I don't think Simpsons paradox plays any role here, if I understand this correctly. In the first graph, the leading digit counts are not the things being summe

I don't think Simpsons paradox plays any role here, if I understand this correctly. In the first graph, the leading digit counts are not the things being summed up across all the regions. The weekly report counts are summed up across all regions and then the leading digits are counted.