Pre-post test designs, also known as repeated measures designs, involve the repeated measurement of the same individuals at two (or more) timepoints. Repeated measures designs allow for a statistically powerful analysis of changes in a measure over time, or to assess the effect of an intervention. The main benefit of repeated measures designs, other than the ability to assess change, is the control of “between-subject” variability. For example, say we were conducting a study for a cheese company that wanted to know how a labelling change would affect participants’ satisfaction with their ever-popular extra sharp cheddar. While we could have different people assess the two different cheese labels, there would be a source of experimental error—the inherent variability between people. Repeated measures designs control for this inherent variability by taking into account each participant’s “starting score.” For example, one individual may already be pretty happy with the cheese brand, so a high post-test score would not mean much, while another participant may be indifferent at pretest, meaning that a high-posttest score would indicate a larger change.
Aligning theoretical framework, gathering articles, synthesizing gaps, articulating a clear methodology and data plan, and writing about the theoretical and practical implications of your research are part of our comprehensive dissertation editing services.
One key aspect of this design is that each participant’s starting score and ending score (e.g., pre and post-test scores) can be linked in some manner. This is vitally important—without matching the scores in your dataset, you will not be able to conduct a repeated measures analysis and control for between-subject variability. This means that you will need to ensure that you assign some sort of confidential identifier unique to each participant.
If you have already conducted your study and did not include a reliable way to link your participants, you may be able to match participant scores based on demographic information that you collected. For instance, if there is only one male participant, it would be pretty easy to ensure that those scores are matched up. However, you would have to use multiple sources of information (education, age, marital status, race) to narrow down the other participants. This does mean, however, that you would have required participants to indicate their demographic information at all points that you measured them. In this case, you also are assuming that the demographic information that participants provide is completely consistent from pretest to posttest.
If you cannot narrow down your participants in that manner, you will not be able to conduct a repeated measures design. In that case, you may still determine differences between pre and posttest scores (by treating “pre” and “post” as groups of your independent variable), but this approach reduces the validity of your findings. Because non-repeated measures analyses assume that all observations are independent (i.e., no one was measured more than once), your results may not be completely valid because this assumption is violated. If you take this approach, make sure that this limitation is clearly noted within your research. The best approach is to put a plan in place to link your participants across your measurements before you collect the data. In doing so, you can be confident that your statistical analysis can be carried out as planned.