6.1.2 Simple Relationships Between Constructs

К оглавлению
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 
17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 
34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 
51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 
68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 
85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 
102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 
119 120 121 122 123 124 125 126 127 128 129 130 131 

When you examine the relationships between elements, you’re addressing

similarities in what’s being said about the elements. Similarly, it’s possible to

look at the relationships between constructs; but here, you’re addressing

similarities in how the interviewee talks about the elements.

Is s/he using the various constructs in similar ways? Do the constructs

represent very different aspects of the interviewee’s thinking, or do you get the

impression that they’re simply minor flavours of one rather obsessive theme

which seems to underlie the way in which the interviewee thinks about that

particular topic?

And that needs stressing. These are all questions dealing with a person’s thinking

about a particular topic; but, because they deal in‘hows’ rather than‘whats’, it’s very

tempting to feel that you’re discovering something about the person him- or herself,

rather thansomethingmore straightforwardabout theperson’sviewsonthetopic.You

may begin to overgeneralise, and start talking about the interviewee’s style, or

personality, both beingrathermore enduring characteristics.

Grids were developed as a way of describing individuals and the characteristic, and

differing, ways in which they construe experience. In that sense, they have been

understood by some psychologists as a form of personality assessment. I’d suggest

that you don’t think of what you’re doing in this way, though. No doubt, if you were to

elicit several grids on different topics from the same person, and found the same

kinds of constructs, hanging together in the same way, you’d be inclined to feel that

you were getting a good idea of the other person as a unique human being. Now, if

you were to spend more time with them, talking in ways that are more spontaneous

and natural than a grid, and possibly, if you had the appropriate training in using

other techniques andmaking clinical judgements, you could certainly describe what

youwere doingasa formof personalityassessment.And there’sno doubt that the grid

is an excellent technique for doing so.

But if you’re simply interviewing a person on one topic, on one occasion, well, then,

that’s all you’re doing. If you choose to see it as a personality assessment, that’s your

affair. I’d rather, though, that you were to think of what’s going on in thisway. In making

their tacit knowledge explicit, you’re construing their construing.Would other people

agree with you? That includes the interviewee! Remind yourself of what was said in

Section 5.2, which provides a different slant on the same issue, and keep checking. I’ll

suggest ways of doing so below.

As with elements, so with constructs: a simple examination of relationships is

an optional part of the analysis, and you might instead choose a more detailed

structural analysis.

You’d choose to analyse simple construct relationships for the same sorts of

reasons:

. because you didn’t have access to a computer

. because you’re uncomfortable with number-crunching any more involved

than the calculation of simple similarity scores

. because you want to explore the relationships between constructs with this

interviewee.

Your approach is very similar to the approach you adopted in looking at the

elements. You make inferences from differences between ratings, but, this time,

you’re counting along the rows of the grid, rather than down the columns.

Take a look at Table 6.4. This is an attempt to make tacit knowledge explicit: to

get inside a department store manager’s head, encouraging her to be explicit

about what it takes to be an effective sales clerk. Do a quick eyeball analysis to

familiarise yourself with it. Particularly, look at the constructs, row by row. Do

you notice anything about the ratings on the first construct, and the fourth one

from the top? Both constructs are shown highlighted in boldface.

Practically identical! Element by element working from left to right, whatever

rating is given to ‘learns the new models quickly’ is also given to ‘awareness of

sizes, colours, and availability’ – with the exception of a difference of ‘1’ in the

case of the third element along from the left, Billie. Let’s examine the two

constructs in detail.

In the example shown in Table 6.5, the sum of the differences, construct 1

against construct 4, is just 1. This is recorded in a new section of the table, in

the right-hand half of the table. (Read along the construct 1 row and find the

intersection with the construct 4 column.)

Is the interviewee using these two constructs as equivalents? Are they linked

in some way, or is it just a coincidence? These and other fascinating questions

can be asked as you explore the interviewee’s knowledge structure. First,

though, let’s run through the procedure. It’s very similar to the one you used

for elements, though it varies in one important point.

(1) Calculate difference in ratings of the first element on the first pair of

constructs. Working with the full grid, as in Table 6.4, take construct 1 and

construct 2 (the top row and the next one down), and find the absolute

difference between the two ratings on the first element.

(2) Sum across the page for the remaining elements. Do the same for the

other elements, working systematically from left to right, cumulating as you

go. Put this sum in the appropriate space at the right, on the intersection of the

appropriate row and column.

