Causal language and practice recommendations in observational clinical psychology research : a thesis presented in partial fulfilment of the requirements for the degree of Master of Science in Psychology at Massey University, Wellington, New Zealand

Abstract

Clinical psychology research often relies on observational designs due to constraints related to random assignments of participants. This means that observational studies are central to understanding relationships between complex variables in clinical psychology. Researchers are often motivated by causal theorising, and knowledge about causal effects is necessary for making inferences about practical implications. However, researchers often use linking language (terms and phrases that connect predictor and outcome variables) in ways that suggest causality, even when their observational designs do not support such inferences. Additionally, practice recommendations (actionable suggestions derived from study findings) frequently rely on these implied causal relationships, which raises concerns about the validity of these recommendations. This thesis investigates the use of linking language in clinical psychology literature, specifically within observational studies. It also assesses the extent to which practice recommendations rely on the assumption of causal relationships between predictor and outcome variables. Conceptually replicating the methods of prior research by Haber et al. (2022) in public health and epidemiology, we analysed 97 articles from leading clinical psychology journals. We conducted a quantitative content analysis to assess the strength of causal implications in both linking language and practice recommendations. We also evaluated the relationship between the strength of causal implications in the linking language and those in the practice recommendations. Our results revealed that while many linking sentences in observational clinical psychology articles implied causality, practice recommendations often carried even stronger causal implications. There was no statistically significant correlation between the causal implication strength of the linking sentences and that of the practice recommendations, indicating a disconnect between the two. Furthermore, researchers often provided practice recommendations that implied causal effects despite including disclaimers stating they could not make causal inferences. This indicates an implicit interest in causal inference that is not explicitly articulated in their papers. Researchers should be transparent about their interests in causality and ensure that their conclusions are supported by the evidence. Practitioners must critically evaluate findings before applying their practice recommendations, while policymakers should exercise caution when interpreting findings that may exaggerate causal relationships.