3 ms·
Sometimes you cherry-pick data not to get the results you are looking for, but because it is easier to do so. No matter the reason for the cherry-picking, it d
by taeric 9y ago
Sometimes you cherry-pick data not to get the results you are looking for, but because it is easier to do so.
No matter the reason for the cherry-picking, it does add uncertainty to the results. In the malicious case, the uncertainty is easy to predict in the direction it will go. In the "ease of picking" case, it is not.
Which is to say, it does not throw out the results. This is a useful paper and more folks should look at it. Hopefully someone has the necessary energy/effort to get all of the data and we can clear up the uncertainty that it introduced.