Statistical analysis
Following data collection, the survey responses were coded and entered into a customized database using the Statistical Package for the Social Sciences (SPSS), Version 24.0 (IBM Corp., Armonk, New York, USA). Descriptive results were presented as means and standard deviations for continuous variables and percentages for qualitative variables. A one way ANOVA test was performed to analyse regional differences in perception scores. All tests were two-tailed. A P-value of <0.05 was considered statistically significant.
Correlation between awareness score (out of 20) and the COVID-19 statistics of cases and deaths announced at the beginning and end of the study period (12th to 22th of April 2020) for the countries which had at least one case at the beginning of the study was also conducted.
Linear regression was used to screen for the factors affecting participants’ awareness score about coronavirus pandemic versus chosen independent variables in the study , i.e. age, area of residency (city and urban areas or rural areas), country, region, having children, educational level, university type (the university where participants had studied and/or are studying at; public versus private), years of experience, number of professional education workshops attended during the last year, work setting, source of previous knowledge about epidemics and pandemics, source of updates about COVID-19 management, and current satisfaction with knowledge about COVID-19. These predictor measures (independent variables) were considered as candidates for linear regression modelling if they had a significance value p ≤0.25 in univariate analyses. The candidate variables were subjected to backward linear regression, where finally only the significant variables (i.e. p ≤0.05) were retained with the model equation constant. Variables were selected after checking their independence, where tolerance values > 0.1 and Variance Inflation Factor (VIF) values were < 10 were selected to indicate the absence of multicollinearity between the independent variables in regression analysis. The homoscedasticity assumption for multiple linear regression was checked using Breusch-Pagan test, with a p≥0.05 indicating the absence of heteroscedasticity.