The study was approved by the local ethics committee of the Department of Psychology of the University of Koblenz-Landau (reference number LEK-303) and has been conducted in accordance with ethical standards as laid down in the 1964 Declaration of Helsinki and its later amendments. The study design and the analysis plan were pre-registered on aspredicted.org: https://aspredicted.org/26B_T73.
Participants and Procedure
Participants were recruited via email lists and social networks. The inclusion criteria were: age ≥ 18 years and sufficient German language skills. The study was designed as an online longitudinal observational study with two measurement time points. Data collection at T1 took place between March 30, 2021 and April 14, 2021. After eight weeks, participants were contacted via email and were asked to complete the second assessment (T2), which took place between May 25, 2021 and June 14, 2021. Both at T1 and T2, the order of the questionnaires was randomised to prevent order effects. At baseline, 258 participants completed the questionnaires through the survey platform www.soscisurvey.de. Of these, 158 participants completed the follow-up survey eight weeks later. One person had to be excluded because the person’s age was below 18 years. As pre-registered, we only used data of people who completed both T1 and T2. Thus, all analyses are based on a sample of N = 157. As an incentive for participation, participants had the chance to take part in a lottery to win gift vouchers (20€).
The mean age of the participants was M = 26.84 years, SD = 10.07, and 73.2% of the participants were female (25.5% male, 0.6% diverse, and 0.6% not willing to provide data on gender). The majority of the sample (66.9%) had school leaving examination (“Abitur” in German) as the highest educational degree, 25.5% of the participants had a university degree, 7% had primary education (German “Hauptschulabschluss” or “Mittlere Reife”), and 0.6% had a PhD as the highest educational degree. Most participants were university students (62.4%), 23.6% were employed, 4.5% worked in civil service, 1.3% were self-employed, 3.8% were in training, 0.6% were unemployed, 1.3% were homemakers, and 0.6% were pensioners. The BDI II sum scores were M = 11.59, SD = 9.43, at baseline and M = 10.35, SD = 9.27, at follow-up, each reflecting minimal symptoms of depression (Beck et al., 1996).
Situational expectations. Situational expectations were assessed using the Depressive Expectations Scale (DES) developed by Kube et al. (2017). This 25-item scale was developed to measure depression-specific expectations with a high level of situational specificity. The DES assesses expectations concerning interpersonal situations, performance-related situations, and negative mood regulation. Each item is formulated in an “if-then” structure, such that the items of the DES reflect situation-specific predictions that could be tested empirically, for instance through a behavioural experiment. Fifteen items of the DES reflect negative expectations, in the sense that a negative outcome is anticipated (e.g., “If I ask someone for help, I will be rejected”). By contrast, 10 items represent positive expectations, meaning that a positive outcome is anticipated (e.g., “If I talk to someone about my problems, I will feel better afterwards”). All items are rated on a five-point Likert scale ranging from 1 (“I disagree”) to 5 (“I agree”). If researchers wish to use the DES total score, the items reflecting positive expectations have to be reversely scored. For the present analysis, we consideres positive and negative expectations as two separate subscales to examine their differential effects on depressive symptoms. In doing so, we report the original (not reversely scored) values of the positive expectations, in order to allow a more intuitive interpretation. That is, high values of the positive expectations subscale reflect positive expectations and high values in the negative expectations subscale reflect negative expectations. Participants completed the DES both at baseline and at follow-up.
In previous studies, the DES total score has shown excellent internal consistency as indicated by Cronbach’s α (Kube et al., 2017; Kube, Glombiewski, et al., 2018; Kube, Herzog, et al., 2019; Kube, Siebers, et al., 2018b). In the present sample, the internal consistency for the entire 25-item DES was Cronbach’s α = .86 at both T1 and T2. Separately for each subscale, the internal consistency of the positive expectations subscale was Cronbach’s α = .61 at T1 and Cronbach’s α = .73 at T2, and, respectively, Cronbach’s α = .84 at T1 and Cronbach’s α = .83 at T2 for the negative expectations subscale. The test-retest reliability with a time interval of eight weeks was r = .808 (p < .001) for the DES total score, r = .718 (p < .001) for the positive expectations subscale, and r = .751 (p < .001) for the negative expectations subscale.
Dispositional optimism. Dispositional optimism was assessed at baseline with the German version of the Life Orientation Test Revised (LOT-R, Scheier, Carver, & Bridges, 1994) by Glaesmer et al. (2011). The LOT-R is a 10-item self-report scale, of which four items are distractor items that are excluded when computing the sum scores. The items are rated on a five-point Likert scale, with higher values of the LOT-R total score indicating more optimistic outcome expectations. The LOT-R has shown good reliability and validity in previous research (Glaesmer et al., 2011; Reilley et al., 2005; Scheier et al., 1994). In the present study, internal consistency of the LOT-R was Cronbach’s α = .80.
Dysfunctional attitudes. Dysfunctional attitudes were assessed at baseline using the Dysfunctional Attitudes Scale (DAS), originally developed by Weissman and Beck (1978) and translated into German by Hautzinger, Joorman, and Keller (2005). Here, we used the 26-item version of the DAS, which comprises two factors: approval by others and performance evaluation. The DAS has shown good reliability and validity (Joormann, 2004; Nelson et al., 1992). Participants rated the items of the DAS using a five-point Likert scale, with higher values reflecting a greater endorsement of dysfunctional attitudes. Cronbach’s alpha in the current sample was Cronbach’s α = .91.
Depressive symptoms. Depressive symptoms were assessed both at baseline and at follow-up with the Beck Depression Inventory-II (Beck et al., 1996). This well-established 21-item scale assesses somatic, cognitive and affective symptoms of depression (sum scores ranging from 0 to 63) with higher scores reflecting more severe symptoms of depression. The BDI-II has shown good psychometric properties (Beck et al., 1996). In the present study, Cronbach’s alpha of the BDI-II was Cronbach’s α = .91 at baseline and Cronbach’s α = .92 at follow-up.
Data screening was conducted according to the recommendations by Tabachnick and Fidell (2019). We first performed a Fisher-z test to examine whether the magnitudes of the correlations of positive and negative expectations with depressive symptoms differ. For the main analyses, we performed a hierarchical linear regression analysis with the BDI-II sum scores at T2 as the criterion. In the first block, we included BDI-II sum scores at baseline as the predictor. In the second block, we added the sum scores of the positive and negative expectations subscales of the DES as predictors. In the third block, we entered the sum scores of the DAS and the LOT-R as additional predictors. In order to test the mediation hypotheses, BDI-II sum scores at T2 were the dependent variable (Y). In the first mediational model, the LOT-R sum score was the independent variable (X), and in the second mediation model, the DAS sum scores were included as the independent variable. In both mediation models, negative expectations (M1) and positive expectations (M2) were the mediator variables. Using model 6 from the PROCESS macro for SPSS v4.0 by Andrew Hayes (Hayes, 2022), we examined the potential mediating effects of negative and positive expectations. In order to test the statistical significance of the indirect effects, we used bias-corrected accelerated (BCa) bootstrapping confidence intervals (CI) with 10,000 bootstrapping samples. This bias-corrected bootstrapping procedure has been recommended by several authors for testing mediation effects (Cheung & Lau, 2008; Hayes, 2009; MacKinnon et al., 2004). To facilitate the comparison of the results across different studies, we provide standardised coefficients according to Preacher and Hayes (2008). For all analyses, IBM SPSS Statistics Version 27 was used.