Meths & Stats


Research design


 

 

Research manifesto


Statistics Index


CERG Home Page  

CERG Resources

Bibliography

Having a research question or hypothesis

Designing a research project is easier if you have a clear research question you wish to pursue - or a set of interrelated questions. You will find that most texts on design will emphasise the need for a research questions.

    If you start with a grape-shot, or scatter gun approach, you have less chance of producing clean and easily reported results.

      But where you are simply exploring an area and hope to focus in the next phase of the research, you can be less precise.

Cross-sectional versus longitudinal

The standard research design reported in the journals is a cross-sectional design.

    Data is collected from a sample at a given time and then comparisons are made using internal groupings.

    You may collect data on the same course in two semesters or years. You would then be comparing different people.

More sophisticated projects will use a longitudinal design.

    You have a longitudinal design if you collect data on the same sample of individuals, at a number of points.

    A project can be called longitudinal if you can analyse the way in which individuals change over time.

Designing for comparison

If you wish to make comparisons between groups or to investigate a given group over time then you must collect biographic information which will allow you to make comparisons. For example, if you wish to analyse gender differences, within a given intake, as they progress through the course, you have to ask for gender on each of your surveys. The exception would be where you are doing a true longitudinal study and are able to identify each individual over time.