56

0.68

0.70

0.53

0.72

0.48

0.77

0.87

0.39

0.85

0.32

0.90

0.75

0.38

0.86

0.41

0.83

0.64

0.45

0.80

0.54

0.70

0.73

0.64

0.59

0.68

0.54

0.86

0.40

0.84

0.42

0.83

0.70

0.64

0.59

0.54

0.70

0.55

0.70

0.50

0.58

0.67

0.84

0.70

0.51

0.39

0.85

0.60

0.74

0.45

0.40

0.84

0.72

0.51

0.74

0.53

0.72

Indicators of Sensitivity to Reward Latent Factor 2. Does the good prospect of 0.43 obtaining money motivate you strongly to do some things? 10. Do you often do things to be 0.49 praised? 0.46 12. Do you like being the center of attention at a party or a social meeting? 14. Do you spend a lot of your time 0.42 on obtaining a good image? 0.55 18. When you are in a group, do you try to make your opinions the most intelligent or the funniest? 0.37 20. Do you often take the opportunity to pick up people you find attractive? 22. As a child, did you do a lot of 0.50 things to get peopleÕs approval? 0.60 24. Does the possibility of social advancement move you to action, even if this involves not playing fair? 26. Do you generally give preference 0.54 to those activities that imply an immediate gain? 28. Do you often have trouble 0.38 resisting the temptation of doing forbidden things? 30. Do you like to compete and do 0.54 everything you can to win? 38. Do you sometimes do things for 0.67 quick gains? 40. Does your attention easily stray 0.40 from your work in the presence of an attractive stranger? 42. Are you interested in money to 0.63 the point of being able to do risky jobs? 44. Do you like to put competitive 0.53 ingredients in all of your activities?

R.M. OÕConnor et al. / Personality and Individual Differences 37 (2004) 985–1002

997

Table 3 (continued)

46. Would you like to be a socially powerful person? 48. Do you like displaying your physical abilities even though this may involve danger?

Sample 1 ðN ¼ 603Þ

Sample 2 ðN ¼ 104Þ

Sample 3 ðN ¼ 160Þ

Factor loadings

Residual variance

Factor loadings

Residual variance

Factor loadings

Residual variance

0.51

0.74

0.56

0.68

0.51

0.74

0.53

0.72

0.44

0.80

0.51

0.74

Note. All factor loadings significant at p < 0:05 except for that indicated with an asterisk (*).

Post hoc probing of the model was done to determine whether freeing error covariances would improve the model fit. Error covariances were freed iteratively, such that in each successive model the covariance that produced the largest standardized residual was freed. This continued until all standardized residuals with a value of 2.50 or greater had been freed. Given the very minimal improvement in the model (v2 ð203; N ¼ 603Þ ¼ 597:58, p < 0:001; v2 =df ¼ 2:94; CFI ¼ 0.85; TLI ¼ 0.88; RMSEA ¼ 0.06) with 11 error covariances freed, and based on the lack of a priori hypotheses, the error covariances were not retained in the final model. To confirm the fit of the final trimmed measurement model, the model was run using the SPSRQ data from Samples 2 and 3. 3.4.2. CFA with Sample 2. Based on Hu and BentlerÕs (1999) criteria for fit indexes, the two-factor model provided a poor fit to the data, however, the v2 =df test suggests support for the model (v2 ð70; N ¼ 104Þ ¼ 120:22, p < 0:001; v2 =df ¼ 1:72; CFI ¼ 0.82; TLI ¼ 0.83; RMSEA ¼ 0.08). Furthermore, as shown in Table 3, all factor loadings were substantial and all loadings were significant ðp < 0:05Þ, suggesting adequate fit to the data. The percent of variance accounted for in the observed measures by the latent factors ranged from 12% to 67%. The covariance between the factors was not significant (sxy ¼ 0:03, r ¼ 0:07; p > 0:05). 3.4.3. CFA with Sample 3. The two-factor model provided a better fit to the data than in Samples 1 and 2 (v2 ð100; N ¼ 160Þ ¼ 137:569, p < 0:01; v2 =df ¼ 1:38; CFI ¼ 0.91; TLI ¼ 0.91; RMSEA ¼ 0.05), and again indices of fit suggested mixed support for the model. In addition, as in Samples 1 and 2, the factor loadings suggested adequate fit (see Table 3). Almost all factor loadings were substantial and significant ðp < 0:05Þ. One exception was the item ‘‘Do you think a lot before complaining in a restaurant if your meal is not well prepared?’’ which loaded weakly on the SP factor (standardized factor loading ¼ 0.09). The SP latent factor accounted for little variance in this item ðR2 ¼ 0:01Þ, relative to the other items. This item was the weakest loading item in the final model run with Sample 1 (standardized factor loading ¼ 0.34); thereby, suggesting that it may be a problematic item. The percent of variance accounted for in the observed measures (excluding the problematic SP item cited above), by the latent factors ranged from 11% to 80%. The covariance between the factors was not significant (sxy ¼ 0:00, r ¼ 0:01; p > 0:05). 3.4.4. Validating the final model. Using the data from Sample 3, new SP and SR scales were created based on the reduced scales. Associations between these new SP and SR scales and the

