Table 1.
Recruitment and translation details for each country in the study.
Fig 1.
Plot of severity value vs. reverse severity rank order for each country under a free-text numeric measurement.
Note that the same rank position may indicate different harmful content for different countries.
Table 2.
Exponential regression results of each country’s data.
Here, p < 0.001 for all parameters. See T1 in S1 File for the full set of regression results.
Fig 2.
Plot of severity value vs. reverse severity rank order for each country under a Likert-scale measurement.
Fig 3.
Plot of explained variance vs. number of principal components in PCA.
Fig 4.
Plot of average silhouette score vs. number of clusters in Gaussian Mixture models.
Table 3.
Pairwise ranking correlation results for all countries.
Table 4.
Country clusters and their ranking characteristics.
Fig 5.
Harmful content’s standard deviation vs. sample-wide ranking.
Fig 6.
Harmful content’s Δrank vs. sample-wide ranking.
Table 5.
Types of harmful content that had max ranking differences (Δrank) of at least 33, or half of the total number of rank positions.
Table 6.
Countries’ median ranking of each type of harmful content.
1 = most severe; 66 = least severe. The World ranking is a country-agnostic ranking instead of an aggregation from individual country rankings.