ABSTRACT
The tendency for individuals to interpret ambiguous information in a threatening way is theorised to maintain anxiety disorders. Recent findings suggest that positive and negative interpretation biases may have unique effects. This study tested the relationships between threat and benign biases with state and trait anxiety and quality of life, and whether individual differences moderate these relationships. N = 699 individuals with elevated trait anxiety symptoms completed web-based measures of interpretation bias, trait anxiety, state anxiety, and quality of life. Results demonstrated that threat interpretations predicted state anxiety, trait anxiety, and quality of life. Benign interpretations also predicted quality of life. However, benign interpretations only weakly (or not at all) predicted state and trait anxiety. Individual differences (e.g. gender, race, ethnicity, age) did not moderate findings. Results emphasise the need to consider benign and threat biases separately, both in cognitive models of anxiety and experimental designs.
Acknowledgments
This research is supported by NIMH grants R34MH106770 and R01MH113752 awarded to B. Teachman. The clinicaltrials.gov number for this study is NCT02382003. Thank you to Emily Holmes, Simon Blackwell, Bundy Mackintosh, Andrew Mathews and their research teams for sharing some of their training and testing materials.
Disclosure statement
No potential conflict of interest was reported by the authors.
Data availability statement
The data that support the findings of this study are openly available via the Open Science Foundation at https://osf.io/a3v9g/.
Notes
1 Note that some online interpretation measures present the positive or negative word prior to the ambiguous text, instead of following the ambiguous text (e.g., Amir et al., Citation2012).
2 Note, the term bias is used throughout based on the interpretation bias measures’ established use as indicators of anxiety-linked selective or preferential processing (i.e., an individual difference to prefer one type of response over another), but there is no “ground truth” for an accurate response on these measures, so the term bias should not be interpreted to indicate accuracy.
3 Rather than conducting correlations and regressions as initially planned, we used SEM to increase parsimony and allow us to examine multiple inter-relationships simultaneously, create latent factors to reduce measurement error, and reduce the number of separate tests given the many indicators. Given the large number of participants needed for SEM, we did not split the sample to do an internal replication. We included imagery Prime as a covariate, rather than testing the role of Prime as a secondary analysis. Finally, in preregistration, we described the study as exploratory, but we include hypotheses that were determined post-registration, but prior to conducting analyses, in this manuscript.
4 Note that we asked participants to report their peak state anxiety experienced during the imagery task, rather than their current level of anxiety right after the imagery task.