# APPENDIX 7

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

AIDE-MEґMOIRE/SUMMARY OF GRID PROCEDURES

This appendix presents the various procedures outlined throughout the

chapters in one place for easy reference while you’re learning them. With

practice, you’ll be able to depart from them, and know the good reasons why

you’re departing from them. Initially, though, each procedure is available in

outline form for reference. You’ll need to check back to the full presentation

occasionally, since some of the outlines are very sparse.

Basic Repertory Grid Elicitation

Section 3.1.2

(1) Agree a topic.

(2) Agree a set of elements.

(3) Explain that you wish to find out how your respondent thinks about the

elements.

(4) Taking three elements, ask your respondent, ‘Which two of these are the

same in some way, and different from the third?’

(5) Ask your respondent why: ‘What do the two have in common, as

opposed to the third?’

(6) Check that you understand what contrast is being expressed.

(7) Present the construct as a rating scale.

(8) Ask your respondent to rate each of the three elements on this scale,

writing the ratings into the grid as s/he states them.

(9) Now ask the respondent to rate each of the remaining elements on this

construct.

(10) Your task is to elicit as many different constructs as the person might

hold about the topic. So, repeat steps 4 to 8, asking for a fresh construct

each time, until your respondent can’t offer any new ones.

Laddering Down

Section 4.4.1

Laddering down is used at stage 6 of the basic grid elicitation procedure.

6b Put a ‘how’ question to the interviewee about the emergent pole of the

original construct.

. Write the answer down below the emergent pole.

. Put a ‘how’ question to the interviewee about the implicit pole.

. Write down the answer below the implicit pole.

. Stop at that point, or repeat the ‘how’ question in more detail still, about the

construct you’ve just written down.

. Apply the remaining steps of the basic grid procedure to the final construct

you arrive at.

Pyramiding

Section 4.4.2

Pyramiding is used as an alternative to laddering down at stage 6 of the basic

grid elicitation procedure.

6b Put a ‘how’ question to the interviewee.

. Write the answer down below the emergent pole of the original construct.

. Ask what is the opposite or contrasting pole of the construct you wrote

down above, and write it down.

. Go back to the implicit pole of the original construct and ask a ‘how’

question to the interviewee about it.

. Write the answer down below the implicit pole of the original construct.

. Ask what is the opposite or contrasting pole of the construct you wrote

down above, and write it down.

. Stop at that point, or repeat the ‘how’ question in more detail still, about the

constructs you’ve just written down, noting the emergent poles, and asking

in each case what their opposite is.

. Apply the remaining steps of the basic grid procedure to all of the constructs

you arrive at.

Eyeball Analysis

Section 5.3.2

Eyeball analysis is used as a preliminary to any other analysis technique.

. What is the interviewee thinking about?

. How has the interviewee represented the topic?

. How does s/he think: what are the constructs?

. What does s/he think: how have the elements been rated on the constructs?

. Examine the supplied elements and constructs and their ratings.

. Draw conclusions.

Characterising Constructs

Section 5.3.3

The following four numbered steps apply in adopting a systematic approach

to analysing the following kinds of constructs: core versus peripheral;

propositional versus constellatory; constructs used pre-emptively; affective,

behavioural, evaluative, and attributional.

(1) Identify constructs which appear to have that characteristic.

(2) Assess the proportion with that characteristic among the others in the

whole grid.

(3) Ascribe significance to this proportion, in context.

(4) Examine relationships: how do these particular constructs relate to other

constructs?

The particular way in which relationships are assessed varies depending on

which sort of construct is being dealt with: see Section 5.3.3.

Simple Relationships Between Elements

Section 6.1.1

This is a suggested approach to the use of element difference sums and

% similarity scores.

(1) Calculate differences in ratings of the first pair of elements on the first

construct.

(2) Sum down the page over the remaining constructs and note the total.

(3) Repeat for all pairs of elements.

(4) Compare the sums of differences, especially the smallest and largest.

