Metadata
- Author: Andrew Gelman
- Full Title:: Multiple2f
- Category:: 🗞️Articles
- Document Tags:: bayesian testing, Bayesian testing,
- URL:: http://www.stat.columbia.edu/~gelman/research/published/multiple2f.pdf
- Finished date:: 2024-05-06
Highlights
we are typically not terribly concernedwithType 1 error because we rarely believe that it is possible for the null hypothesis to be strictly true. Second, we believe that the problem is not multiple testing but rather insufficient modeling of the relationship between the corresponding parameters of the model. (View Highlight)
The classical perspective worries primarily about Type 1 errors, and we argue that these should not be the focus of our concern (View Highlight)
Amore serious concern might be that we claim τj > 0when in fact τj < 0, finding a positive effect when in fact the effect is detrimental. (View Highlight)
Classical inference in essence uses only the information in each site to get the treatment effect estimate in that site and the corresponding standard error. 9 (View Highlight)
Amultilevel model, however, recognizes that this site-specific estimate is actually ignoring some important information—the information provided bythe other sites (View Highlight)
we are typically not terribly concernedwithType 1 error because we rarely believe that it is possible for the null hypothesis to be strictly true. Second, we believe that the problem is not multiple testing but rather insufficient modeling of the relationship between the corresponding parameters of the model (View Highlight)
This overall risk of error (formally, the probability that we have any rejections when the null hypothesis in fact holds) is sometimes referred to as the familywise error rate (Tukey, 1953). (View Highlight)