Selecting Statistical Tests - GROUPS

We now have to work out whether your study has repeated measures, also called within-subject factors. Repeated measures occur when you measure your participants on more than one occasion. This typically occurs for two types of research design: when you measure your participants for the same dependent variable at (1) more than one time point, and/or (2) under more than one condition/treatment. Please remember that at this stage we are only looking to determine whether repeated measures exist in your study design and your overall study design might be much more complicated.

Do you have repeated measures?   Not sure, keep reading on...

(1) how do you know if you have more than one time point?

This usually means that you have some form of time-line or time-course; so you measured some variable(s) at one time point and then again at a later time point (and perhaps many more time points). A typical example of this type of research design is when you are measuring a variable pre- and post-intervention to find out if the intervention caused a change in this variable. A general schematic is shown below:

http://statistics.laerd.com © Lund Research Ltd 2010

Examples

Here are some sample examples that represent this type of study design:

1. Male volunteers aged 45 to 65 years old agree to undergo a 3-month exercise training programme to determine if exercise can reduce blood pressure. Blood pressure is measured before (pre-) and after (post-) the exercise intervention.

Do you have repeated measures?   Not sure, keep reading on...

(2) how do you know if you have different conditions?

There are advantages to using the same participants to measure the effectiveness of different treatments or conditions on a given dependent variable. A typical example is the testing of different types of drug on the same participants. Lets assume that a pharmaceutical company wants to test a new drug against an existing drug in the treatment of diabetes to see if the new drug is more effective at reducing blood sugar levels. In this repeated measures study design, each participant will take BOTH the existing drug for a set period of time and ALSO the new drug for a set period of time (but NOT at the same time). At the end of each drug treatment, blood sugar levels will be measured and a comparison made between the two. A general schematic of this is shown below:

http://statistics.laerd.com © Lund Research Ltd 2010

Do you have repeated measures?   Not sure, keep reading on...

A special case: the "Matched Pairs" case.

Why does it matter whether we are using repeated measures or not? Well, each individual is unique and they will react in a slightly different way, so by using the same individual it is possible to eliminate the variation between individuals. This makes our statistics more powerful. Part of this reason is that we do not know how one individual will react to our study compared to another as we usually know very little about them: but what if we did? What if we took different participants and matched them on criteria we thought were important. For example, blood pressure-reducing drugs are most likely to be affective in individuals that have high blood pressure and a host of other similarities. This is called a "matched-pairs design" and although the individuals are different between groups they can be treated as being the same individuals.

Do you have repeated measures?