(3a) Repeat for all pairs of constructs. In other words, for rows 1 and 3, 1 and

4 . . . 2 and 3, 2 and 4 . . . etc., noting the sums on the right as you go.

Table 6.6 shows the results of these first three steps. The sum of differences for

construct 1 against construct 2 is 10; for construct 1 against construct 3 is 18; for

construct 1 against construct 4 is . . . (work it out for yourself: 575 plus 171

plus 271 plus 171 plus 474 plus 272 equals) . . . 1. And so on.

And in among all these sums of differences, we see that the two constructs we

identified by eyeball inspection earlier are indeed very highly matched for the

manager: the sum of differences between construct 1 and construct 4 is just 1,

Table 6.4 Grid interview with the manager of the clothing section of a department

store (an attempt to get at her tacit knowledge of effective sales performance as part of

the development of a competency system)

1 Jane Ann Billie Ian Alma May 5

Con 1 Learns the new

models quickly

5 1 1 1 4 2 Takes a while to learn

the features of new

lines

Con 2 Too forward in

pushing a sale:

tends to put

customers off

3 4 3 1 2 1 Good balance between

active selling and just

being helpful

Con 3 Could be more

interested in

after sales

1 5 4 4 1 3 After sales (alterations,

other bespoke

elements) well handled

Con 4 Awareness of

sizes, colours,

availability

5 1 2 1 4 2 Availability and

choice knowledge

poor

Con 5 Pleasant and

easy-going

3 1 2 5 4 3 Takes it all very

seriously

Con 6 Overall, an

effective

salesperson

5 1 2 4 4 2 Overall, a less effective

salesperson

Table 6.5 Grid interview with the manager of the clothing section of a department store, examining the simple relationship

between two constructs

Simple construct analysis

1 Jane Ann Billie Ian Alma May 5

Against

Con 1

Against

Con 4

Con 1 Learns the new

models quickly

5 1 1 1 4 2 Takes a while to learn the

features of new lines

– 1

Con 4 Awareness of sizes,

colours, availability

5 1 2 1 4 2 Availability and choice

knowledge poor

Table 6.6 Grid interview with the manager of the clothing section of a department store, examining the simple relationship between

constructs

Simple construct analysis

1 Jane Ann Billie Ian Alma May 5

Against

Con 1

Against

Con 2

Against

Con 3

Against

Con 4

Against

Con 5

Against

Con 6

Con

1

Learns the

new models

quickly

5 1 1 1 4 2 Takes a while to

learn the

features of new

lines

– 10 18 1 8 4

Con

2

Too forward in

pushing a sale:

tends to put

customers off

3 4 3 1 2 1 Good balance

between active

selling and just

being helpful

– 10 9 12 12

Con

3

Could be more

interested in

after sales

1 5 4 4 1 3 After sales

(alterations,

other bespoke

elements) well

handled

– 17 12 14

Con

4

Awareness of

sizes, colours,

availability

5 1 2 1 4 2 Availability and

choice knowledge

poor

– 7 3

Con

5

Pleasant and

easy-going

3 1 2 5 4 3 Takes it all very

seriously

– 4

Con

6

Overall, an

effective

salesperson

5 1 2 4 4 2 Overall, a less

effective

salesperson

and that’s the lowest sum of differences in the grid. There certainly seems to be

some shared meaning here.

Moreover, if we want a single indicator of how she construes ‘effectiveness

overall’ (a construct which happens to have been supplied by the interviewer –

construct 6), we can see that the construct which is being used most similarly

is construct 4. (Look in the rightmost column at the right-hand side of the

table, the column for construct 6; and glance downwards) ‘awareness of sizes,

colour, and availability versus availability and choice knowledge poor’ has the

lowest sum of differences of just 3. This supervisor appears to construe

effective sales performance in terms of the extent to which the salesperson is

informed about what’s available. (There is more on this kind of analysis,

against overall, summary constructs, in Section 7.3.2.)

I suggest you take a break now. You don’t have

to do an exercise at this point, but you do need

to take some time out. Put the kettle on. Take

the dog for a walk. Whatever appeals.

Now let’s consider a small additional step: a very important way in which the

procedure for analysing construct relationships differs from the procedure,

discussed earlier, for analysing element relationships.

(3b) Repeat step 3 for all pairs of constructs with one set of ratings reversed.

Unlike an element, a construct is bipolar, and can express the same meaning

with ratings which run from ‘1’ to ‘5’, or with ratings which run from ‘5’ to ‘1’,

so long as the words at each pole of the construct are reversed from left to

right. Your construct analysis needs to take this into account.