(5) Discuss these relationships with the interviewee.

(6) Examine relationships with supplied elements, if any.

(7) Ensure comparability with other grids by turning the difference scores

into % similarity scores.

Use the formula:

element % similarity ј 100 _ рfSD=ЅрLR _ 1Ю _ C_g _ 100Ю

where SD is the sum of differences, LR is the largest rating possible on the

scale, and C is the number of constructs in the grid; or refer to the table in

Appendix 3.

Simple Relationships Between Constructs

Section 6.1.2

This is the analogous procedure for constructs.

(1) Calculate the differences in the ratings of the first element on the first

pair of constructs.

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

(3a) Repeat for all pairs of constructs.

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

(4) Find the smallest sum of differences among all the comparisons you’ve

made, unreversed and reversed.

(5) Discuss these relationships with the interviewee.

(6) Examine relationships with supplied constructs, if any.

(7) Ensure comparability with other grids by turning the difference scores

into % similarity scores.

Use the following formula:

construct % similarity ј 100 _ рfSD=ЅрLR _ 1Ю _ E_g _ 200Ю

where SD is the sum of differences, LR is the largest rating possible on the

scale, and E is the number of elements in the grid; or refer to the table in

Appendix 4.

Cluster Analysis: Elements

Section 6.2.2

Follow the example shown in Table 6.11 and Figure 6.2.

This is one possible approach for making sense of the output from a clusteranalysis

package.

(1) Examine the elements, and notice which elements have been reordered.

(2) Examine the shape of the element dendrogram.

(3) Identify construct similarities and differences.

(4) What does this mean in terms of the way in which your interviewee is

thinking?

(5) Find the highest % similarity score.

(6) Examine the remaining scores.

Cluster Analysis: Constructs

Section 6.2.2

See also the example shown in Table 6.12 and Figure 6.2.

This is the corresponding procedure for constructs.

(1) Examine the constructs: notice how they have been reordered.

(2) Look at the shape of the construct dendrogram.

(3) Identify element similarities and differences.

(4) What does this mean? Discuss the implications with your interviewee.

(5) Find the highest % similarity score.

(6) Examine the remaining scores.

Principal Components Analysis

Section 6.3.2

Here’s just one procedure for making sense of the output of a principal

components analysis package. There are many others, depending on what

you’re looking for, but it’s one I find convenient.

(1) Determine how many components you’ll need to work with: how many

do you need to account for 80% of the variance?

(2) Examine the shape of the lines representing the constructs: how tightly are

they spread?

(3) Identify any similarities in the meaning of the constructs which make up

each ‘sheaf’:

. by inspection

. by examining their relationship to any supplied construct.

(4) Note the position of any meaningful groupings with respect to the two

principal components: the vertical axis and the horizontal axis.

(5) Check your interpretations with the interviewee. Resist the temptation to

pronounce about them.

Content Analysis: Bootstrapping – Core-Categorisation

Procedure

Section 7.2.1

This is the core procedure which is used as steps 1 and 2 of the contentanalysis

procedures which follow. Each item being categorised is compared

with each of the others.

(1) If an item is in some way like the first item, the two are placed together

under a single category.

(2) If an item is different to the first item, they’re put into separate categories.

(3) Remaining items are compared with each of the categories and allocated

to the appropriate one if an appropriate category exists.

(4) A new category is created if required; when a new category is created,

the possibility that existing categories need to be redefined (combined,

or broken up, with their items reallocated accordingly) is

considered.

(5) This process continues until all the items have been classified.

(6) However, any unclassifiable items are placed in a small category marked

‘miscellaneous’.

(7) No more than 5% of the total is regarded as such.

Content Analysis: Bootstrapping – Generic Content-Analysis

Procedure

Section 7.2.1

Here’s the full process. It should incorporate the reliability procedures shown

below.

(1) Identify the categories.

(2) Allocate the constructs to the categories, following the core-categorisation

procedure, steps 1 to 7 above.

