Figure 2
From: The pervasiveness and policy consequences of medical folk wisdom in the U.S.
Psychometric properties of the MFW scale (Study 2). Note (a) Presents the distribution of the raw count of folk theories endorsed in the sample—i.e., the count of respondents who indicated that each theory was “definitely" or “probably" true—displayed as a histogram. (b) Presents percentages of respondents who endorsed each specific theory (again coded dichotomously), displayed as a bar chart. (c) A scree plot derived from an unrotated principal components analysis (PCA) assessing the factor structure of all 11 folk theories. The large (i.e., greater than 1) Eigenvalue associated with a one-factor solution is suggestive of unidimensionality. Finally, (d) plots item characteristic curves resulting from the 2PL IRT model referenced in the text. S-shaped curves indicate that people who endorse each item tend to have a high probability (y-axis) of being classified as scoring highly on the latent MFW scale (x-axis), while those who do not endorse these items tend to have a low probability of doing so.
