In your IA you may wish to use a parametric test of difference if your data meets all of the parametric assumptions. This is often unlikely to be the case though as usually the sample size is too small. If you have been very conscientious and collected LOTs of data then you may be justified in doing a t-test.
Before you decide to go ahead and do a t-test make sure you can answer yes to a few important questions.
A t-test is only justified if…
- You have data that is at least interval (i.e. it is in standardised units), to be ratio you will need to have a true zero (i.e. you cannot have minus values)?
- You have homogeneity of variance, i.e. your standard deviations are relatively similar, meaning samples have a similar degree of variation between the scores?
- The data in both groups or conditions is normally distributed, meaning 68% of scores are within one standard deviation and 95% of the scores are within two standard deviations? This doesn’t need to be exact but you should get a “relatively” bell-shaped curve when the scores are plotted. If you have samples of around 30 you can assume normal distribution.
Before going ahead you will also need to make sure you are using the right sort of t-test.
- If you have an independent measures design you will need an unrelated t-test.
- If you have a repeated measures design you will need an related t-test.
Remember when you are looking at your p value, it is essential that you check the correct figure…
- if your hypothesis is directional, you need to check the p value for a one -tailed test.
- if your hypothesis is non-directional, you need to check the p value for a two -tailed test.
psychologyrocks 