Teach Yourself Repertory Grid: Part the Third, Construct Elicitation (Two)

Teach Yourself Repertory Grid, Part the Third: More on Constructs and Laddering.

In this post I’m going to go into a little more detail about constructs: how to make sure that you’re getting constructs appropriate to your purpose, how you can ‘unpick’ a construct’s component parts and how you can use a person’s constructs to get closer to their deepest value system.

I’m hoping that you really did go through the various exercises I suggested in previous posts, and I hope you tried the ‘trick question’ – give me a construct that isn’t yours. It’s not possible. The best that anyone can do, if they’re very empathetic and they know their subject well, is to think themselves into the shoes of Elizabeth Bennet or President Nixon or Professor Dumbledore, but we also know that we’ll only be able to penetrate so far and that the characters will retain the capacity to surprise us. Your construct system is personal to you, and it represents the distinctions that you’ve needed to make about your world in order to navigate around it. So there are lots of Grid-based projects that depend on listening to a person’s (or an organisation’s) construct system.

So, let’s whet your appetite with a few examples of consultancy projects that have relied on construct elicitation to research the issues:

Three projects using construct elicitation for problem definition:

A: Middle Managers in an Oil Company.

In the 1980s I was responsible for a project in which middle managers in a very large oil company were asked to produce constructs about their colleagues (anonymously, of course). We interviewed over 200 managers – far more than was necessary – and then did a simple content analysis of the constructs. By far the largest group of constructs could be classified as ‘Knowing the right way to relate to Head Office,’ and the next largest were about extracting the same behaviour from their subordinates. Then there came some constructs about technical knowledge and analytical power. But there were no constructs at all – not one – that mentioned the customer; there were hardly any that mentioned relationships with other departments, team-work, and collaboration; and hardly any about financial discipline.

I’m afraid that this was one of those occasions where the company felt unable, or unwilling, to see the need to make some changes in how managers construed their own and others’ effectiveness. About 20 years later, they were involved in a huge scandal which the official enquiry traced back to an unwillingness of subordinate managers to give their bosses bad news, and a lack of interest in sharing information between the different functions in the business.

B: Managers in a Retail Bank.

This story has a happier ending. When I was introduced to the Bank, they knew that they had to make at least 30% of their managers redundant, plus they had to make the change from being credit experts to having retail skills and become aware of the competitive environment. Again, we interviewed a sample of managers asking them to produce constructs about colleagues (anonymously) and the first content analysis revealed some very interesting findings: there were five major construct groups, in summary:

Effectiveness (which was expressed as activity level, not skill)

Being Mr/Ms Nice Guy with the staff

Timetabling – you couldn’t call it Planning, it meant knowing what you’d be doing on Friday, nothing about priorities

Being a credit expert

Presence in the local community

We fed this information back to the senior managers in a session whose purpose was to ask ‘What’s the survival value of this formulation for effectiveness?’ and go on to plan for changing it. I’ll tell you more about this in a later post, but for the moment let’s say that the senior managers realised that there needed to be drastic changes in how effectiveness was perceived.

Jump to the end product: we ran Assessment Centres for all the existing managers (450 in total) based on the new formula for effectiveness. Although 30% of them were, in fact, offered redundancy, we had only three grievances raised and these were settled quickly and amicably; and an 18-month follow-up study by the local university showed a significant improvement in performance from the managers assessed on the new criteria.

C: Problems with Government Reform.

I spent a considerable slice of my life working with the New Zealand government in the 1980s and 1990s; it was painfully obvious that the way the government worked needed root-and-branch reform (New Zealand was headed to becoming the world’s first white third-world economy) and a small team of brilliant visionaries pioneered many changes in the management of the public service. One of these changes meant that Heads of Government Departments no longer were assured of jobs-for-life, but put onto performance contracts; they were given much more freedom to manage their departments than under the old, creaking bureaucracy; but many of them found it pretty stressful.

I did a nifty little construct elicitation study that compared the constructs of three different groups of people about the job of Chief Executive of a government department. One group consisted of ministers and control agencies (such as Treasury) to whom the Chief Executives reported; one group consisted of Chief Executives themselves; and the third group was composed of senior public servants who’d been around a long time, were perceptive, and didn’t have ambitions to become Chief Executives themselves. The results: Ministers and control agencies judged the Chief Executives on two major criteria – the quality of their policy advice, and not landing them in trouble. Whereas the Chief Executives and the Old Hands used these criteria, for sure, but also a major preoccupation was the business of managing an (often large) department. They were having to work it out for themselves – there were very few helpful examples from the private sector, the reforms were new and there was no established body of knowledge for them to consult, and their ministers and controllers (to whom they had to account for their stewardship of the country’s resources) had a blind spot about many of the issues that preoccupied them.