Confused? Let’s put that in a different way: take a look at Table 6.7. Suppose

you’re doing a grid in which the elements are people known to the

interviewee, ‘my friends’ being the topic. If you run through steps 1 to 3 for

these two constructs (272) + (571) . . . and so on, you’ll get a sum of

differences of 10. Quite a large difference, really. There seems to be little

relationship between these two constructs.

Table 6.7 Relationship between two constructs about six people

E1 E2 E3 E4 E5 E6

Against

Con A

Against

Con B

1 5

[ [ [

Con A Friendly 2 5 4 1 3 4 Remote – 10

[ [ [

Con B Shy 2 1 2 5 3 4 Outgoing –

But do you remember the basic grid elicitation procedure (Section 3.1.2)?

Whatever the two elements have in common is written down on the left, and

the contrast to this, which characterises the odd one out, is written down on

the right. For the first construct, the triad E1, E3, and E4 was offered, E1 and E4

were said to be similar in that they are both rather ‘friendly’, and so this

emergent pole was written down on the left, defining the ‘1’ end of the scale

for the first construct.

Likewise, elements E2, E3, and E4 were offered for the second construct, E2

and E3 were reported as similar in that they are both ‘shy’, and these were

written down on the left, defining the ‘1’ end of the scale for the second

construct.

Now, suppose that the triad offered in the case of the second construct had

been E2, E4, and E6? (Remember, which triads are offered is arbitrary: all that

matters is that they should present different combinations of elements each

time, to encourage fresh constructs.) Well, if E4 and E6 are both construed as

relatively outgoing (as indeed they are: see Table 6.7) as opposed to E2, then

‘outgoing’ would have been written down on the left, and would have defined

the ‘1’ end of the scale.

The meaning being conveyed by the second construct would then have been

expressed by the ratings as shown in Table 6.8. Aha! They are, indeed, strongly

related, with a sum of differences of just 4! Nothing has changed. Both versions

express the same meaning.

An element that is rated ‘5’ on a scale that runs from ‘shy = 1’ to ‘outgoing = 5’

must receive a rating of ‘1’ if the scale runs from ‘outgoing = 1’ to ‘shy = 5’, if

the same meaning is to be expressed. And so on, for all the other ratings.

But the first variant obscured the relationship that was there, for the entirely

arbitrary reason of the particular triad that was presented when construct B

was elicited.

Table 6.8 Relationship between two constructs about six people, showing a reversal

E1 E2 E3 E4 E5 E6

Against

Con A

Against

Con B

1 5

[ [ [

Con A Friendly 2 5 4 1 3 4 Remote – 4

[ [ [

Con B Outgoing 4 5 4 1 3 2 Shy

And so, when calculating construct similarity, you have to do it twice, once with

each construct as it stands; and once with the ratings of the first construct of

the pair being the same, but the rating of the second construct of the pair being

‘reversed’. On a 5-point scale, 1 reverses to 5, 2 to 4, 3 stays the same, and so

on: whatever the rating is, just subtract it from 6 to reverse it. (If you happened

to be using a 6-point scale, you’d subtract the rating from 7 to reverse it; on a 7-

point scale subtract the rating from 8 to reverse it; etc.)

Now return to our main worked example, the grid about sales clerks’

performance. Take a look at Table 6.9. This is the same as Table 6.6. The sums

of differences for each construct compared with each construct, both being

‘unreversed’, are shown, as before, in the top-right corner of the right half of

the table. However, the spare space in the right half of the table, below the

diagonal indicated by the dashes, has been used to record the sums of

differences for each construct ‘unreversed’ compared with each construct

‘reversed’. The actual reversed ratings are shown in parentheses below the

unreversed ratings in the left half of the table.

This may seem like a lot of calculations, but it isn’t, really. You’d calculate the

unreversed comparisons as normal. Then you’d reverse the ratings of

construct 1 and work out sums of differences between those, and all the

remaining constructs from construct 2 onwards. Then you’d reverse the

ratings of construct 2 and work out the sums of differences between those, and

all the remaining constructs from construct 3 onwards: it’s quicker than you

think. The result would be a table that looks like Table 6.9.

(4) Compare these sums of differences, the smallest sum of differences, and

the largest sum of differences, unreversed and reversed. The smallest

difference, indicating the two constructs which are being used most similarly,

and the largest difference, indicating the two whose meanings are most

dissimilar, are particularly useful to examine. Search for these, smallest and

largest, in the whole of the right-hand side of the table, and note whether the

relationship is an unreversed or a reversed one.

