Measuring Movements of Electors Using Election Results

AuthorAlan Taylor
Published date01 June 1974
Date01 June 1974
DOI10.1111/j.1467-9248.1974.tb00019.x
Subject MatterArticle
MEASURING MOVEMENTS
OF
ELECTORS
USING
ELECTION
RESULTS
ALAN
TAYLOR
University College
of
Swansea
Tmmovements ofelectors from one party to another, and to and from abstention,
can be studied from two data sources. A certain amount of information may be
obtained from opinion surveys, but this is not satisfactory on its own because
electors are often unable or unwilling to recall their past voting accurately.
Therefore such evidence requires checking from the only reliable source
of
evidence on electoral behaviour-the election results themselves. This becomes
more and more imperative as the gap between interview and voting act grows,
The problem is to find a method by which such information can be obtained from
election results. Care is necessary when comparing the results from aggregate
data such as elections with those from opinion surveys. The former can only
show net movements of election, while the latter show gross movements. In
addition it is hazardous to make inferences about movements of electors from
election results since there is the danger of drawing unfounded conclusions due
to the well known ecological fallacy. The patterns which appear in election
results may not appear at the individual level.
Methods
of
obtaining information on this topic from election results have been
suggested by Miller’l Irwin and Meeter,2 and Ha~kes.~ Though they are
of
considerable potential value, they are open to methodological criticism, do not
avoid the ecological fallacy and have yet to be perfected. However,
a
slight
lowering
of
sights would permit the use
of
less ambitious and simpler techniques
which are open to fewer objections but which still produce useful results. In
particular, the ecological fallacy can be avoided by limiting the scope
of
the
conclusions.
I
studied relationships between changes in each party support,
basing the analysis
on
first-order single linear regression coefficients of the changes
in one party’s strength on the changes in each other party’s strength. These
coefficients will show the extent to which changes
in
support for each party can
be predicted from changes in support for each other party.
For
instance, they
will show to what extent the Conservatives do badly where the Liberals
do
well.
This method does not measure transition rates.
A
negative coefficient indicates
a consistent pattern of apparent net movements between the two parties con-
cerned. Since these are not voter transition rates, it does not matter that they may
be negative. This avoids a major problem encountered by
It
does mean
that the results are not directly comparable with transition rates calculated from
opinion surveys and also that conclusions about the size of movements of electors
W.
L.
Miller,
‘Measures
of
electoral
Change using Aggregate Data’,
Jorrrnal
of
rhe Royal
Statistical Society,
A,
135,1972,
122-142.
G.
A.
Irwin
and
D.
A.
Meeter,
‘Building
Voter
Transition Models
from
Aggregate
Data’,
Midwest Journal
of
Political Science,
13,
1969, 545-566.
A.
G.
Hawkes,
‘An
Approach
to
the Analysis
of
Electoral
Swing’,
Journal
of
the Royal
Statistical Society,
A,
132,
1969, 68-79.
‘Miller, loc. cit.
Politid
Studies,
Vol.
XXU,
No.
2
(204-209)

To continue reading

Request your trial

VLEX uses login cookies to provide you with a better browsing experience. If you click on 'Accept' or continue browsing this site we consider that you accept our cookie policy. ACCEPT