There are three ways to design an experiment, and the design we choose can have a major impact on the validity of our research.
Independent Groups design – every participant experiences only one level (or condition) of the IV, allowing comparison between the results from different groups.
Repeated Measures design – every participant experiences every level (or condition) of the IV, allowing comparison between each participants scores in each condition.
Matched Pairs design - every participant experiences only one level (or condition) of the IV, allowing comparison between the results from different groups. HOWEVER – participants in one condition are carefully ‘matched’ (i.e. same age, gender, socio-economic group etc) with participants in the other in order to reduce individual differences.
Problems and Solutions of Experimental Design
Order Effects – the possibility that experience of one level of the IV will change behaviour in another level of the IV.
Practice Effects – a specific type of order effect which presumes that experience of one level of the IV will improve performance in another level of the IV.
Counter-balancing – a technique used to limit the impact of order effects in repeated measures. Different groups of participants experience levels of the IV in different order.
Randomisation – a technique used to limit the impact of individual differences in independent groups. Participants are randomly allocated to levels of the IV.
Strengths and Limitations
Each of these designs are vulnerable to confounding variables which may reduce the internal validity
Independent Groups – vulnerable to individual differences as comparisons between each level of the IV will mean comparisons between different individuals. This can be reduced by using randomisation. However, there is no opportunity for order effects to occur and there is less chance of demand characteristics as participants have only experienced part of the procedure.
Repeated Measures – vulnerable to order effects as each participant experiences consecutive levels of the IV. This can be reduced by using counter-balancing. Also has an increased chance of demand characteristics as participants have experienced the complete procedure. However, there is no opportunity for individual differences to occur as each participant is compared with themselves.
Matched pairs – similar to independent groups, although individual differences are reduced by the matching process. However, matching can be difficult and some individual differences will remain.