'Research methods' fall into two design categories in psychology. Research methods
that are experimental in design include the laboratory, field and quasi-experiment.
Non-experimental methods include the observational, survey, interview and case study
methods. Experimental methods produce measurable quantitative data. Non-experimental
methods can sometimes give you quantitative data but information is more likely to be
descriptive or qualitative in nature. The type of data produced by a particular
method affects the validity and reliability of the research results.
|Research Hypotheses: Two types
The experimental hypothesis, which only applies to laboratory, field, and quasi-experiments.
The correlational hypothesis: a statement predicting a relationship between two covariates
in a correlation.
Both types share the feature of the null hypothesis.
Null hypothesis is a statement of no effect (experiment), or no relationship (correlation).
Often tested to a degree of significance e.g. 0.05. By counter intuition if null hypothesis rejected,
experimental/correlational hypothesis is accepted/supported. Or vice versa.
|Samples & Sampling
The participant/subject selection technique that allows for generalisation of results onto the target population from
which your sample is drawn.
You can infer thoughts, feelings and behaviours from a sample and thus for example how
target population might think, feel, behave.
A sample must therefore be as representative of the target population as
possible. Done by adopting a particular type of sampling technique.
|Opportunity sampling sees your sample made up from whoever is available and around at the time.
If you need 20 participants an opportunity sample are the first 20 people you find willing to assist.
Random sampling is where every
member of a target population has an equal chance of being chosen to take part in your research e.g. 1: 10
Stratified sampling occurs
when you look at your target population and decide to make up a sample for your research reflecting the make-up of the target population.
|If the random sample is large enough, random
sampling gives the best opportunity for everyone in a target population to
participate in your research. In a target population of 2000, a representative
random sample of 1: 20 would be more representative than one of 1: 200. 1: 20
as a representative random sample of 2000 would be a unbiased random sample.
1: 200 would be a biased random sample.
|The bigger the target population the more
difficult it is to randomly sample in it. You are unsure who the target population
is. If it were a town of 60 000 for example, looking in the phone book and choosing
every 1000th person would be a problematic random sample. Not everyone is in the
phone book, people are ex-directory, thousands of people only use mobiles whose
numbers are not listed etc.
|When it is important that characteristics/subcategories/strata
of a target population be investigated stratified sampling is most useful. Stratified sampling
gives you a truly representative sample of your target population on the basis of those
identified characteristics you want to investigate.
|Stratified sampling is time consuming because characteristics
in the target population have to be identified, and a calculation of their ratio of occurrence
worked out. This is to ensure the correct ratios in your stratified sample.
|Opportunity sampling is extremely quick and economical.
It is the most common method of sampling because it is convenient.
|It is an unrepresentative method of sampling. There is a difficulty
when using opportunity sampling to generalise your results to a meaningful target population.
If your opportunity sample was 10 first year pupils from a large secondary school anything you
might infer from a survey could only be applied to this small unrepresentative group.
|Quota sampling is a quick and efficient way to gather information
on specific strata within a population. If you are a consumer intelligence firm and a client was a
large fashion chain catering to females in the 16-25 age group, quota-sampling females in this age
group above would be ideal from the point of view of efficient market research.
|How the quota sample is chosen is often left up to the researcher.
If 100 16-25 year old females were to be the quota sample, an opportunity sample of 100 16-25 year
old female students might be used. This quota would not reflect all 16-25 year old females in the
<<< Back to Research Methods Home
Forward to the correlational technique >>>