Experimental Design – Matched Pairs

There are three types of experimental design these are independent groups, matched pairs and repeated measures.

The matched pairs design focuses on matching one participant up with another, but the person they are matched with will share similar or exact traits such as IQ, age or ethnicity. It would be easier just to use twins because they would share the exact same genetics however twins are hard to come across in research experiments and the sample size would be too small. In a way the matched pairs design is like finding your twin (but they don’t always look the same).

Here are some examples of how people can be matched:

Megan Fox & Lady Gaga- Age 25, Keira Knightley & Natalie Portman-Looks, Steven Gerrard & Paul Mccartney-Male, Liverpool

 Megan Fox & Lady Gaga- Age 25, Keira Knightley & Natalie Portman-Appearance, Steven Gerrard & Paul Mccartney-Male, Liverpool.

Matched pairs are difficult to find, as you can see above individuals may match in one aspect but could be completely different in another. This weakens the reliability of the research as one of the participants would be put into the control group and the other in the experimental. The experimenter will try to find participants with the same variables, the more they share the more support there will be for evidence. Although they might just focus on what they are testing so sometimes they only need to match on one variable such as IQ. Some variables can affect the outcome of the study, using matched pairs can ensure a degree of consistency. However it is impossible to match every single variable. For example what if the participant did not have much sleep the night before the study and their matched pair did, this variable could affect the outcome of the study.

Matched pairs can be beneficial within the field of biology and drug treatments. Such benefits have been found in research into breast cancer patients and the drug reserpine. Reserpine is a drug to control high blood pressure and to relieve psychotic symptoms. Fifty five female participants diagnosed with breast cancer were matched with patients who did not have breast cancer. They were matched on age, psychiatric diagnosis, race and religion. These patients had all been given a psychiatric diagnosis and the drug reserpine was being tried to relieve the symptoms. All of the patients who used reserpine were monitored over a period of time. There were no significant increases of breast cancer. The drug could then be used to treat patients who had a psychiatric illness and also those with the illness and breast cancer (Laska, Meisner, Siegel,Fischer and Wanderling 1975).

Compared to a repeated measures design matched pairs has an advantage. During a repeated measures experiment the participants will have to carry out all of the conditions. There are no separate groups. For example, condition one could require the participants to run around a course. Condition two is that the participants must first drink a can of red bull before they run the course. The idea is that the participants will do better in one condition than the other. However order effects may occur such as learning and fatigue. Participants may run the course quicker in the second condition because they have learnt the course. If they run slower in the second condition this could be because they are tired. It may have nothing to do with what the experiment is trying to research which in this case is the red bull. There are no order effects in matched paired experiments. If it was easy to get participants to match up on almost every single variable matched pairs would be the best experiment out of the ones I have mentioned to use. But there is a way to get rid of order effects in repeated measures, by using counterbalance. This is where half of the participants will carry out condition one first and the other half will carry out condition two first. Thus making the results of a repeated measure design reliable.

Reference

Laska, E., Meisner, M., Siegel, C., Fischer, S., & Wanderling, J., (1975). Matched-pairs study of reserpine use and breast cancer. Which appears in The Lancet. 306, 7929, 296-300. DOI 10.1016/SO140-6736(75)92731-2

Advertisements

One thought on “Experimental Design – Matched Pairs

  1. I agree that matched pairs is a useful experimental method when one particular variable is being tested in two conditions. For example if testing the effects of eating a healthy large breakfast before taking an exam matched pairs would be a good option. Participants would be matched on IQ and although independent groups could be used it could be argued that this would lower the reliability of the study as orfer effects could occur and learning could take place between each test and the results could be effected by other factors instead of whether they did or didn’t have a healthy breakfast.

    However, i’m not convinced that matched pairs is always necessary. Firstly when matching up variables you can never match people exactly, there will always be individual difference. Also, it takes much time to match people which isn’t always successful. Also if many participants are simmilar would it be as representative as a completely random sample? As in my above example, if independent groups are used, there will be a mix of IQ’s in both groups and a random sample of people could of been used for the study. Unless it’s definately necessary this would be much easier to carry out and it would be easier to use a large sample size which would therefore make the results more generalizable and where possible its important to use a randomized sample, (Donald, 1974).

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s