998

R.M. OÕConnor et al. / Personality and Individual Differences 37 (2004) 985–1002

Positive and Negative Affect scales (Watson et al., 1988) and the BIS/BAS scales (Carver & White, 1994) were examined. The SP scale was hypothesized to correlate positively with the BIS scale and the NA scale, and the SR scale was hypothesized to correlate positively with the three BAS scales (Drive, Fun Seeking, and Reward Responsiveness) and the PA scale. The correlation matrix is presented in Table 4. Included in the matrix are the original SP (SPorig) and SR (SRorig) scales. The new and old SP and SR scales did not appear to differ in terms of correlations with these other measures. The SR scales significantly and positively correlated as expected with Carver and WhiteÕs three BAS scales (Reward Responsiveness, Drive, and Fun Seeking), and with Watson et al.Õs PA Scale. SR was unrelated to NA and the BIS scale. The original and new SP scales were similar in that they both correlated positively and significantly with the BIS scale and NA scale and significantly negatively correlated with the PA scale, and the BAS Fun Seeking scale. These correlations were in the expected direction and support the validity of the SP and SR scales.

4. Discussion The SPSRQ developed by Torrubia et al. (1995) is proposed as a self-report measure of BIS and BAS, the two motivational systems described by Gray (1975, 1982, 1987a, 1987b). The authors identify two orthogonal scales, Sensitivity to Punishment and Sensitivity to Reward, which corresponded to the BIS and BAS, respectively. Previous interpretations of an exploratory factor analysis of this measure generally supported a two-factor solution, however, the data presented from the analysis suggested some problems with the measure (Torrubia et al., 2001). The purpose of the present study was to further assess the underlying factor solution of the English version of the SPSRQ in three samples of college students. Notably, the samples were relatively heterogeneous with respect to ethnicity. The relative heterogeneity of the sample is considered a strength of the current study, as the validity of the scale is not only assessed across different languages, but also across different ethnicities. Demonstrating consistent strengths and weaknesses of the structure of the measure across samples has the potential to inform how the measure might be improved to maximize generalizability. The exploratory factor analysis run in this study replicated previous results (Torrubia et al., 2001) in that several poorly loading items were apparent. A CFA of the proposed measurement model also suggested a problem with the two-factor model using the 48 items. Once problematic items were trimmed from the model, a final measurement model with 18 SP and 17 SR items suggested mixed support for the two-factor model. Some fit indices suggested poor fit, however, v2 =df suggested adequate fit across two of the samples and the pattern of factor loadings suggested adequate fit across all three samples. Some researchers have argued that CFA and associated fit indices are not useful for evaluating the factor structure of models with a large number of items. According to this view, one must rely on other means of assessing the adequacy of the model, such as size of factor loadings and R2 . In the current study, these criteria suggest that the two-factor solution is adequate for the reduced pool of items. The validity of the final trimmed measurement model was supported, however, there did not appear to be any increase in the validity of the new, reduced scales over the original SPSRQ scales. In sum, our attempts to improve the factor structure of the SPSRQ resulted in some improvements. It should be noted that the items excluded from the final model did not appear to

N ¼ 160

SP

SR

SPorig

SRorig

PA

NA

BIS

Drive

Fun

Rew.

