FIGURE 1.

Variance importance plots showing important covariates in accurately classifying potential donors as a willing donor. (a) The largest mean decrease in accuracy was found with the removal of compensation, showing that compensation plays the most important role in accurately classifying potential donors as willing or not. Removing compensation (b) or both compensation and opinions as to whether helping others outweighs the inconvenience of donation (c) from the data lead to different results when applying the random forest method, confirming that the results are obtained by chance. Some of the covariates that appeared as important are prior knowledge of how faecal microbiota transplants (FMTs) could help the others, gender and prior knowledge of what a stool collection kit looks like.