Repeated Measures, Matched Pairs and Independent design: which design has fewer flaws?
When conducting research in psychology, in the form of an experiment you need to decide on an experimental design. The three main designs used are: repeated measures, matched pairs and independent design. All three have flaws but is it possible that one can bring higher validity to an experiment than others?
Independent measures design can minimise the chances of order effects occurring as different participants are used, however a main problem of this design is individual differences / participant variables. How do we know if our results are reliable or if there was a difference between the ability of the different groups that showed through the different variables?
Repeated measures dismiss’ the problem of individual differences which highers the validity of the study. However, by using the same participants in each condition, order effects are likely to occur such as: the participants getting bored, feeling fatigue or fed up by the time of the second condition, or even getting more knowledgeable to the requirements of the test which in turn may lead to demand characteristics. Charles Stangor(2010) wrote that “repeated measures research designs represent a useful alternative to standard between participants designs in cases where carryover effects are likely to be minimal.” This shows that although repeated measures has a big flaw, order effects, in some cases this may sill be the best option for the experiment.
Matched pairs helps to combine the both to decrease problems such as individual differences and order effects: two sets of participants are used but they are matched on factors such as age, sex and social background. This can be seen as effective, however, no two participants can be matched exactly- even identical twins may have different thinking or life experiences. This is also very time consuming.
As all three have factors with lower validity this can become a problem for the experiment: we need to know that we are measuring what we say we are measuring and not order effects or individual differences.
Overall I think that repeated measures is the most important and can have fewer flaws especially when counterbalancing is used: meaning that this method should be used more often in experimental research. However, there are some cases in which it can only be an independent measure is used- For example in Maguire et al’s research on taxi drivers’ brains, he two conditions were those who were taxi drivers in London and those who weren’t.
When an option is available to choose from I feel that repeated measures design is the best method to use as it can higher validity.
psychologyblogaimee 
I thought this was a very interesting and clear blog, it would have been nice to have some more in depth examples such as a hypothetical experiment but I thought you did a good job of covering the different designs and I liked the fact that you concluded which one was the most useful. I do think matched pairs is a good design however because order effects are also eliminated as well as participant variables, even though this takes longer and requires more participants.
This blog is very interesting and you have done a great job of covering the advantages/disadvantages of each of the different designs. I also agree with “rhinon99” and would say that “matched pairs” is the best design to use. This is not really because I am very enthusiastic towards this method, but because there are too many flaws with the other two designs. Firstly, an independent design is fairly similar to matched pairs as the participants are only used once, however would you not agree that in the 2nd, 3rd or 4th test that there would be much more of an advantage to have participants who are very similar to each other (not the same, but similar!) than to have a complete, new random sample of participants who will very likely range in age, gender, intelligence etc? As for repeated measures, well as you have clearly pointed out many of the negative factors, it is wide open to gaining unreliability in results as things like “the participants getting bored, feeling fatigue or fed up by the time of the second condition, or even getting more knowledgeable to the requirements of the test which in turn may lead to demand characteristics,” all threaten the reliability of the results. This is not found in matched pairs. The only issue with matched pairs is that the participants are EXACTLY the same, however there may be a few strategies to try and reduce this problem. We can categorise people very quickly and efficiently by their physical attributes eg. age, gender, height, weight etc. however it is much more difficult to do this by mental characteristics. Therefore to try and overcome this problem, it may be logical to distribute a number of questionnaires or IQ tests before the study begins, to group participants into groups where there results have proven similar. Although this method may seem very time consuming, I believe that it is the most valuable design to use, if it is conducted in the correct manner.
interesting topic of choice and i really enjoyed your analysis of flaws and positives of each experimental sample design. Personally i find independent measures to be the most effective and useful experimental sample design as although it lacks in ability to generalise to the rest of the population, it more then makes up in its ability to avoid order effects and sample bias which the other two i feel are more prone to.
The advantages and disadvantages of each of these experimental designs really help decide which is the best to use and in my opinion Independent participants(measures) would be my first choice, as it is effective and reduces order effects. However, the fact that it is harder to generalise to the general population becomes a problem. But the other two designs, in my opinion, have more flaws which outweigh the benefits of the independent measures design.
I think that they can all have the flaws and matched pairs can be seen as the better choice because of its advantages of having no order effects. However, It can be very time consuming and we still don’t know if we are measuring what we say we are or just individual differences. If counterbalancing is used, for example half the participants take part in variable 1 first and the other half take part in variable 2 first, and then switch we can analyse the results to decide whether there s a significant difference in our results or whether we are just measuring order effects. In this way by using repeated measures but with counterbalancing we can minimise order effects as well as having no individual differences. This method also requires less participants than independent measures and less time consuming than matched pairs but with fewer flaws. So even though the other two methods are good and can be more useful for certain experiments I still think that this method is more reliable and has fewer flaws.