A comprehensive meta-analysis of the self-serving bias in schizophrenia spectrum disorders compared to non-clinical subjects

Highlights

• Comprehensive meta-analysis of self-serving bias in schizophrenia spectrum disorders.

• Greater self-serving bias in schizophrenia spectrum disorders than in controls.

• Self-serving bias seems specific to persecutory delusions and paranoid ideation.

• The effect is a function of the method of measurement and cultural background.

• Self-esteem, depression, and jump to conclusion bias are influencing factors.

Abstract

We conducted a comprehensive meta-analysis to investigate the self-serving bias SSB in schizophrenia spectrum disorders (SSD) to examine its specificity for persecutory delusions, paranoia, and to explore sources of heterogeneity in previous findings. We included 56 studies with n = 2501 patients with SSD and n = 2601 non-clinical controls in the main random-effects model using Hedges’ g. Data quality and risk of bias were assessed. Results showed that SSB was a function of the method of measurement, and the cultural background of the participants, thereby identifying important sources of heterogeneity in measuring SSB in patients with SSD. The SSD group was somewhat more prone to SSB (g = 0.17), especially, when in a state of active persecutory delusion (g = 0.32/0.57), when they scored high in paranoid ideation (r = 0.25), or tended to jump to conclusions (r = 0.43). High self-esteem in the SSD group was associated with high SSB (r = 0.34); depression with low SSB (r=-0.39).

Introduction

Converging evidence suggests that social cognition and cognitive biases are important for the formation and maintenance of symptoms in patients with psychosis (Freeman, 2007; Halverson et al., 2019; Pinkham et al., 2014). Thus, psychological interventions have been tailored to target cognitive biases to reduce positive symptoms (Moritz et al., 2014) and foster social cognition (Grant et al., 2017) in patients with psychosis and persons at a clinical high-risk for psychosis (van der Gaag et al., 2012).

Among the cognitive biases, a self-serving attributional style has also been suggested to be particularly relevant in the research of psychotic symptoms and persecutory delusions in particular (Bentall et al., 2001, 1994). Heider (1958), an originator of the attributional theory noted that people, in general, are more likely to attribute positive events to themselves, however, negative events, are more likely attributed to other external causes (Heider, 2013). For instance, a student who fails to pass an exam may attribute the causality for his lack of success to the professor’s poor didactic skills or the extraordinary difficult test questions (both external attributions). However, passing an exam the student is likely to attribute his good performance to his high IQ, or efforts (both internal attributions) (Greenwald, 1980). This attributional pattern has been termed the self-serving bias or within the literature on psychosis, the externalizing bias (Kinderman and Bentall, 1996). The self-serving bias refers to the tendency to attribute success to oneself, and failure to external factors (Mezulis, Abramson et al., 2004). Typically, the self-serving bias is derived from attributional measures, such as the Attributional Style Questionnaire (Peterson et al., 1982). In these measures, participants are asked to imagine themselves in a series of hypothetical situations. Half of these events have a positive outcome (for example, becoming very rich); the other half have a negative outcome (for example, being fired from a job). The participants should provide the major cause of each hypothetical situation if it happened to them. Possible answers include internal causal explanations, or external causal explanations, which should be rated by the participants on a scale from “Is the cause of this due to something about you or something about other people?”

A self-serving attributional style has been suggested to be particularly relevant for the development of psychotic symptoms, especially persecutory delusion (Bentall et al., 2001, 1994). This assumption is termed delusion as defense theory. The delusion as defense theory holds that blaming others for negative outcomes is intended to protect patients with persecutory delusion from low self-esteem. However, the tendency to attribute negative events to others may come at the expense of a distorted perception of social environment to a hostile intention of others towards their person, which may foster paranoid ideation. The initial studies designed to test the delusion as defense theory supported the theory by showing a pronounced self-serving attributional style in patients with persecutory delusions when compared to non-clinical controls and depressed patients (Fear et al., 1996; Kaney and Bentall, 1989). However, more recent studies, which tried to replicate this finding, failed to do so (Diez‐Alegría et al., 2006; Mehl et al., 2014). The new evidence led to a reformulation of the delusion as defense theory. The revised form postulated that self-esteem should be more unstable in persons with paranoia and that there should be a significant difference between a lower implicit and higher explicit self-esteem in these patients (Bentall et al., 2001). The findings that patients with persecutory delusions would show an excessive self-serving attributional style were challenged by narrative reviews, which included up to 44 individual studies (Garety and Freeman, 2013a).

