Likely ”report abuse” submissions are just one variable in a model that is used to determine propability of app being malicious.

Or it might be that humans are just reviewing top N apps, sorted by number of reports.