Experimental Design
Independent Groups Design
- Experimental design refers to how the participant sample is used in combination with the different levels of the independent variable (IV) e.g. if there are two conditions of the IV, will participants experience both conditions or only one?
- There are three main types of experimental design:
-
- Independent Groups design
- Repeated Measures design
- Matched Pairs design
-
-
- In an independent groups design participants only experience one condition of the IV for example:
- Participant A learns a poem with music playing (condition 1)
- Participant B learns the same poem in silence (condition 2)
- Participants are randomly allocated to each condition to avoid researcher bias
- Independent groups design generates unrelated data (because the two groups are separate to one another)
- The performance of the group in condition 1 is compared to the performance of the group in condition 2, for example:
- Participants who have learned a poem with music playing (condition 1) are asked to write down as much of the poem as they recall in 5 minutes
- Participants who have learned the same poem in silence (condition 2) are asked to write down as much of the poem as they recall in 5 minutes
- The dependent variable (DV) is measured as the number of words correctly recalled from the poem
- The number of words correctly recalled by participants in condition 1 is compared to the number of words correctly recalled by participants in condition 2
| Strengths of Independent Groups Design | Limitations of Independent Groups Design |
|
Less likely to have demand characteristics as participants only take part in one condition, meaning they are less likely to guess the purpose of the study and how to behave for that purpose. |
There can be participant variables/individual differences. If more participants with a particular characteristic are all assigned randomly to one condition, this can affect the results i.e. they are not a true measure of the IV's effect on the DV. |
|
Due to participants only taking part in one condition, it means there are less likely to be order effects, meaning they cannot predict what happens next and change their behaviour. |
More participants are need to ensure there are enough to take part, which can sometimes be problematic if enough participants are not easy to find. |
Repeated Measures Design
-
- In a repeated measures design participants experience all conditions of the IV for example:
- Participant A learns a poem with music playing (condition 1)
- Participant A learns a different poem in silence (condition 2)
- The same participants complete each of the experimental conditions
- Repeated measures design generates related data (because the two scores come from the same participant)
- Participants act as their own control group as their performance in condition 1 can be compared to their performance in condition 2, for example:
- Participant A learns a poem with music playing (condition 1) and is asked to write down as much of the poem as they can recall in 5 minutes
- Participant A learns a different poem in silence (condition 2) and is asked to write down as much of the poem as they can recall in 5 minutes
- The dependent variable (DV) is measured as the number of words correctly recalled from the poem
- The number of words correctly recalled by the participant in condition 1 is compared to the number of words they correctly recalled in condition 2
- To avoid order effects researchers should use counterbalancing:
- Half of the participants experience condition 1 followed by condition 2
- The other half of the participants experience condition 2 followed by condition 1
| Strengths of Repeated Measures Design | Limitations of Repeated Measure Design |
|
There are no individual differences as the same participants are used in each condition, meaning participant variables do not affect the measurement of the IV s |
Order effects can be a problem as participants take part in all conditions. This can result in a better performance in the second (or third or fourth etc.)condition due to practice or learning what is expected, or, it could result in worse performance as the participant could be bored or tired/lethargic. Counterbalancing can help control this, by splitting the participants into groups and changing the order of the conditions per group e.g. half the sample complete condition A and then B; the other half complete condition B and then A. |
| Fewer participants are required as each participant generates (at least) two scores | Demand characteristics are more likely as the participants are more likely to guess the purpose of the research due to taking part in multiple conditions. |
Matched Pairs Design
- Participants (usually) achieve just one score as they (usually) only take part in one condition
- Participants are assessed and matched on the characteristic(s) that are important for the particular research they are taking part in, e.g. age, ethnicity, gender etc.
- Often MZ (Monozygotic/ identical) twins are used for this design as they create the perfect matched pair (one twin can be assigned the experimental condition and the other twin the control condition)
- The matched participants are then randomly allocated to one condition each
- As each participant is related to their pair this design produces related data, for example:
- In an experiment on the social learning of aggression, participants would be matched on a scale according to how aggressive they already are
- So: participant A who ranked 10 for aggression would be matched with participant B who also ranked 10 for aggression
- Participant A takes part in condition 1 of the experiment; participant B takes part in condition 2 of the experiment
- This matching helps to factor out aggression as a possible confounding variable in the experiment i.e. any difference in scores should be due to the effect of the IV and not on natural aggression levels
| Strengths of Matched Pair Design | Limitations of Matched Pair Design |
| Due to participants only taking part in one condition, it means there are less likely to be order effects | Matching is difficult and it is often impossible to match all characteristics, especially when the unmatched characteristic could be important to the results of the research. Also, even well matched participants could have different levels of motivation in the study, affecting the outcome. |
| Almost factors out individual differences as a confounding variable as the researchers have striven to find a 'match' per participant i.e. participant variables are controlled for to some extent. | More participants are need to ensure there are enough to take part in the different conditions. |
| Matching participants is very difficult and time consuming. |
