The internally-assessed experiment (IA) that you conduct in Psychology is worth 20% of your marks if you are an HL student and a whopping 25% if you are an SL student, so it is worth getting it right. Here are some thoughts to help you along the way.
Choose the right study. Attached is a list of studies that are all suitable for replication, with a link to the original study in each case. The originals are sometimes hard to read, but you should be able to find a good summary online. At least read the abstract and the introduction if possible. You should be thinking about using studies into memory or perception as they have few ethical problems and are relatively easy to conduct.
Keep it simple
Your IA should be a replication or partial-replication of a published study. If you are an HL student you may significantly modify the study; if you are a student at either level it would be expected that you would simplify it. For example Loftus and Palmer’s 1974 study into eyewitness testimony used 5 groups of students, 7 video clips and had two parts to it. There was also a total of 45 participants in the first part experiment and 150 in the second. For your IA you would be expected to use only two groups of about 10 participants (total number 20) and replicate just the first part of the experiment, using one video clip of a car crash.
Generate statistics at the ordinal level or above
You will be required to calculate one measure of ventral tendency (mean, median or mode) and one measure of dispersion (standard deviation, variance, inter-quartile range) on your data. This is much easier if you have data that is ranked or conforms to a scale in some way, rather than categorised nominal data which makes comparison using these measures difficult. Moreover, HL students will have to determine if there is a statistically significant difference between the means of both groups. This cannot be done if we are comparing nominal data. If you are not sure what all this means, see this site. If you need some help in your statistical analysis, see here.
If you have separate groups (independent measures design) you will probably use the Mann-Whitney test. If you are using the same participants in both experimental conditions – or in the control and the experimental condition – then you will be using the Wilcoxon test, as this is a repeated measures design.
HL students should find this video helpful
Keep it ethical
No experimenting on animals or young children. Nothing that could cause psychological or physical harm, and gain full informed consent from your participants and ensure that they know their rights to withdraw themselves and/or their data.
Finally, have fun. Be well prepared and organised. This is “real psychology” in action!