If the smallest sum of differences was from a reversed comparison, make sure

you swap the two ends of the constructs round when you’re reporting the two.

In Table 6.9, for example, while the smallest difference, indicating the two

most highly matched constructs, is, as we already know for constructs 1 and 4,

1. Learns the new models quickly – Takes a while to learn the features of

new lines

4. Awareness of sizes, colours, etc. – Availability and choice knowledge

poor

there are some quite small differences, indicating a strong relationship,

between reversed constructs. Take a look at the sums of differences for

Table 6.9 Grid interview with the manager of the clothing section of a department store, examining the simple relationship

between constructs, and showing reversals

Simple construct analysis

UNREVERSED

1 Jane Ann Billie Ian AlmaMay 5

Against

Con 1

Against

Con 2

Against

Con 3

Against

Con 4

Against

Con 5

Against

Con 6

Con

1

Learns the new

models quickly

5

(1)

1

(5)

1

(5)

1

(5)

4

(2)

2

(4)

Takes a while to

learn the features

of new lines

– 10 18 1 8 4

Con

2

Too forward in

pushing a sale:

tends to put

customers off

3

(3)

4

(2)

3

(3)

1

(5)

2

(4)

1

(5)

Good balance

between active

selling and just

being helpful

12 – 10 9 12 12

Con

3

Could be more

interested in

after sales

1

(5)

5

(1)

4

(2)

4

(2)

1

(5)

3

(3)

After sales

(alterations, other

bespoke

elements) well

handled

4 12 – 17 12 14

Con

4

Awareness of

sizes, colours,

availability

5

(1)

1

(5)

2

(4)

1

(5)

4

(2)

2

(4)

Availability and

choice knowledge

poor

19 11 3 – 7 3

Con

5

Pleasant and

easy-going

3

(3)

1

(5)

2

(4)

5

(1)

4

(2)

3

(3)

Takes it all very

seriously

12 4 6 11 – 4

Con

6

Overall, an

effective

salesperson

5 1 2 4 4 2 Overall, a less

effective

salesperson

16 8 4 15 14 –

R

E

V

E

R

S

E

D

construct 3 and construct 4, for example. Unreversed, the sum of differences is

17, practically negligible, you would imagine. But if you reverse one of them,

3. Could be more interested in after – After sales (alterations, other

sales bespoke elements) well handled

4. Reversed

Availability and choice knowledge – Awareness of sizes, colours,

poor availability

you can see that the two constructs are highly related, with a sum of

differences of just 3. Whenever the manager thinks of a sales clerk as ‘lacking

interest in after sales’, she also tends to think of them as having relatively ‘poor

knowledge of availability and choice’, and whenever she thinks of them as

‘handling after sales well’, she also construes them as ‘aware of sizes, colours,

and availability’.

(5) Discuss these relationships with the interviewee. And that would be the

way to put things. ‘Have you ever noticed that, whenever you say ‘‘x’’, you

tend to say ‘‘y’’?’ Your interviewee may find this unremarkable, since it’s as

plain as a pikestaff to him/her and always has been. ‘Of course!’ Alternatively,

the relationship may come as a surprise, since the knowledge involved is tacit

but should, on reflection, make sense. After all, your analysis is based directly

on the ratings you were given by the interviewee.

In other words, as with the elements, there should be some sense of ownership

of the relationships you indicate. (If there isn’t immediately, point to the

ratings for the particular two constructs and go over them pair by pair.) This

kind of analysis is particularly fruitful in counselling, guidance, personal

development, and decision-making situations, as you explore the ramifications

of the interviewee’s thinking, and point to the implications of those particular

ways of construing.

If you think about it, this is an excellent technique to use in counselling, since you’re

pointing out the implicational connections of your interviewee’s own construct

system directly, rather than doing what usually happens; that is starting with your

own construct system to identify implications in the interviewee’s construct system!

Jankowicz & Cooper (1982) isworth looking at in this respect.

There are three possibilities to examine. Glance at Table 6.9 again, where the

two most highly matched constructs are, as we noted, construct 1 and

construct 4 (unreversed). The fact that these two constructs are highly matched

(a small sum of differences) may reveal:

(a) a belief about the causal influence of construct A on construct B. Thus, the

manager believes that the sales assistant who learns the new models

quickly is, as a result, aware of the different sizes and colours in the range.

