How to Decide That Voters Decide
| Author | Patrick Dunleavy |
| Published date | 01 October 1982 |
| Date | 01 October 1982 |
| DOI | http://doi.org/10.1111/j.1467-9256.1982.tb00070.x |
| Subject Matter | Article |
24
HOW TO DECIDE THAT VOTERS DECIDE
PATRl
CK
DUNLEAVY
Patrick
Dunleavy.
In a recent book, four social psychologists, Hilda Himmelweit, Patrick Humphreys,
Marianne Jaeger and Michael Katz, set out an uncompromising and distinctive view of
How
Voters Decide. They project 'the individual voter' back into the centre stage,
describing her straightforwardly as a 'political consumer', and unabashedly pro-
claiming the reality of consumer sovereignty in this as in other fields. 'The voter
of today is
.
. .
someone who is not simply conforming to his or her own past or
follows other peoples' example, but makes up his own mind' (p.202). 'The psychology
of the voter today
.
. .
the psychological processes involved in arriving at a
decision' can be best described in terms of market analogies. 'Instead
of
buying
goods the individual purchases a party', searching 'for the best
fit
or
least misfit
between his or her views and preferences and the parties' platforms'. Tlhe 'individual
voter' emerges as rationalistic, well informed, voting on issues, increasingly
discriminating inher choices over time, and willing
to
adapt her views fairly exten-
sively to take account of a changed situation. The only social structural influences
on voting behaviour are two residual and weaker pressures: 'the first of these is
the habit
of
mting for a party, like the loyalty people develop for a particular
brand or shop, and the second is the pressure or example of significant reference
groups, like the way one's friends and style of living affect one's own pr.eferences'
(p.
198).
on the basis of the evidence and explanation developed in How Voters Decide because
of problems in the sampling used in the study, in the data collected, and in the
analysis.
I
argue that the book is simply the latest in a now lengthy sequence of
studies which have been misled into ideological reasoning by a critical failure to
examine the assumptional underpinnings of economic theories of voting, from which
the use of market analogies derives.
In this critique
I
shall seek to demonstrate that these views cannot be sustained
Samp
Zing
Prob
Zems
to the behaviour of the electorate as a whole there must be some assurance that the
data set is .representative in social background terms. But with the Himmelweit
et.al. study there are grounds for doubting whether this basic requirement is met.
Their research grew out of a non-political study of adolescent boys in grammar
and secondary modern schools in the late 1950s. Starting with
600
plus initial
respondents, heavily weighted towards grammar schools for extraneous reasons
to
do
with their initial project, Himmelweit et.al. resurveyed progressively smaller
proportions of the people involved in 1Gnd at the six subsequent elections up to
1975.
Their core data set is an eventual 'longitudinal sample' of just
178
people
successfully recontacted at every stage, and this group is used in all the key
analyses presented. But the initial bias
of
the data set, plus wmulatively lower
response rates from manual people, plus very high levels of intra-generational
mobility among the data set respondents combined to produce a final 'sample' with
just
31
manual workers apd nearly 80 per cent of respondents in the 'upper middle'
or 'middle middle class' (Table
1).
Their occupational class categories are nowhere
defined by the authors,
so
comparison with other studies is difficult].
safe
to
suppose that there are around ten times as many higher non-manual people
in the Himmelweit et.al. data set as there should be given the numbers of manual
workers. Thus
it
seems legitimate to ask in what sense the study is based on a
sample, rather than just an apparently arbitrary selection of data.
The authors never convincingly face this criticism, presenting data on social
backgrounds only in an appendix, with no details of the categories used, and with
a number of quite misleading pleas in mitigation. Chief of these is the suggestion
that since their respondents' attitudes and alignments were quite close to those of
If
we are to have any confidence in extrapolating from findings in one survey
But
it
seems
To continue reading
Request your trial