We implemented a structural equation modeling analysis in Mplus 8 (Muthén and Muthén, 2017) to estimate parameters and test hypotheses concerning the relationships depicted in Figure 1. Ten latent variables were defined according to the coding scheme for composite scores (see “Protective behaviors,” “Risk perception,” “Attitude,” “Experience,” “Cultural worldviews,” and “Normative conducts”). Thus, hierarchy–egalitarianism, individualism–communitarianism, and direct and indirect experience of COVID-19 were the exogenous latent variables; social norms, affective attitude, feelings of risk, risk analysis, promoting hygiene and cleaning, and avoiding social closeness were the endogenous ones. Because we measured all the latent variables by multiple Likert-type (or ordered categorical) items, we carried out the analysis using robust weighted least squares estimators (WLSMV). This method makes no distributional assumptions and is recommended to handle ordinal data (Rhemtulla et al., 2012).

Besides model χ2, we assessed the model’s fit using other descriptive indexes: comparative fit index (CFI), Tucker–Lewis index (TLI), root mean square error of approximation (RMSEA), and standardized root mean square residual (SRMR). According to Hu and Bentler (1999), CFI and TLI greater than 0.95 indicate a good fit of the model, with values above 0.90 deemed acceptable. A good fit is also supported by RMSEA and SRMR lower than 0.06 and 0.08, respectively. We relied on four criteria to assess the quality of the measurement model. First, all empirical indicators should load on the corresponding latent variables above 0.50 (indicator reliability). Second, the composite reliability (CR) of each latent variable was expected to be greater than 0.60 or better above 0.70 (construct reliability). Third, the average variance extracted (AVE), an index of the proportion of variance in the indicators that was accounted for by the corresponding latent variable, should be greater than 0.50 or higher (convergent validity). Lastly, the square roots of the AVE for each latent variable should be greater than the estimated correlations of that latent variable with other variables in the model (discriminant validity).

The significance of indirect effects was tested using bias-corrected bootstrap confidence intervals with 1,000 resamplings. Each indirect effect represents the average increase in protective behavior accounted for by direct and indirect experience of COVID-19 and hierarchical and individualistic worldviews through specific intermediate variables, like affective attitude, risk perceptions, and social norms. Standardized indirect effects around 0.02, 0.13, and 0.26 represent small, medium, and large effect size thresholds, respectively (Hayes and Rockwood, 2017).

Note: The content above has been extracted from a research article, so it may not display correctly.

Please log in to submit your questions online.
Your question will be posted on the Bio-101 website. We will send your questions to the authors of this protocol and Bio-protocol community members who are experienced with this method. you will be informed using the email address associated with your Bio-protocol account.

We use cookies on this site to enhance your user experience. By using our website, you are agreeing to allow the storage of cookies on your computer.