John Garry1, Rob Ford2, Rob Johns3. 1. Queen's University Belfast, Belfast, Northern Ireland. 2. University of Manchester, Manchester, UK. 3. University of Essex, Colchester, UK.
Abstract
BACKGROUND: Freeman et al. (2020a, Psychological Medicine, 21, 1-13) argue that there is widespread support for coronavirus conspiracy theories in England. We hypothesise that their estimates of prevalence are inflated due to a flawed research design. When asking respondents to their survey to agree or disagree with pro-conspiracy statements, they used a biased set of response options: four agree options and only one disagree option (and no 'don't know' option). We also hypothesise that due to these flawed measures, the Freeman et al. approach under-estimates the strength of the correlation between conspiracy beliefs and compliance. Finally, we hypothesise that, due to reliance on bivariate correlations, Freeman et al. over-estimate the causal connection between conspiracy beliefs and compliance. METHODS: In a pre-registered study, we conduct an experiment embedded in a survey of a representative sample of 2057 adults in England (fieldwork: 16-19 July 2020). RESULTS: Measured using our advocated 'best practice' approach (balanced response options, with a don't know option), prevalence of support for coronavirus conspiracies is only around five-eighths (62.3%) of that indicated by the Freeman et al. approach. We report mixed results on our correlation and causation hypotheses. CONCLUSIONS: To avoid over-estimating prevalence of support for coronavirus conspiracies, we advocate using a balanced rather than imbalanced set of response options, and including a don't know option.
BACKGROUND: Freeman et al. (2020a, Psychological Medicine, 21, 1-13) argue that there is widespread support for coronavirus conspiracy theories in England. We hypothesise that their estimates of prevalence are inflated due to a flawed research design. When asking respondents to their survey to agree or disagree with pro-conspiracy statements, they used a biased set of response options: four agree options and only one disagree option (and no 'don't know' option). We also hypothesise that due to these flawed measures, the Freeman et al. approach under-estimates the strength of the correlation between conspiracy beliefs and compliance. Finally, we hypothesise that, due to reliance on bivariate correlations, Freeman et al. over-estimate the causal connection between conspiracy beliefs and compliance. METHODS: In a pre-registered study, we conduct an experiment embedded in a survey of a representative sample of 2057 adults in England (fieldwork: 16-19 July 2020). RESULTS: Measured using our advocated 'best practice' approach (balanced response options, with a don't know option), prevalence of support for coronavirus conspiracies is only around five-eighths (62.3%) of that indicated by the Freeman et al. approach. We report mixed results on our correlation and causation hypotheses. CONCLUSIONS: To avoid over-estimating prevalence of support for coronavirus conspiracies, we advocate using a balanced rather than imbalanced set of response options, and including a don't know option.
Entities:
Keywords:
Coronavirus conspiracies; compliance; mistrust; public opinion; survey design
Authors: Johannes Langguth; Petra Filkuková; Stefan Brenner; Daniel Thilo Schroeder; Konstantin Pogorelov Journal: Int J Data Sci Anal Date: 2022-05-27
Authors: Pascaline Van Oost; Vincent Yzerbyt; Mathias Schmitz; Maarten Vansteenkiste; Olivier Luminet; Sofie Morbée; Omer Van den Bergh; Joachim Waterschoot; Olivier Klein Journal: Soc Sci Med Date: 2022-03-17 Impact factor: 5.379