BAS

SP SR SPorig SRorig PA NA BIS Drive Fun Rew. BAS

1.00 0.01 0.98 0.03 )0.29 0.47 0.54 )0.04 )0.16 0.01 )0.08

1.00 0.04 0.96 0.20 )0.01 )0.10 0.32 0.22 0.17 0.30

1.00 0.05 )0.25 0.49 0.53 )0.03 )0.18 )0.01 )0.09

1.00 0.18 0.02 )0.05 0.34 0.24 0.20 0.33

1.00 )0.22 )0.13 0.18 0.21 0.36 0.31

1.00 0.35 0.05 )0.07 )0.00 )0.01

1.00 0.11 0.02 0.25 0.16

1.00 0.49 0.38 0.79

1.00 0.53 0.83

1.00 0.79

1.00

4.44 7.52

3.54 9.47

5.25 10.06

4.19 13.63

0.66 3.39

0.68 2.30

0.54 2.75

0.58 2.78

0.53 3.05

0.44 3.32

0.41 3.07

SD M

Note. SP ¼ Sensitivity to Punishment scale based on final model solution; SR ¼ Sensitivity to Reward scale based on final model solution; SPorig ¼ Sensitivity to Punishment scale with original 24 items; SRorig ¼ Sensitivity to Reward scale with original 24 items; PA ¼ Positive Affect scale from PANAS; NA ¼ Negative Affect scale from PANAS; BIS ¼ BIS scale from BIS/BAS scale; Drive ¼ BAS Drive scale from BIS/BAS scale; Fun ¼ BAS Fun Seeking scale from BIS/BAS scale, Rew. ¼ BAS Reward Responsiveness scale from BIS/BAS scale; BAS ¼ Composite of BAS Drive, BAS Fun Seeking, and BAS Reward Responsiveness. * p < 0:05; ** p < 0:01.

R.M. OÕConnor et al. / Personality and Individual Differences 37 (2004) 985–1002

Table 4 SP and SR Correlated (Pearson correlations) with Watson et al.Õs (1988) PANAS and Carver and WhiteÕs (1994) BIS/BAS scale (Sample 3)

999

1000

R.M. OÕConnor et al. / Personality and Individual Differences 37 (2004) 985–1002

have any underlying common theme. Moreover, post hoc exploratory factor analysis run with the deleted items did not support a meaningful factor structure. In conclusion, Torrubia et al. (1995, 2001) approached the development of a self-report scale of GrayÕs BAS and BIS in accordance with GrayÕs theoretical model, and they have been successful at doing so. However, psychometrically the original SPSRQ has some limitations, as evident in Torrubia et al.Õs (2001) findings, and as shown in the present study, using an English version and ethnically diverse population. Our results in three independent samples suggest caution should be used when implementing the SPSRQ. Some improvement in the factor structure is gained by reducing the item pool without sacrificing the validity of the scales. In future research it might be useful to write additional items or reword current items to improve the factor structure of this measure. Improving the factor structure will likely lead to further improvements in validity. Perusal of the items suggests that some include complicated sentences that might be confusing for some participants. For another study we are administering the SPSRQ by reading the items aloud to participants and initial observations suggest that some items are confusing. Rewording some of the more complicated items might be particularly useful. It is noteworthy that the SPSRQ administered in the current study was an English translation of the original non-English version. The psychometric data presented by Torrubia et al. (2001) was based on a non-English version (Catalan) of the SPSRQ. It is possible that poor fit of our CFAs with the original 48 items is attributable to awkward wording of items on the English version, as well as, cultural differences. However, consistencies in the factor structure (i.e., similar problematic items) are evident across data collected using the English version, which was administered with an ethnically heterogeneous sample, and using the non-English version, as evident by the findings presented here and those presented in Torrubia et al. (2001). Thus, problematic items do not appear to be attributable to our use of an English version or to cultural differences. Although the SPSRQ would benefit from further refinement, it appears to be a promising measure, which has a strong theoretical application as a measure of GrayÕs BIS and BAS. Future research should validate the English version in a wider age range. Changing the wording of some awkward items, and perhaps considering new items to replace the ones we identified as problematic are also important directions for future research.

References Arbuckle, J. L. (1996). Full information estimation in the presence of incomplete data. In G. A. Marcoulides & R. E. Schumaker (Eds.), Advanced structural equation modeling: Issues and techniques (pp. 243–277). Mahwah, NJ: Lawrence Erlbaum Associates.  Avila, C. (2001). Distinguishing BIS-mediated and BAS-mediated disinhibition mechanisms: A comparison of disinhibition models of Gray (1981, 1987a, 1987b) and of Patterson and Newman (1993). Journal of Personality and Social Psychology, 80(2), 311–324. Brebner, J. M., & Martin, M. (1995). Testing for stress and happiness: The role of personality factors. In C. D. Spielberger, I. G. Sarason, J. Brebner, E. Greenglass, P. Laungani, & A. M. OÕRoark (Eds.), Stress and emotion: Anxiety, anger, and curiosity (Vol. 15, pp. 139–172). Washington, DC: Taylor and Francis. Byrne, B. M. (2001). Structural equation modeling with AMOS: Basic concepts, applications, and programming. Mahwah, NJ: Lawrence Erlbaum Assoicates.

