Explore BrainMass

Justification for an Alpha Level in Hypothesis testing

Do alphas differ among industries? How do ethics affect your chosen level of significance? Can you not reject and reject the null hypothesis? Why or why not?

Solution Preview

Yes, Alphas do differ by industry and even between studies within the same industry. The alpha level is really the amount of time you are willing to accept a type I error (accepting a result as true when it is really false). So, if you are doing a study on the causes of heart disease, it would be critical to find every possible correlation and you would want a low alpha level (.001). But remember that Type I errors and Type II errors are inversely correlated. If you lower the type I error you increase the risk of a type II error (accepting a result as false when it is really true). To minimize the possibility of a type II error you will want to increase the power of the study as much as possible. This is done through the largest sample size possible and minimizing the extraneous variance as much as possible. On the other hand, if you are doing a study on language perception, you ...

Solution Summary

Why do some industries/scientists choose a stringent Alpha level of .001 and other scientists choose a more liberal Alpha level of .05. This article describes the rationale behind choosing an Alpha level and how this choice can change your experimental error rate.