Enough examples to make your mouth water? Perhaps you’ll permit me to get back to some practice on the details of effective Grid interviewing. Here’s some guidelines:

Anal-Retentive Corner:

When I’m teaching Grid in a classroom, I make no apology for insisting as strongly as I can that people follow the rules … because if you’ve learned to follow the rules and do as ‘pure’ a Grid interview as possible, you can then depart from them when the project requires it in the full knowledge of what you’re doing. It’s like being a musician: you need to keep up with your scales and arpeggios, and if you’re a composer you need to understand the basic disciplines of key structure, harmony, etc., before you depart from them – otherwise you’ll simply produce a Confused Noise. So, forgive my emphasis on the rules; they’re there for a purpose.

Why does Grid use triadic comparison?

People often ask about the purpose of using three elements to generate a construct. Answer: the triadic comparison process is the only way you can be certain of getting two poles, so there isn’t really any consistently successful alternative to offering the elements in triads. However, should your interviewee produce bipolar scales by any other means, if they’re suitable to the purpose then accept them. What’s important is that …

Constructs should be bipolar.

A construct is a scale on which all, or nearly all, of the elements can be rated. Therefore it must have two poles. This may seem to be a very elementary place to start, but one occasionally sees ‘constructs’ described by just one word - for example ‘efficiency’ or ‘happiness’, usually in order to fit them into an oversimplified analysis. The reason why one-word descriptions aren’t a good idea is that …

Both ends of the construct should carry equal weight.

In other words, both poles should be equally well-defined. A construct like X - not X will prove awkward to use when rating; but more important is the fact that you want to know how the contrast pole is defined by the client. If one pole of the construct is creative the constrast pole might be practical, or dull, or realistic or disciplined ... or any of half a dozen other notions, depending on the interviewee's experience; and you will only know what’s meant by creative if you see its constrast pole. So it would not be useful to let pass a construct creative - not creative, nor to write the construct as creativity without bothering to write both poles. Ask for the other pole to be defined, with a question like ‘how would you describe the other(s) by contrast?’ (Don't, please don't, ask what's the opposite? because you might get the dictionary opposite, but not the quality that makes the other pole different by contrast).

Also, there’s a very old piece of psycholinguistic research in which the authors took a series of adjectives, such as happy, interesting, well, etc., and used the techniques of semantic differential analysis to examine how much ‘weight’ was implies when the opposite word was expressed in a variety of ways, e.g. not happy, unhappy, and sad; not interesting, uninteresting, and boring, etc. They found that in all cases the ‘Not X’ or ‘UnX’ formulation carried less information than the formulation X-Y, where Y is the dictionary opposite. This is one reason why good Grid practitioners always ask for the contrast pole to be expressed in its own right, rather than using the emergent pole and tacking a negative on the front of it. You want a scale, a construct, where both poles carry equal weight.

Don’t infer or supply the contrast pole.

If you don’t elicit the contrast pole from the interviewee, you are left to infer it; or if a number of people’s constructs are to be combined or shared, then it is up to someone else to infer it. And it could be dead wrong. For example, imagine that you are interviewing someone about her significant others and she produces one pole which she labels manic. Is this one end of a construct manic - depressive? Or is it one end of a construct manic - even-tempered? Her construct, which of course is personal to her and intended to reflect her experience, could be either, depending on how she experiences the other significant people in her life. If the fully-expressed construct is manic - depressive, then the even-tempered people in her life will be rated at the middle of the scale. If the fully-expressed construct is manic - even-tempered, then the even-tempered people in her life will be rated at the end of the scale. Of course, you can explore the range of her construct manic - ???? and later you may want to re-write or split it (wait until we chat about the Range of Convenience of constructs), but if you leave the contrast pole unexamined then you are not doing justice to the construct system.

Constructs should be appropriate to the purpose.

You need to ensure that you are exploring the domain of constructs appropriate to the purpose. For example, you could interview a manager about his or her team members with a counselling contract, in which case you would want constructs about how the manager feels about the team, relates to them; or you could use the same element set but with an agenda of drawing up a person-specification, in which case you would want constructs about performance. The way to take care of this is with the qualifiers - the ‘... in terms of...’ questions you ask when laying down the triads.

Physical presentation is important.

