3 ms·
No. The statistics here are absurd: "We compiled data on 250+ US public companies involved in corporate securities frauds identified in 1,000+ Securities and E
by gwern 6y ago
No. The statistics here are absurd:
"We compiled data on 250+ US public companies involved in corporate securities frauds identified in 1,000+ Securities and Exchange Commission filings over 2005–2013; we randomly selected a comparable control group of 500+ US public companies from Compustat."
- djbebs 6y agoYeah their entire study is "we looked at a list of companies who comitted fraud, and can see that the percentage of companies in that list that committed fraud is higher than that of the average company" Nothing was discovered here, and the title is not supported by their methodology
- cycomanic 6y agoWhat do you mean this is a perfectly valid statistical method to employ. It's being used in all sorts of contexts, e.g. violent crime, men are X times more likely to commit violent crime. That number comes from instances of violent crime and then you related the percentage of instances committed by men to the percentage of men in the population. That's the exact same methodology, except here in this study they use the control group to find what the average size of a public corporation is.
- _8091149529 6y agoJust posting to lament the fact that there are two blatantly garbage-tier research articles on the front page at the moment (this and the Ramanujan machine).