R.M. OÕConnor et al. / Personality and Individual Differences 37 (2004) 985–1002

1001

Carver, S. C., & White, T. L. (1994). Behavioral inhibition, behavioral activation, and affective responses to impending reward and punishment: The BIS/BAS scales. Journal of Personality and Social Psychology, 67, 319–333.  Caseras, X., Avila, C., & Torrubia, R. (2003). The measurement of individual differences in behavioural inhibition and behavioural activation systems: a comparison of personality scales. Journal of Personality and Individual Differences, 34, 999–1013. Clark, L. A., & Watson, D. (1988). Mood and the mundane: Relations between daily life events and self-reported mood. Journal of Personality and Social Psychology, 54, 294–306. Clark, L. A., & Watson, D. (1991). Tripartite model of anxiety and depression: Psychometric evidence and taxonomic implications. Journal of Abnormal Psychology, 100, 316–336. Comrey, L. A., & Lee, H. B. (1992). A first course in factor analysis (2nd ed.). Hillsdale, NJ: Lawrence Erlbaum Associates, Publishers. Eysenck, S. B. G., & Eysenck, H. J. (1978). Impulsiveness and venturesomeness: Their position in a dimensional system of personality description. Psychological Reports, 43, 1247–1255. Gray, J. A. (1975). Elements of a two-process theory of learning. London, England: Academic Press. Gray, J. A. (1981). A critique of EysenckÕs theory of personality. In H. J. Eysenck (Ed.), A model for personality (pp. 246–276). New York: Springer. Gray, J. A. (1982). Precis of the neuropsychology of anxiety: An enquiry into the functions of the septo-hippocampal system. The Behavioral and Brain Sciences, 5, 469–534. Gray, J. A. (1987a). The neuropsychology of emotion and personality. In S. M. Stahl, S. D. Iversen, & E. C. Goodman (Eds.), Cognitive neurochemistry (pp. 171–190). Oxford: Oxford University Press. Gray, J. A. (1987b). The psychology of fear and stress (2nd ed.). Cambridge, England: Cambridge University Press. Harmon-Jones, E., & Allen, J. J. B. (1997). Behavioral activation sensitivity and resting frontal EEG asymmetry: Covariation of putative indicators related to risk for mood disorders. Journal of Abnormal Psychology, 106, 159– 163. Hawk, L. W., Jr., & Kowmas, A. D. (2003). Affective modulation and prepulse inhibition of startle among undergraduates high and low in behavioral inhibition and approach. Journal of Psychophysiology, 40, 131–138. Heubeck, B. G., Wilkinson, R. B., & Cologon, J. (1998). A second look at Carver and WhiteÕs (1994) BIS/BAS scales. Personality and Individual Differences, 25, 785–800. Hu, L., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling, 6, 1–55. Jorm, A. F., Christensen, H., Henderson, A. S., Jacomb, P. A., Korten, A. E., & Rodgers, B. (1999). Using the BIS/ BAS scales to measure behavioural activation: factor structure, validity and norms in a large community sample. Personality and Individual Differences, 26, 49–58. Kline, R. B. (1998). Principles and practice of structural equation modeling. NY: The Guilford Press. Muthen, L. K., & Muthen, B. (2002). Mplus: The comprehensive modeling program for applied researchers. Los Angeles, CA: Muthen and Muthen. Patterson, C. M., & Newman, J. P. (1993). Reflectivity and learning from aversive events: Toward a psychological mechanism for the syndromes of disinhibition. Psychological Review, 100, 716–736. Sutton, S. K., & Davidson, R. J. (1997). Prefrontal brain asymmetry: A biological substrate of the behavioral approach and inhibition systems. Psychological Science, 8, 204–210. Tabachnick, B., & Fidell, L. (2001). Using multivariate statistics (4th ed.). Boston, MA: Allyn and Bacon.  Torrubia, R., Avila, C., Molt o, J., & Caseras, X. (2001). The sensitivity to punishment and sensitivity to reward questionnaire (SPSRQ) as a measure of GrayÕs anxiety and impulsivity dimensions. Personality and Individual Differences, 31, 837–862.  Torrubia, R., Avila, C., Molt o, J., & Grande, I. (1995). Testing for stress and happiness: The role of the behavioral inhibition system. In C. D. Spielberg, I. G. Sarason, J. Brebner, E. Greenglass, P. Langani, & A. M. OÕRoark (Eds.), In stress and emotion: Anxiety, anger, and curiosity (Vol. 15, pp. 189–211). Washington, DC: Taylor and Francis. Watson, D. (1988a). Intraindividual and interindividual analyses of positive and negative affect: Their relation to health complaints, perceived stress, and daily activities. Journal of Personality and Social Psychology, 54, 1020– 1030.

1002

R.M. OÕConnor et al. / Personality and Individual Differences 37 (2004) 985–1002

Watson, D. (1988b). The vicissitudes of mood measurement: Effects of varying descriptors, time frames, and response formats on measures of positive and negative affect. Journal of Personality and Social Psychology, 55, 128– 141. Watson, D., Clark, L. A., & Tellegen, A. (1988). Development and validation of brief measures of positive and negative affect: The PANAS scales. Journal of Personality and Social Psychology, 54, 1063–1070.