Until now, you’ve been working with just three elements. In practice, Grid interviews use a larger sample of elements (more on element selection in later posts) and I’ve seen people try to short-cut the process by giving their interviewee a list of elements and asking them to look at numbers 2, 5 and 7. This doesn’t work, even with very perceptive interviewees. It really is important to write the elements on cards so that they can be shuffled around, or otherwise present them three at a time. Seeing them in a physical relationship to one another makes the process much easier.

Full Context Presentation.

One way of being certain that you’ll get a construct is to place all the elements in front of the client and ask for the two which are most similar, and then ask what it is they have in common; then ask for the element which is most different on that dimension. This almost always breaks the log-jam, and if it doesn’t then you need to go back and think about whether the client is comfortable with the contract and the purpose.

Laddering (Up and Down):

If you go back to the second post, where you can eavesdrop on a real session using cars as elements, you’ll observe that at some point the interviewer takes a construct and asks which pole the person prefers, and why. This process is called Laddering Up, and it takes you (very quickly) to the interviewee’s core constructs. Give it a try with some of the constructs that you’ve already produced and see what happens … betcha that after three or four levels of ‘why?’ questioning you’ll be uttering your own version of ‘that’s how it is for me … that’s what I believe … that’s what I was always told …’

According to Kelly, everybody has a hierarchy of constructs, with core constructs at the centre and a network of constructs that eventually leads to the periphery. What’s important for now is that we don’t mess around with people’s core constructs unless it’s with their permission and we’re skilled counsellors. A person’s core constructs – their value system – are deeply-held beliefs, usually non-negotiable, and for most purposes folk like thee and me have got no business asking about them.

By the way, you can also ask the Laddering Up question by taking a construct and asking ‘Is this an important distinction to make about (your field of enquiry)?’ Variations on this theme include asking your interviewee to sort their constructs into High, Medium, and Low priority, and asking for the criteria used in this sorting; and you could keep your detailed questioning to the high priority constructs only.

Also in the interview about cars, you’ll hear the interviewer take the construct flashy- wolf in sheep’s clothing and ask for more detail about how cars that are flashy differ from cars that are wolf in sheep’s clothing … what would the person see or feel? And the response ‘unpicks’ the construct’s component parts: Flashy cars are covered in unnecessary gadgets; have auxiliary brake lights halfway up the back; have personalised number plates (yuck!); go-faster stripes; fancy hubcaps. Wolves in sheep’s clothing don’t have gadgets; orthodox brake lights; standard fittings.

This process is called Laddering Down, and it’s an essential skill. It’s the means by which you move from an abstract, high-level construct towards the features that you can actually see, hear, feel, touch – it’s how you answer the question What does this construct mean in practice? If you think about a construct system as a hierarchy, there’s no telling at what ‘level’ your two-against-one question is going to enter said hierarchy. Ask me about MONTY, JIM, and PAUL and I might respond with good drivers – bad driver, which is a pretty important construct but by no means as important to me as responsible – careless. If you laddered down from responsible – careless you’d get good drivers – bad driver, which for me is a sub-set of responsible – careless; and you could, of course, ladder down from good drivers – bad driver to learn more about what that distinction means to me.

How to write elements and constructs:

It’s worth keeping to a convention about how to write elements and constructs. In a printed text I always write elements in capitals, and constructs in italics. I’ve already mentioned the importance of writing each element on its own card or piece of paper; when writing constructs, I usually take a fresh index card (8” by 5” or thereabouts) for each construct and begin by writing the construct roughly half-way up the card, with the pair on the left and singleton on the right, with the numbers of the associated elements at the side. This gives you room to write any information from Laddering Up above the construct, and information from Laddering Down below it.

Why so many cards, you might ask? Well, because there are lots of applications of Grid that involve sorting the constructs into groups – by priority, or with a content analysis. Having each construct on its own card makes the sorting easier.


Homework:

We’ve got homework to do. It’s a rainy Friday afternoon, and I ought to figure out how to create a picture of a typical construct card and post it into the blog. That’s not going to be easy because (i) my camera’s broken, and (ii) if I can make a mess of doing something new, you can guarantee it’ll happen. Hold thumbs.

What I’d suggest for you is to get some practice in laddering up and down, and in writing the constructs. If you can persuade someone to let you practise on them, then use a set of more than three elements (nine is a good number, but they should all be from the same ‘family’) and notice how, with a bit of practice, you can co-ordinate your writing with the interviewee’s thinking. There’s likely to be a pause when you’ve presented a new triad, or shuffled the existing triad differently, and you’ll find that you can use this thinking time to tidy the construct card you’ve been working on and be ready to listen to the next.

Enjoy!!

Valerie.