1996 Australian election
Did the 1996 Federal Election See a Blue-Collar Revolt Against Labor? A Queensland Case Study
Review of J.
Singleton, P. Martyn & I. Ward. 1998. ‘Did the 1996 Federal Election See a
Blue-Collar Revolt Against Labor? A Queensland Case Study’ in Australian Journal of Political Science V33,
no.1. Australasian Political Studies Association.
(based on an honours thesis written by J. Singleton
in 1996).
This
article tests the hypothesis that the ALP’s traditional support base, defined
as blue-collar workers, deserted it in 1996. This is argued as the most
significant reason why Labor lost the 1996 federal election, and by such a big
margin. It uses triangulation by using three different data sets in combination
with ABS demographic data:
1.
Australian Electoral Survey
(AES), 1993 & 1996
2.
AEC voting data, 1993 & 1996
3.
Survey of ex-ALP party members,
1996
The
authors choose Queensland in particular: it had the biggest swing of all the
states, 8.7%. They argue that if there is any truth to the hypothesis, it
should be seen in Queensland because of the accentuated swing. Their
alternative hypothesis, H1, is that
blue-collar voters across Australia did
swing disproportionately away from the ALP. You can reject H1
on the basis of a lack of evidence from Queensland, but you can never accept it
on the basis of evidence from Queensland: it is an extreme example (big swing).
This is a methodological problem with the paper.
The
first data set contains contrary statistics from AES survey data which show
that the blue-collar swing in Queensland was only 9.6%, compared with 10.1%
nationally. They hypothesise that the biggest reason for the big overall swing
was the blue-collar swing. Yet, in Queensland, the state with the biggest
overall swing, the blue-collar swing was less than the national average. Even
small-ish discrepancies such as this should be taken more seriously than the
authors did.
The
authors also use ABS demographic data from the 1991 census, by suburb. The
second data set comes from AEC voting statistics from both the 1993 and 1996
elections, arranged by booth.
The
third data set is from a survey of ex-ALP members, asking about the reasons why
they left the ALP. The response rate was fair (61%) and reflected the gender
composition of ex-party members and the correct proportion of the samples from
Brisbane proper and south-east Queensland generally. However, disappointingly,
we do not find out if these two groups of ex-members (women and Brisbane
residents) are over- or under-represented as a proportion of those leaving the
ALP. The authors are trying use this survey as a basis to make predictions
about why voters in the general
community swung against Labor.
Question
5 is ‘Which of the following reasons listed below best describes why you left
the Australian Labor Party?’ (see p127). The first possible response, ‘Internal
organisational problems’, is probably not important to the average voter. The
second response, ‘Leadership style of party’s political leaders’, would be more
relevant to average voters who are not as well informed about policy as party
members are. The fourth response, ‘ALP polices didn’t help workers’ is also
potentially problematic. It seems easy to imagine a high proportion of
blue-collar workers responding favourably to that question, how could they not?
Perhaps the survey designers could have tried to obtain this information
differently, possibly by asking about attitudes to Industrial Relations or
Welfare Policy, or some other way. These
points surely call into question the claim that the data from the survey (of
ex-party members) shows a disproportionate number of blue-collar voters in the general community probably voted
against the government because its policies ‘didn’t help workers’.
There
are possible epistemological concerns with the survey here. It assumes that
voters are sophisticated and rational. It also assumes that they are
well-informed about politics and that they know what they are doing. When blue collar voters swing to the Liberal
or National parties you really have to question if they are aware of what they
are doing. It is almost certainly in the interests of at least most blue-collar
workers to vote for, or preference, Labor. Are there issues of hegemony or
false consciousness here, or political campaigning tricks etc.? If you want to
know why blue-collar voters deserted Labor, frankly, I’m not sure that
blue-collar people are going to tell you the answer, because they probably do
not actually know. A psychologist or sociologist is probably needed here!
For
the overall research project the authors narrow the parameters of their
research. They claim to be trying to deal with facts to establish links between
voting statistics and demographics. The authors are looking to be objective,
matching statistics with other statistics, in a presumably ‘value-free’ way. It
is easy to understand why they might ignore epistemological considerations;
they are following the lead of previous studies of electoral behaviour, which
are quite common. Such studies have a sophisticated, specific format which has
been perfected over the years and is well established.
There
are epistemological concerns, however, when the authors form their hypotheses.
They make certain choices about which variable cleavages should be studied.
This seems to be based on ‘gut feelings’ and comments from ‘the media’ about why people swung against Labor.
No literature relevant to the 1996 election is
reviewed. The 1998 paper is only a short research note and is based on a thesis
written in 1996. However, a review of other literature probably would have
offered other suggestions as to possible reasons for the huge swing. There may be a particular bias towards forming the view that
blue-collar voters were the main reason for the big swing. It seems to be
presumed, certainly by the media commentators, given the emphasis of Howard
during the campaign on fighting for the ‘battlers’.
