STATS 408 Lecture 5: Types of Biases
Document Summary
Things that are published in journals are surprises. Observed an outcome, which if the status quo were true (under normal circumstances), should only happen rarely: it"s a surprise that it happened. If the status quo is true, the probability of an observation as or more extreme than the one we observed is called the p value of the test. What proportion of people will get a significant result even if the treatment has no effect: 5 percent of the time, type 1 error reject status quo when it was really true. When you read a paper, you don"t know if you are reading a type 1 error or something that is truly important: just because it"s published, doesn"t mean it"s true. More apparent in some circumstances more than others. In studies with a large number of observations, everything is statistically significant: doesn"t mean they are practically significant. The variance of the proportion gets smaller as the sample size increases.
