Читать книгу The Research Experience - Ann Sloan Devlin - Страница 126

Research in the social and behavior sciences typically uses inferential statistics, which means that we use samples to make informed guesses about the characteristics of the population from which the sample is drawn. In other words, we don’t know for sure about an answer to a particular outcome because we haven’t asked or assessed every person in the population of interest. We have asked what we hope is a representative sample. But we could be wrong because we are using inferences about our statistical hypotheses. Type I and Type II errors describe the ways in which we could be wrong. In a Type I error, we claim that we have a significant statistical result when that is not the case. Formally, we reject the null hypothesis (of no statistical difference or relationship between groups) when we should not have done so. In a Type II error, we have missed a finding that is there. Formally, we fail to reject the null hypothesis when there is a statistical difference, that is, when we should have done so.

Figure 3.5 Actual State of Affairs

Figure 3.5 represents the four possible outcomes. We can be correct in two ways: We are correct when we reject the null hypothesis when there is a finding; we are correct when we do not reject the null hypothesis because there is no finding. We can also be incorrect in two ways (our Type I and Type II errors). We are incorrect when we reject the null hypothesis (say there are statistical differences or relationships; a false alarm) when there are none (Type I error). We are also incorrect when we do not reject the null hypothesis and we should have, that is, such statistical differences or relationships are there (Type II error; a miss).

Bonferroni adjustment: Adjustment for Type I error by dividing the alpha level (.05) by the number of statistical tests performed to create a new more stringent alpha level.

Two-tailed significance test: When the critical region for significance is spread across both tails of the distribution.

One-tailed significance test: When the critical region for significance is limited to one tail of the distribution (more lenient than two-tailed tests).

Both Type I and Type II errors should be avoided, but Joseph Simmons et al. (2011) view Type I as more problematic. Their argument is once these false positives or incorrect statements that a finding exists appear in the literature, they tend to stay there. Such findings can encourage other researchers to investigate the phenomenon further, which may be a waste of resources.