>I'm all ears. I'd be interested in hearing about the project, and
>especially how you are defining and measuring "values".
Okay! The short answer is "Laddering"; here's the longer answer!
The project involves 48 employees of a social security agency, each one of
whom has provided a repertory grid on the topic of internal and external
fraud and security breaches. The elements (8) were brief case descriptions
of typical security breaches/frauds, some 200 words each, devised by the
Internal Security branch of the agency as typical of the cases they
encounter, made available to respondents in advance and referred to during
the grid interview as "case no 1, case no. 2", etc. Each respondent
provided 6-8 constructs and rated elements on constructs on a 5-point
scale. I supplied one construct to each respondent, viz. "overall more
serious breach - overall, less serious breach" and constructs were
aggregated across the sample using Honey's content analysis technique,
(which has the useful property of retaining the individual provenance of
each construct while classifying under a set of categories relevant to the
whole group: see EPCA Newsletter No. 1 for a procedural guide, or Honey
1977). The content analysis was highly reliable: (Perrault-Leigh 1993)
reliability index of 0.96, 95% confidence interval +/- 0.025.
As for values: not sure I want to say that I'm "measuring" them, since this
connotes a precision which is probably unwarranted. However, they were
defined according to Donald Super's notion of a value as _a personal
objective which is held as self-evident for its own sake, without further
justification_. They were obtained by a simple resistance-to-change
technique by
a) starting with each construct
b) asking the interviewee which end of the construct s/he preferred
c) exploring the reasons for the preference, in terms of "why, _for you_",
is this the preferred pole
d) iterating steps b) and c) until the self-evidence for preferring the
pole of the current value was agreed by myself and the interviewee.
(Yes, following pcp approaches the value is regarded as itself a construct,
albeit at a high level of abstraction and generality, and thereby involving
expression as a _bipolar_ entity just like any other construct.)
Duplicate values were eliminated at the prompting of the interviewee; the
remainder were then put into a priority order by a series of successive
pairwise comparisons in which the interviewee had to choose between
positive pole of value (i) - negative pole of value (j) and
negative pole of value (i) - positive pole of value (j).
For example: if two of your values are
Predictability- Chaos (value (i)) and
Social support - Living and Being alone (value (j))
the left-hand pole of each being preferred,
the particular pairwise comparison for these two values would be presented
as the choice: "Suppose you were offered the chance of living in a new
world. Two worlds, actually! World number one is Predictable; but living in
it, you are Alone; World number two provides you with all the Social
support you want; but it's a Chaotic world. I know it's difficult: but
which world would you prefer to live in?"
Comparing all the values pairwise and summing the preference choices
results in a hierarchy of values for the individual, in which the
most-frequently- chosen are those which are, by definition, more resistant
to change, and the less-frequently-chosen are those which (albeit values,
and thereby still important to the individual!) comparatively more
resistant to change.
I understand from Helen Jones, to whom I owe my introduction to this
technique, that it's a direct variant of Hinkle's resistance-to-change grid
(and I'd give my right arm for a complete copy of this reference: Hinkle's
PhD thesis is one of those seminal references that one just can't get hold
of!)
As you'd expect, different people:
a) expressed different values (with some overlaps, of course);
b) different numbers of values;
c) and in different priority orders.
The way I aggregated these across all 48 respondents involved another
content analysis
d) _in which the individual preference orderings of the value hierarchies
was preserved_.
Eek! If you want to know how I did the aggregation despite a), b) and c)
while achieving d) immediately above, please contact me for more details.
Best way of doing so is to e-mail me directly for a copy of Jankowicz 1996,
where the procedure is outlined. (Eek again: don't everyone ask!! But if
you really need it, and ask me for it enclosing a snail-mail address, I
promise to mail it on to you. Or wait until the autumn, since the
organisers of the conference in which I gave this paper will be publishing
_Proceedings_ later this year.)
Kind regards,
Devi Jankowicz
References
Honey P. (1977) "The repertory grid in action" _Industrial and Commercial
Training_ 11, 11, 452-459.
Jankowicz A.D. (1966) "Personal values among public sector employees: a
methodological study". A paper given at the 3rd biennial Conference of the
European personal Construct Association, University of Reading, U.K.: April.
Perrault W.D.J., & Leigh L.E. (1989) "Reliability of nominal data based on
qualitative judgements" _Journal of Marketing Research_ XXVI, May,
135-148.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%