In a recently published random-effects meta-analysis, Murphy et al. aimed at testing both versions of the delusion as defense theory (2018). Their analysis had a broad scope and included the SSB in persons with persecutory delusions, among other outcomes (explicit self-esteem, implicit self-esteem, the discrepancy between implicit and explicit self-esteem, and self-esteem instability). Murphy et al. (2018) included studies in their analysis if at least 50 % of the sample had current persecutory delusions. Results from a subgroup set (k = 27; N = 1128) showed a greater SSB in samples with persecutory delusions compared to non-clinical controls (g = 0.48, 95 % CI 0.23−0.73), individuals with depression (g = 1.06, 95 % CI 0.48–1.63), and patients with psychosis without persecutory delusions (g = 0.40, 95 % CI 0.12−0.68). Moreover, a small correlation between the SSB and paranoia severity (r = 0.18, 95 % CI 0.08−0.27) was detected. There was large heterogeneity in the meta-analytic estimate of the SSB, pointing to unknown moderators that may have influenced the results. To explore sources of heterogeneity in their data Murphy et al. (2018) performed two moderator analysis (matching of the compared groups and depression). Both had no significant influence on the results. However, the existing literature in the field of attribution research points to a considerably higher number of possible moderators (Mezulis et al., 2004). Thus, mixed evidence on the SSB reported in the literature and heterogeneity in the meta-analysis by Murphy et al. (2018) might have resulted from moderating variables such as different measurement methods and cultural backgrounds of the participants included in the studies (Western, especially US samples, typically show a higher SSB than Asian samples; Mezulis et al., 2004). Moreover, other moderating variables such as socio-demography, IQ, and self-esteem should also be controlled for, as these covariates are highly likely to be correlated with the SSB (Mezulis et al., 2004; Sweeney et al., 1986). Thus, it can be concluded that methodological aspects and moderating variables that influence the SSB in persons with schizophrenia spectrum disorder have so far been largely neglected by reviews and meta-analysis.

Taking in account the evidence in the field, it is surprising to us that Murphy et al. (2018) did not find an influence of depression on SSB in patients with persecutory delusions, as depression is known to change the attribution style from a self-serving to a self-blaming attribution style (the self-blaming attribution style is characterized by the tendency to attribute negative events to one's person, whereas positive events are attributed to external causes; Sweeney et al., 1986). Thus, we suggest exploring the association between depression and SSB in people with psychosis using a different methodological approach. The same applies to the association between paranoia and SSB. Here Murphy et al. (2018) found a small correlation. The question is whether this effect can be replicated using another approach.

The part of the analysis by Murphy et al. (2018) relevant to the SSB included 27 studies. The inclusion criterion of at least 50 % of participants with current persecutory delusions may have led to the exclusion of a significant number of studies from the meta-analysis. However, narrative reviews by Garety and Freeman (2013a) and by Bentall et al. (2008) indicate that there are more studies available. In addition, there is evidence from the literature that the SSB may not be specific to persecutory delusion in patients with schizophrenia, but that SSB also occurs in samples with a broader schizophrenia spectrum (So et al., 2015). Therefore, there is a need for a comprehensive meta-analysis that takes into account all available studies.

The present meta-analyses aims: (1) to determine the overall effect size of the SSB among people with schizophrenia spectrum disorders compared to non-clinical controls by considering all available data; (2) to examine if theSSB is specific to the active state of persecutory delusion and other paranoid conditions, or if the SSB is also linked to other psychotic symptoms; (3) to explore whether there is an effect of method of measurements and cultural background; (4) to explore the impact of different covariates such as age, sex, depression, IQ, paranoid-ideation, other psychotic symptoms, and self-esteem on SSB.