(3) Tabulate the result.

(4) Establish the reliability of the category system (using the procedure shown

below).

(5) Summarise the table; first, the meaning of the category headings.

(6) Summarise the table: next, find examples of each category heading.

(7) Summarise the table; finally, the frequency under the category headings.

(8) Complete any differential analysis which your investigation requires.

(9) Complete any statistical tests on this differential analysis as required.

Content Analysis: Bootstrapping – Reliability

Section 7.2.1

This is for inclusion at stage 4 of the above procedure. These steps improve the

reliability of the content analysis, and measure the degree of reliability

achieved, for reporting purposes. The former is essential and the latter

advisable.

(4) Establish the reliability of the category system.

(4.1) Involve a colleague: ask a colleague to repeat steps 1 to 3

independently.

(4.2) Identify the categories you both agree on, and those you disagree on.

(4.3) Record your joint allocation of constructs.

(4.4) Measure the extent of agreement between you.

. Index A: the number of constructs lying along the diagonal for the

categories you have both agreed on as a percentage of all the

constructs in the whole table.

. Index B: the number of constructs lying along the diagonal for the

categories you have both agreed on as a percentage of the

constructs allocated to the categories you have both agreed on.

(4.5) Negotiate over the meaning of the categories.

(4.6) Finalise a revised category system with acceptably high reliability.

(4.7) Report the final reliability figure:

. agreement on all the category definitions, and 90% successful

allocation (or, if you’re using a reliability coefficient, 0.80 or above).

Content Analysis: Honey’s Procedure

Section 7.3.2

(1) Obtain ratings on a supplied ‘overall’ construct.

(2) Compute sums of differences for each construct against the ‘overall’

construct, using the procedure shown earlier, ‘simple relationships

between constructs’ (see Section 6.1.2).

(3) Ensure comparability with other grids, turning sums of differences into %

similarity index.

(4) Take the individual’s personal metric into account: annotate each

construct with the H-I-L index.

(5) Label each construct with these two indices.

(6) Identify the categories.

(7) Allocate constructs to categories, following the core-categorisation

procedure (see above).

(8) Tabulate the result.

(9) Establish the reliability of the category system (Section 7.2.1, steps

4.1–4.7).

(10) Summarise the table: first, the meaning of the category headings.

(11) Summarise the table: find examples of each category heading.

(11.1) Within each category, order the constructs from top to bottom

with respect to their % similarity scores.

(11.2) Looking at all the constructs within a category, identify personally

salient constructs (referring to the H-I-L indices) on which there is

consensus in the group.

(11.3) If there are subthemes within a category, group them according to

the meaning being expressed.

(12) Summarise the table: state the frequency under the category headings.

(13) Complete any differential analysis which your investigation requires.

(14) Complete any statistical tests on this differential analysis, as before.

Laddering Up to Arrive at Values

Section 8.1.1

(1) Take the first construct in the grid.

(2) Ask the interviewee which pole s/he prefers.

(3) Ask the interviewee to describe the basis for this preference.

(4) Note the answer immediately above the preferred pole of the original

construct, as a new construct.

(5) Identify the contrasting pole of that new construct.

(6) Note it, above the implicit (non-preferred) pole of the original

construct.

(7) Repeat steps 2 to 6 for this new superordinate construct.

(8) Repeat step 7 until the interviewee can’t go any further.

(9) Take the next construct in the original grid; repeat steps 2 to 8 for

it.

(10) Do step 9 for each of the remaining constructs in the original

grid.

Recognising a Personal Value

Section 8.1.2

In deciding when to end the laddering procedure, note that a personal value is

characterised by several of the following attributes:

. abstraction

. universality

. intimacy

. self-reference

. self-evidence

. explicit information (that is, ask the interviewee!).

Prioritising Personal Values: The Resistance-to-Change

Technique

Section 8.2

(1) Take the first two personal values identified by the laddering technique

(Section 8.1.1), calling them A–Contrast A and B–Contrast B.