 (b) alternatively, a belief about a causal relationship which runs the other way

round: construct B influences construct A. For example, the manager may

be telling us that she believes that sales clerks’ general trade knowledge

about what kinds of sizes and colours are available at different times of

year will help them to learn this year’s new models quickly.

(c) that the high match is a sheer coincidence. The manager’s ratings on both

constructs may be unrelated. Other factors may influence the values they

take. How quickly a sales assistant learns new lines may relate to their

intelligence as perceived by the manager – how quick they are as learners

– while the awareness of sizes, colours, and availability may reflect

something very different; for example, how long the sales assistants have

been working for this particular company.

And, as you will expect by now, the best way to examine the possibility of

associations (and, just possibly, causal relationships of this kind) is to draw

them to the interviewee’s attention and ask which one might be the case!

Let’s continue with our standard procedure for examining the simple

relationships between constructs.

(6) Focus on relationships with supplied constructs, if any. As suggested in

Section 4.2.7, these will most probably be of two kinds.

(a) Relationships with a supplied construct which summarises the topic of the

grid: the ‘overall summary’ construct. A performance-appraisal grid for a

PA job, for example, might have eight PAs as its elements, and the

qualifying statement ‘from the point of view of what it is that they do that

makes them effective in their job’. After eliciting the interviewee’s own

questions, you would ask him/her to give an overall impression of each

PA on the supplied construct, ‘overall, more effective – overall, less

effective’.

You would then look to see which constructs had ratings which were

similar to this one. In a sense, this is what the interviewee means when s/he

characterises someone as, overall, an effective, or less effective, employee –

his or her implicit definition of effectiveness in that particular job. (Do this

with 20 other interviewees and you have the makings of a usable

performance-appraisal questionnaire: see Section 7.2.3.)

(b) Relationships with a supplied construct which summarises some belief of

your own which you want to test about your interviewee’s construction

system.

This may be a relatively informal impression you’ve gained from the earlier

constructs offered (e.g. ‘she seems to think of people largely in terms of how fit

and healthy they are; if I’m right and I supply her with a ‘fit – unfit’ construct,

it should have similar ratings to the others’).

Alternatively, it may be a more formal belief, identified before you ever met

this, or other, interviewee, which you wish to test as part of a research project.

For example, you may be interested in the tacit knowledge which bankers

have about business lending. A reasonable way of researching this would be to

ask, ‘Which kind of constructs match closely with the supplied construct, ‘‘I

would lend them money – I wouldn’t lend them money’’, when bankers think

about different examples of business loan applications?’

Indeed, it’spossible to identify the constructswhich bank commercial lending officers

on the one hand, and venture capitalists on the other hand, have in mind when they

think of good prospects. It turns out that these are not simply the‘objective’ factors to

dowithbalance sheets, businessplans, profit andlossaccounts, etc.Of course, these

‘objective’ indicators play an important part: but only in getting the applicant through

the banker’s door, as it were.After that, the lending officer’s knowledge, expressed in

the form of intuitions, and judgements of more ‘subjective’ factors play a huge part.

Moreover, bankers and venture capitalists differ in the constructs they associate with

a good prospect as opposed to a poor prospect. If this interests you, have a look at

Jankowicz (1987b) for bank lenders, and Hisrich & Jankowicz (1990) for venture

capitalists.

As you may recall from Section 5.3.3, propositional constructs are often used in

this way.

At this point, it would be useful to have a little practice at computing construct

difference scores and handling reversal.

Please do Exercise 6.4 before continuing.

(7) Ensure comparability with other grids. Finally, you may need to turn

your sums of differences into % similarity scores, just as you did for

element comparisons, and for a similar reason; that is, you want to

compare similarities in grids which have different numbers of elements. You

may be talking to a trainer about the teaching aids she uses before

attending an in-service continuing development course, and after attending

it.

When you repeat the grid after she’s attended the course, to see what, if any,

impact the course has had, you may find she wants to talk about two new

teaching aids she’s played around with during the course, resulting in a grid

which has two additional elements.

And so, because the sums of difference scores are based on a sum across a

different number of elements, they can’t be compared directly but must be

turned into a % similarity score. The procedure is almost identical to the one

you followed for element % similarity scores, as follows:

. The largest possible rating is LR71 as before, but this time multiplied by the

number of elements, E. So the proportion of actual sum of differences to

largest possible sum of differences is

SD

рLR _ 1Ю _ E

. Now, remembering that constructs have two poles, you need some way

of signalling that a reverse relationship may be involved. The

conventional way to do this is to spread the range of possible

percentages over a 200-point scale; this means that you multiply the

proportion by 200, not 100.

