On Research Design, Purpose, and Analysis

 

Now I’ve got all the information, how do I analyse it?

 

There are many, many different ways of analysing Grid data, from simple frequency counts to complex statistical matrix manipulations, and I don’t propose to list them all here. What I want to discuss is the importance of having a good research design so that you know which analysis method or methods will be the most suitable.

Grid is a powerful empty procedure

One of my mentors in Grid technique was Professor Laurie Thomas, who was then at Brunel University. Laurie was easily the best teacher of Grid I have known, and also one of the most inventive in term of research design. One of his key phrases was ‘Grid is a powerful empty procedure’; by which he meant not only that it was content-neutral, but also that whenever you do a Grid (unless it’s on a topic about which the subject has no experience at all) you will get data. The potential problem, which you have to anticipate by good research design, is that the data might not be suitable for your purpose.

What’s your purpose?

For example, you could interview managers about critical incidents in their jobs. The element set would be critical incidents. But what’s your purpose? Is it a counselling contract, where your job is to help the manager gain insight into what kind of incidents are the most troublesome?. Is it part of a competency analysis, where your job is to understand how people construe the demands on their skills? Is it part of a project to see whether there are any differences between people who find it easy to get on with new technology and those who don’t? Are you trying to find out whether the organisation is more or less stressed than it was two years ago? And so on.

Each of these purposes demands its own form of configuration and analysis. If this isn’t considered beforehand, they you stand a fair chance of finishing up with furniture-vans full of useless data.

Five thoughts on research design

So I’d like to offer five thoughts, thus:

1. Get to know the range of analysis programs which are available for analysing Grids. They’re all different; they each look at your data in different ways.
   
   
2. Focus on the design of the project. When we wrote the Business Applications of the Repertory Grid, which was published in 1988, the first half was in effect ‘teach yourself Grid’. The second half was about individual applications and purposes, and underlying each chapter there’s a more-or-less hidden theme which says ‘think how much better we could do this with a computer’.
   
   Nonetheless, what we learned was the huge variety of low-tech ways in which you could accomplish a Grid project when all you had was paper and pencil. They’re still valid. And of course they made us focus more clearly on the design of the project and the subsequent analysis because they made us think more clearly about the cost-effectiveness of our efforts. Because it made us think about the ‘front end’ of the project, we couldn’t kid ourselves that somewhere at the end of the rainbow was a computer program which would somehow impose order on badly-planned data.
   
   
3. Pilot, pilot, pilot. Don’t commit yourself to a huge research programme without first working through a small-scale version and seeing for yourself whether it looks as if it will work and meet your purposes. Before you use up other people’s time and resources (and create expectations which you may not be able to fulfill) work it through in miniature.
   
   For almost any Grid purpose, there are several different configurations which will do the job; so you can adapt. Anything other than to find yourself buried under useless data, probably followed by the discovery that things are no clearer once you’ve fed them into the computer.
   
   
4. Feedback. It doesn’t necessarily follow that relying on a computer program excises the message that Grid is a conversation, that a ‘first pass’ over the interviewees is just that and they should be given feedback and the chance to comment and the opportunity to help with interpretation. But it does no harm to stress its importance.
   
   I remember someone who’d collected 45 Grids from people all over the country, retreated into the computer, and was asking for help choosing a form of computer analysis. He didn’t have any plans for giving feedback. Not only was this discourteous to the people who’d given their time, it meant that he was cutting himself off from the chance to question them about the analysis and ask them what conclusions they could draw. He was cutting himself off from one of the best sources of analysis he had: which is not a computer program, it’s the people who gave you the data.
   
   
5. Vulnerability to a poorly-stated research hypothesis. Finally, Grid is a research method just like any other research method in that it is vulnerable to a poorly-stated research hypothesis. A poor hypothesis will lead to an inappropriate research design, difficulty with configuration and analysis, and little or nothing added to the collective wisdom.

Prepared by Dr Valerie Stewart

Search this Site

---

Related Resource

• The repertory grid technique: Its place in empirical software engineering research. Personal construct theory (applied via the repertory grid technique) supports interpretivist research in a structured manner and, as such, has relevance for researchers conducting studies focused on the human and organisational aspects of software engineering. Helen M. Edwards, Sharon McDonalda, and S. Michelle Younga, Department of Computing, Engineering and Technology, University of Sunderland, UK. David Goldman Informatics Centre, St. Peter's Way, Sunderland, UK.

Suggestions

• Download a free evaluation copy of Enquire Within here
• Email this page to a colleague.
• Post this page to: del.icio.us
• Want to know more? Join the experts at the RepGrid User Group.
• Want to keep in touch? Join our mailing list.
• A detailed log of developments at Enquire Within is at http://twitter.com/EnquireWithin. Join us there.
• Translate this page to your language? Try Google Toolbar.

Back to Hints Index