Читать книгу Abnormal Psychology - William J. Ray - Страница 138

When performing research studies, we use inferential statistics to determine whether the IV influences the DV. By using statistics, we ask, if we performed the same experiment 100 times, what is the probability we would obtain the results seen in the present study? Actually, with statistics, the convention is to ask the question in the other direction. That is, how many times would we expect the results not to be the same? If the answer is less than 5 in 100 times (p <.05), we say that the results of the study are statistically significant.

statistically significant: the probability that the independent variable (IV) influences the dependent variable (DV) by chance alone

clinically significant: the characterization of the results of a study when, beyond being statistically significant, the findings indicate clinically important outcomes

When considering medical or psychological disorders, we also want to know if the results of the study are clinically significant. For example, if you did a study related to dieting and everyone in the experimental group lost .5 pounds, while everyone in the control group gained 2 pounds, the results would be statistically significant. However, clinically, you would not recommend a treatment that only resulted in weight loss of half a pound.

Consider a study in which a researcher wants to determine whether performing exercise would reduce depression. In this study, the experimental group would receive the exercise training for 2 months and the control group would not. Assume that the exercise did indeed reduce the depression score on a particular measure of depression from 21 to 20, whereas the control group’s depression did not change. Statistics might show a significant relationship between the two sets of data. However, clinically, it would not be worth the effort of having participants exercise for 2 months to have depression change by only 1 unit. Thus, a distinction is often made in clinical work between results that are statistically significant and those that are also clinically significant.

effect size: the measured magnitude indicating the influence that a treatment has on the dependent variable (DV)

One way to measure the magnitude of effect that a treatment has on the DV is referred to as effect size. A common measure of effect is Cohen’s d. In essence, Cohen’s d reflects the difference in the mean scores of the control and experimental groups divided by the standard deviation of the measure from the two groups. Effect size measures are important to clinical researchers for two reasons. First, they describe in quantitative terms the influence of the treatment, and second, they aid a researcher in knowing how many participants need to be included in a research study to determine an effect. Effect size is an important measure of the effects of a treatment on a mental disorder. One could compare two different types of psychotherapy, for example, or even a psychotherapy combined with a particular medication.