SD

рLR _ 1Ю _ E _ 200

or, if you prefer it all on one line, {SD/[(LR71)6E]}6200.

. And so, when you turn this percentage sum of differences into a % similarity

score by subtracting it from 100 (and not 200!), you achieve two things. You

turn percentage difference into percentage similarity, and you scale the

range of possible scores neatly between 7100 and +100.

100 _

SD

рLR _ 1Ю _ E _ 200

or, all on one line, 1007({SD/[(LR71)_E]}_200).

And there you have it. If you check back to Section 6.1.1, step 7, you’ll see how

similar it is to the procedure for calculating element % similarity scores. Now

you can compare the extent to which pairs of constructs are matched, in grids

which have differing numbers of elements.

As before, I’ve provided a ready reckoner to save you some time in doing the

calculations which turn sums of differences into % similarity scores. It’s in

Appendix 4. If the grids you’re working with are different, you’ll need to

calculate the % similarity scores yourself. Follow the bulleted procedure

shown above, or just use the formula directly.

Table 6.10 shows the same grid as Table 6.9, with some of the sums of

differences turned into % similarity scores. That sum of differences of 1 shown

in Table 6.9, between

Learns the new models quickly – Takes a while to learn the features of

new lines

and

Awareness of sizes, colours, – Availability and choice knowledge

availability poor

corresponds to a % similarity score of just under 92%, 91.67% to be absurdly

precise, in the first row and the fourth column in the right-hand side of the

table.

The very large difference of 18 between

Learns the new models quickly – Takes a while to learn the features of

new lines

and

Could be more interested in – After sales (alterations, other

after sales bespoke elements) well handled

in Table 6.9 translates to a % similarity score of 750% in Table 6.10, which

makes you suspect that reversing one of these two constructs may show a

higher relationship. Indeed, you can see from Table 6.9, in the ‘reversed’ part

of the table, that there is a sum of differences of 4 between

Learns the new models quickly – Takes a while to learn the features of

new lines

and

After sales (alterations, other – Could be more interested in after

bespoke elements) well handled sales

which corresponds to a % similarity score of just under 67% in the ‘reversed’

part of Table 6.10. The relationship between these two constructs should

indeed be read this way round, reversed.

Now do Exercise 6.5 and complete Table 6.10!

In doing the exercise, you may have noticed that the values of % similarity

scores for reversed constructs aren’t symmetrical about the zero value with

respect to unreversed constructs. That’s simply a consequence of the way in

which the maximum possible difference is expressed in the formula, and

doesn’t matter. The thing to hold on to is that you always choose the higher of

the two values – the bigger and more positive, the better – in expressing the

relationship between any two constructs, regardless of whether they lie in the

unreversed or the reversed side of the table. Then, if the value in question

Table 6.10 Grid interview with the manager of the clothing section of a department store, examining the simple relationship

between constructs, and showing reversals (% similarity scores)

Simple construct analysis

UNREVERSED

1 Jane Ann Billie Ian AlmaMay 5

Against

Con 1

Against

Con 2

Against

Con 3

Against

Con 4

Against

Con 5

Against

Con 6

Con

1

Learns the new

models quickly

5

(1)

1

(5)

1

(5)

1

(5)

4

(2)

2

(4)

Takes a while to

learn the features

of new lines

– 16.67 750.00 91.67 33.33 66.67

Con

2

Too forward in

pushing a sale:

tends to put

customers off

3

(3)

4

(2)

3

(3)

1

(5)

2

(4)

1

(5)

Good balance

between active

selling and just

being helpful

0 – 16.67 25.00 0

Con

3

Could be more

interested in

after sales

1

(5)

5

(1)

4

(2)

4

(2)

1

(5)

3

(3)

After sales

(alterations,

other bespoke

elements) well

handled

66.67 0 –

Con

4

Awareness of

sizes, colours,

availability

5

(1)

1

(5)

2

(4)

1

(5)

4

(2)

2

(4)

Availability and

choice knowledge

poor

758.33 8.33 –

Con

5

Pleasant and

easy-going

3

(3)

1

(5)

2

(4)

5

(1)

4

(2)

3

(3)

Takes it all very

seriously

0 66.67 –

Con

6

Overall, an

effective

salesperson

5 1 2 4 4 2 Overall, a less

effective

salesperson

733.33

R

E

V

E

R

S

E

D

comes from the reversed side of the table, you would report the relationship

with the direction of one of the two constructs reversed.