Within-subjects statistics are used to compare observations of an outcome across time or within-subjects. There are independent observations of the outcome for the same person or unit of measure at different time points or "within" the same person or phenomena. Within-subjects statistics are often used to assess how outcomes change or stay stable across time or as the result of an intervention. A baseline observation of an outcome is taken and then compared to subsequent observations of the same outcome at different points of time or within the same person.
Within-subjects statistics, by far, possess the most statistical power in that significant treatment effects are easy to detect. This is primarily because each participant serves as their own control in within-subjects designs. Here are the questions to be answered when choosing the correct statistical test:
1. How many observations of the outcome are going to collected in the within-subjects analysis?
2. What is the scale of measurement for the outcome in the within-subjects analysis?
When asking research questions and answering them using within-subjects statistics, researchers are asking how do participants change across time or "within-subjects" for each observation of outcome?
Within-subjects statistics, by far, possess the most statistical power in that significant treatment effects are easy to detect. This is primarily because each participant serves as their own control in within-subjects designs. Here are the questions to be answered when choosing the correct statistical test:
1. How many observations of the outcome are going to collected in the within-subjects analysis?
2. What is the scale of measurement for the outcome in the within-subjects analysis?
When asking research questions and answering them using within-subjects statistics, researchers are asking how do participants change across time or "within-subjects" for each observation of outcome?
Baseline frequency, baseline median, baseline observation
McNemar's test, Wilcoxon, repeated-measures or paired t-test
Cochran's Q, Friedman's ANOVA, repeated-measures ANOVA