Is
it possible to test the hypothesis without looking for other, potentially more
significant, correlations? The project focuses on class only. This would
obviously lead to the conclusion that being blue-collar is a major determinate
of members leaving the party. Maybe, however, it is possible that other (or
even all) cleavage groups which traditionally support the ALP also abandoned
them; maybe blue-collar workers were only one of these groups. If this is the
case, you could be testing whether blue-collar voters abandoned Labor but you
would not necessarily prove that this was the
reason why Labor’s vote collapsed – maybe in 1996 everybody abandoned the Labor
party!
Is
it possible that women swung against Labor in disproportionate numbers to men?
Certainly this could be a result of the perceived arrogant leadership style of
Keating, which could possibly affect the
female vote. Age is also an important cleavage that is usually analysed: there
is no attempt to explain the under-representation of retirees in the group of
ex-Labor party members.
Figure
2 (p120, see attached) plots blue-collar composition of suburbs against the
swing away from Labor. On the right hand side of the graph, where blue-collar
proportion is over 35%, it looks roughly like there could be two distinct trend
lines. This could possibly explain the low variance (R²) of 0.25. This suggests that maybe there are other important variables
that are related to the size of the swing. This could be related to age. It
could be that older voters are less likely to be swinging voters because of
higher levels of party identification. Alternatively, older people could be
more racist, accounting for a bigger reaction against Keating’s ‘big picture’
policies on Mabo, Reconciliation, and ‘integration with Asia’.
These
possible divergent trend lines could represent any of a number of other voting
patterns if not age-related patterns. This should have been investigated or at
least discussed especially with such a low R² value. This would have been easy
to do because ABS demographic data usually includes statistics on age, gender,
ethnicity / aboriginality and immigrants. South-east Queensland would be a poor
choice of case study if it has less immigrants than other areas like Sydney or
Melbourne.
The
authors may have gained an insight into other possible causes of swings if they
had conducted some preliminary qualitative research. For example, a focus group
or two might have revealed age, gender or other factors were relevant. Some
other preliminary surveys might also have helped here.
The
authors also make the claim that blue-collar ex-ALP members (cf. all ex-ALP
members) were the most likely to take any dissatisfaction to the ballot box.
This is an important finding in terms of the overall hypothesis. However, not
voting for the ALP as a first preference is not necessarily linked to Labor
losing seats if preferences go to Labor. It is stated that among blue-collar
ex-ALP members, 47% voted against Labor – 12%
for the coalition plus 35% for other parties. This comes from a survey of 211
ex-members where 59 are blue-collar – this means that 12% represents only 7 people. 59 is quite a small sample
size; where a statistic represents only 7 people we must question if this is
valid. Those 7 people could easily be, say 2 or 3, if you took a survey of
ex-members from states with smaller swings than Queensland or surveyed members
from non-metropolitan areas.
The
authors also comment on Paul Kelly’s argument that the result of the 1996
election shows the decline of class based politics (see p123). The authors,
unfortunately, think they have evidence to support this claim by analysing
evidence of a single swing from the
1993 to the 1996 election. Unfortunately, you cannot predict trends from one result! One swing of
voters in 1996 represents a ‘snap-shot’ in time, not a trend. You need to
analyse several elections in a row before you can make claims about trends like
that.
Again
there are epistemological issues here. Kelly purports to explain why blue-collar voters swung against
Labor: voters in general are showing lower and lower levels of strong party
identification (see p117). This is over-simplifying a complex phenomenon and it
erroneously assumes that voters are rational, well-informed and not influenced
by hegemony or other forces.
Overall
the article presents some good data but it does not adequately test the claims
of the media commentators as it purports to. It does use triangulation to try
to add weight to the arguments; this is probably its main saving grace.
BIBLIOGRAPHY
x
S. Castles, M. Kalantzis, B. Cope & M. Morrissey. 1988. Mistaken Identity: Multiculturalism and the Demise of Nationalism in Australia, 2nd Ed. Pluto Press: Sydney.
D.
de Vaus. 2002. Surveys in Social Science
Research, 5th Ed. Allen and Unwin: Sydney.
C.
Hughes. 1980. The Government of
Queensland. University of Queensland Press: Brisbane.
P.
John. 2002. ‘Quantitative Methods’ in D. Marsh & G. Stoker (eds.) Theory and Methods in Political Science,
2nd Ed. Palgrave: New York.
D.
McNabb. 2004. Research Methods for
Political Science: Quantitative and Qualitative Methods. M.E. Sharpe: New
York.
W.
Neuman. 2003. Social Research Methods:
Qualitative and Quantitative Approaches, 5th Ed. Allyn and
Bacon: Boston.
M. O’Neill. Hard Labor (video recording). 4 Corners, ABC: Sydney. Broadcast 24 February 1997.