4 ms·
>represent statistical significance at the 1%, 5%, and 10% levels, respectively. Isn't this data dredging? Don't you define the test parameters in advance, and
by itsoktocry 3y ago
>represent statistical significance at the 1%, 5%, and 10% levels, respectively.
Isn't this data dredging? Don't you define the test parameters in advance, and then it's either significant or not? The p-value doesn't represent the "strength" of significance.
- carbocation 3y agoI wonder if that's just journal style? Some of the P values in the paper are significant at levels that would survive almost any amount of multiple testing correction. E.g., in Table 3, the P value for DairyCheese is P=1.3E-139. While I agree that the "strength" of the P value is not the thing of interest (which is, generally, the magnitude of the effect), the probability that you'd see results like this if there was no relationship between transaction size and the dairy category is ~0.