Having presented or described data (using graphs, tables, averages etc.) the psychologist must now analyse the data using statistical tests.
Statistical tests are used to decide whether any pattern found in a set of data is significant or whether, on the other hand, it was likely to be caused by chance.
The meaning of a “significant result” is often confusing at first. It is all to do with deciding whether your results may have been due just to chance factors, or whether you really have discovered a relationship of some sort. Probability is used to decide whether your results are indeed significant or not.This 5% significance level is the most commonly used. This is used to see if there is less than a 1 in 20 chance of our results occurring by chance (5% = 5/100 = 1/20 = 0.05). Psychologists have agreed that this is an acceptable cut-off point, an acceptable measure of “unlikelihood”. If a result is significant at the 5% level, this means that it is very 18 unlikely (there is less than a 1 in 20 chance) of the result being due purely to chance factors. You have probably found a genuine pattern/relationship/difference.
So, if your result is found to be significant at the 5% level, it means that the probability (or likelihood) of your result being a fluke is less than 0.05. This is written as p<0.05.
Sometimes psychologists use a 1% significance level. This is more rigorous than the 5% level. If a result is found to be significant at the 1% level, it means that there is only a 1 in 100 chance of the result having been a fluke. Thus the probability of your result having been a fluke is less than 0.01 (p<0.01).
Deciding on which statistical test to use depends upon the type of data you have collected. There are three decisions you have to make:
Categorical data is made up of frequencies – the number of people who do something– male/female, smoker/non-smoker.
Numerical data is where each subject has a number you’ve measured – e.g. a memory score for a subject (N.B. in your coursework when you describe your statistics test you use the word ordinal not numerical)
A difference is where you’re looking at two groups to see whether the average scores are different from each other e.g. a memory test, before and after coffee (2 conditions, 1 group of subjects) or between two groups of subjects in the same condition.
A correlation is where you usually have a single group of subjects and you are looking for a correlation between two variables, such as a high stress score goes with a high illness rating score.
Related data is where you have two scores for one set of people e.g. memory with and without coffee or matched pairs where each member of one group is matched with an individual from another group.
Unrelated data is where you have two different groups of subjects e.g. a memory test – one group with coffee, one group without coffee.
Here is an actvity to learn about statistics in psychology.
The handouts explains errors in statistical research.
A handout detailing how to use SPSS.
Back to Homepage