For example, the original item shown in Table 1 became the following separate items: (a) my job evaluations in the future will be affected by the same reason that caused this negative evaluation, and (b) the reason for this negative evaluation will not impact on my future job evaluations.

The negative consequences item (the likelihood that other negative things would result) was maintained as a single item in the adapted version of the CSQ. As shown in Supplementary Material: Appendix 1, for each scenario, participants rated cognitive style in terms of internality (items 1 and 6), globality (items 2 and 7), stability (items 3 and 8), negative consequences (item 4), and self-worth implications (items 5 and 9). All items were rated using the same 5-point Likert scale ranging from ‘strongly agree’ to ‘strongly disagree’. Items were scored so that higher selleckchem scores indicated more negative cognitive style. The third modification involved removing the positive scenarios, thus halving the length of

the instrument. Our rationale was that depression is more strongly related to inferences for negative scenarios than those for positive scenarios (Alloy et al., 2000 and Alloy et al., 2006). Indeed, an ad hoc strategy of presenting only the negative scenarios has already been employed in some studies (e.g., Gibb, Alloy, Abramson, Beevers, & Miller, 2004). However, omitting the positive items from the CSQ in the absence of any further adaptations is potentially problematic. Haeffel et al. (2008) identified two reasons for the original inclusion of positive items in the CSQ: (a) to assess the individual’s selleck chemicals “enhancing inferential

style… the tendency to make stable, global attributions and infer positive consequences and self-worth characteristics for positive (rather than negative) life events” (p. 826), and (b) to reduce the chances of a response set bias. While omission of positive items is unlikely to be problematic if negative cognitive style is the focus of research, response bias remains a potential threat to reliability and validity. Allowing all items to be rated on the same Likert scale enabled us to reduce the probability of response set bias by including reverse-worded items ( Cronbach, 1970). Thus, to indicate negative cognitive style consistently, GNA12 participants would have to agree with some items but disagree with others. Supplementary Material: Appendix 1 shows how reverse-worded items were included to rate a scenario. The final adaptation was to include the original practice scenario (“you and your parents are not getting along well”) as an additional test scenario in order to broaden the scope of social relationships focused upon. There were thus 13 scenarios (the practice scenario and 12 test scenarios from the original CSQ) in the first iteration of our revised CSQ (the CSQ-13), which had nine response items for each scenario.