Visual Inspection in the Repertory Grid Interview Process

 

The suggestions which follow are only a selection of what you could do; and always remember that you and the interviewee must know your joint purpose and you have a good contract.

Use Cards

Arrange to collect your information in a way which makes it easy to move things around on a desktop. So: use small cards for your elements (one per card, of course) and larger cards for your constructs (one per card). Write each construct about halfway down the card, so that you can record the laddering up information above it, and the laddering down below it. If you think that your interview is going to go beyond construct elicitation and involve some rating, get yourself one or two pieces of stout paper with an empty Grid ready-drawn (leave space at the top and side for writing in the names of the elements and constructs). Apologies if this paragraph sounds too basic, but it comes from early experience of my own.

Elements and Constructs Will be Limited

Reconcile yourself to the fact that you probably won't be able to handle as many elements and constructs as you would in an unlimited Grid, and that most of your information will come from:

• content analysis of the elements (perhaps)
• content analysis of the constructs (certainly)
• comparing only the ratings assigned to elements, probably in pairs or threes

Constructs Are Bipolar and Not Always Evaluative

Visual comparison of constructs gets difficult because they are bipolar; analysis programs inspect to see whether you would get a better correlation by reversing the constructs, but you won't be able to do this visually. (Don't think you can get around this by asking for the 'positive' pole of the construct to be always on the right. Not all constructs are evaluative, nor should they be).

Content Analysis of Elements

Content analysis of elements can be useful if you are interviewing a large number of people, using element creation questions which remain the same between interviews.

For example, if you're interviewing all the members of a team to learn more about the difference between projects which went well and projects which didn't, you might have the first two questions as 'Name two projects which in your view were successful,' followed by two questions asking for projects which were less successful. Then you might ask for one or two examples of projects which they thought would be difficult but turned out to be easier, and vice-versa; a couple of routine projects; and one 'ad lib' which your interviewee thinks would help complete the picture of life in this team. Compiling the answers to each of these questions could give you some very useful information in its own right.

Content Analysis of Constructs

Content analysis of constructs is simple, and again can lead to instant insights. Ask your interviewee to arrange the constructs in groups which have some similarity to one another. (Or you can do it yourself, but remember that Grid is supposed to be a conversation). Then you inspect the groupings in terms of your purpose - looking for big groups, little groups, and (this is the most difficult, but often the most revealing) groups that aren't there at all.

As an example of 'missing' groups, I'm fond of quoting a construct elicitation project which we did with 200 middle managers in a oil company in the 1980s. When the team began the content analysis, I put £5 on the table for the first person whose data included any mention of the customer. It remained unclaimed.

Group Size and Reliability

As with content analysis of elements, content analysis of constructs is more dependable if you have interviewed a number of people from a homogeneous group, rather than just one person, but you and the interviewee can learn a lot from the content analysis of one interviewee alone. (By the way, the figure of 200 middle managers wasn’t our idea. With a homogenous group of people and the same purpose you stop getting much new data about after 20 interviews. The client wanted 200 and this in itself illustrated the risk-averseness which was part of their problem).

Working With One Person

So, if you're doing content analysis of just one person, this is where you can ask some questions depending on your purpose: for example, do the proportions feel like an accurate reflection of the way the interviewee thinks about the world? Is there anything important missing? What's the highest priority group? You can ask for the constructs to be arranged into three piles - high, medium, and low priority - and ask why the interviewee assigned them like that. At that point, good interviewing skills take over.

Comparing one or two elements is the other option. You do this physically; once you have all the elements rated on all the constructs, you prudently make a photocopy or two. (Note: consider using a simple tick/cross - in effect a two-point scale - for your rating. It loses detail but makes the visual inspection very much easier). Then with your trusty scissors you cut out the description of one element and move it next to the element you want to compare it with. At this point the need for good planning of the session becomes apparent, because you need to think about how many of these physical comparisons will be useful.

So, for example, if the purpose of your interview is to help someone whose parents seem to have unrealistic expectations of him or her, you would need to include the elements MYSELF, MYSELF AS MY PARENTS WANT ME TO BE, and perhaps MYSELF AS I WANT TO BE. The other elements in the set would obviously be the person's significant others, but in a sense they would be there as 'carriers' to allow you to get a good range of constructs; you would start work by comparing these three elements and discussing what the comparisons meant, who had the problem, etc. - i.e. moving into a classic counselling interview.

Other inventive ideas might suggest themselves to either of you - for example, creating and rating a completely new element, or cutting all the elements into columns and arranging them in families. But your planning of the session should bear in mind that you want most of the 'weight' of the conversation to come from a small number of element comparisons, and choose your elements accordingly.

There Once Were No Computers

Those of us Gridders who find that the policemen are looking younger nowadays can remember times when this was all we had. Nonetheless, we managed to get some good work done. Hope this helps.

Prepared by Dr Valerie Stewart

Search this Site

---

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