(2) Present the interviewee with a choice of A at the cost of Contrast B, or B

at the cost of Contrast A.

(3) Record the personal value which is preferred.

(4) Compare value A with the next value, C, following the above two steps.

(5) Repeat step 4, comparing value A with each of the remaining values.

(6) Now compare value B with the next value, C.

(7) Repeat step 6, comparing value B with each of the remaining values.

(8) Repeat step 7, comparing each of the remaining values with each other.

(9) Count the number of times that each personal value was preferred over

another.

(10) Record the outcome as a hierarchy of personal values.

Identifying Personal Change: The Simple Change Grid

Section 9.1.1

Where elements and constructs are identical in two grids from the same person,

(1) Elicit two grids in succession from the same interviewee, with the same

elements and constructs.

(2) Ensure that elements and constructs are written in the same position in

both grids.

(3) Cell by cell, record the difference between the ratings in corresponding

positions in the two grids, using either the absolute value (ignoring minus

signs) or the arithmetic value of the difference.

(4) At the bottom of each column of the change grid, sum the differences in

that column.

(5) Consider a process analysis (Section 5.3.1), do an eyeball analysis

(Section 5.3.2), and characterise constructs (Section 5.3.3) in the change

grid.

(6) Consider any of the analysis procedures described in Chapter 6, though

(a) simple relationships between elements

(b) simple relationships between constructs

(c) cluster analysis

are likely to be more useful than principal components analysis.

(7) Discuss all these changes with the interviewee.

Identifying Personal Change: The Messy Change Grid

Section 9.1.2

Where elements and constructs differ in two grids from the same person,

(1) Elicit two grids in succession from the same interviewee.

(2) Before analysis, make sure that any elements and constructs which both

grids have in common are in the same position in both grids.

(3) If there are sufficient common elements and constructs to make it

worthwhile, treat that portion of both grids as a change grid (see above,

and Section 9.1.1).

(4) Examine the function of the new constructs (and elements, if any) with the

interviewee.

(5) Explore the process by which some of the constructs have been dropped;

focus, first and foremost, on the constructs and their meaning: what’s

being said, and what has changed?

(6) Drawing on the C-P-C cycle, the Experience Cycle, and the Creativity

Cycle, change models, identify the kind of change which is taking place in

your interviewee’s construing of the topic, and act accordingly.

Facilitating Mutual Exploration: Simple Partnering

Section 9.2.1

This is suitable for work in a group.

(1) Negotiate a confidentiality contract.

(2) Agree a set of elements sensible to all.

(3) Agree appropriate anonymity arrangements for elements.

(4) Elicit the repertory grids.

(5) Put the interviewees into pairs, and ask each pair to swap grids,

discussing similarities and differences in their pairs. Eyeball analysis

(Section 5.3.2) and laddering down (Section 4.4.1), or pyramiding (Section

4.4.2) is likely to be useful here.

(6) Process the outcomes in a plenary session.

(7) Agree an appropriate action plan.

Entering Another Person’s World: The Exchange Grid

Section 9.2.2

(1) Negotiate a confidentiality contract.

(2) Agree a set of elements sensible to all.

(3) Agree appropriate anonymity arrangements for elements.

(4) Elicit the repertory grids.

(5) Photocopy each grid, and Tipp-Ex out the ratings on the photocopy.

(6) Put the interviewees into pairs, calling one person in each pair ‘A’, and the

other ‘B’. Get A and B to exchange the photocopies of their own grids.

(7) Either ask each interviewee to fill out the other’s grid as s/he thinks the other

filled it out, or ask each interviewee to fill out the other’s grid as themselves.

(8) Ask each interviewee A to compare B’s attempt at being A with A’s

original grid, discussing the attempt. Then swap round, with each

interviewee B comparing A’s attempt at being B with B’s original grid.

Use the simple change grid subtraction procedure (Section 9.1.1).