| Sumario: | Part of a special section on the advantages and disadvantages of null-hypothesis significance testing in psychological research. The writer argues that null-hypothesis significance tests (NHSTs) are useful in research. Both NHSTs and confidence intervals can be useful in indicating whether there is sufficient evidence to be confident of the sign of the population effect. However, confidence intervals are subject to the same Type I, II, and III error rates as NHSTs, without being able to provide some of the information that NHSTs can supply. Moreover, the proposed alternatives to NHSTs are just as susceptible to misinterpretation, and banning NHSTs is not going to guarantee that misunderstandings regarding sampling error and the strength of evidence needed to support a conclusion will not continue to occur. It is concluded, therefore, that problems of bias can be handled by providing archives for all methodologically sound data sets and that NHSTs can be reserved for making interpretations and conclusions.
|
|---|