Three basic questions must be answered:

1. WHAT shall we study? (variables)
2. WHO will be studied? (participants)
3. HOW will the data gathered be used to demonstrate a real difference? (statistics)

Basically, psychological research involves manipulating one variable in order to affect another. For example, if I think that lack of sleep adversely affects performance on psychology tests, then the sleep would be the variable I would manipulate, and the performance on the test would be the variable I would measure. Sleep is the independent variable and performance on the test is the dependent variable.

OK, now think about these questions:

• How would you know that it was indeed lack of sleep that affected the performance, and not something else, like the subjects just not being good at tests, or having eaten too much just before the test, etc?
• How many groups would you test, and how much sleep would you let each group have?
• How many people would you have in each group?

How would you select the people?

We can choose our participants in many different ways. Things to consider are:

• Convenience: (you wouldn’t, for example, decide to study Japanese immigrants to Tanzania if you are a student at IST who speaks no Japanese and has no connection with the Japanese community here). You will probably use IST students. This is an opportunity sample.
• Numbers: if the numbers of participants are too small, then you cannot generalise your results (remember the ‘infection’ hypothesis concerning anorexia nervosa?) Optimum size is about 25 to 30 people.
• Sample: the particular group of people in which we are interested, like students, managers, or elderly people, is our target population. From them we choose our representative sample.
• Bias: a truly representative sample is an ideal. The best way to get near this is to take a random sample. Any ideas how we would get a random sample of: students on this campus, 5 year olds living in Dar es Salaam, male teachers over 40 years old?
• Stratified sampling: if you wanted a representative sample of students in this school, you might decide to take G6 —G12 students in proportion to their numbers in the school. For example, if G6 students make up 10% of the school population, then 10% of your sample will be made up of G6 students.
• Quota sampling: used most amongst market researchers. A certain number from each group that makes up the population being surveyed is taken. For example, if the target population is teenage girls at IST between 13 and 16 years of age, and you know that there are 120 13 year olds, 110 14 year olds, 90 15 year olds and 100 16 year olds, then a quota sample might be 12 who are 13, 11 who are 14, 9 who are 15 and 10 who are sixteen.

Data: there are 3 levels of data — nominal, ordinal and interval.

Some data is qualitative and some is quantitative. Qualitative data cannot be measured and give the answer to "how much". Think about the statement "I’m hungry". If I said "How hungry?" I would not expect the answer "9". Hunger cannot be quantified. If we produce a scale and try to quantify it, how do we know that one person’s 6 out of 10 (with 10 being the hungriest) isn’t the equivalent of another’s 2 out of 10? Quantitative data, like height, age, time taken to remember something, numbers of items remembered, etc. can be counted.

Different methods tend to produce different data. Unstructured interviews — as in a conversation — tend to produce qualitative data. Large-scale surveys produce quantitative data. Why? Observations produce either kind. Think of an observational study you might do that would produce both qualitative and quantitative data.

Go back to methods page

Go back to homepage