Opinion polls, crystal balls and other agents of augury

Opinion polls, crystal balls and other agents of augury

Joel D. Palmer, Jakarta

With its first ever direct presidential election campaign in
full swing, Indonesia is suddenly discovering the marvel of
public opinion polling, now an integral part of democratic
elections nearly everywhere in the world. Hardly a day goes by
without news of yet another poll predicting the outcome of the
five-way race that will culminate on Election Day, July 5.

However, while these polls may provide plenty of entertaining
diversion from the serious business of choosing a leader, there
is a real question as to how accurate they really are.

The recent election in India is a case in point. As the
Congress Party took the reins of power, public opinion pollsters
were still busy wiping the figurative egg from their embarrassed
faces.

According to pre-election polls, Prime Minister Atal Behari
Vajpayee and his ruling BJP were expected to be shoo-ins for
another term in office. But Indian voters proved to be impervious
to the pollsters' conclusions, choosing instead the opposition
Congress Party of Sonia Gandhi by a considerable margin.

Does this colossal failure send a signal that there is an
inherent flaw in public opinion polls, rendering them unreliable
as predictors of winning candidates? Not necessarily, but it does
send the message of "buyer beware" when it comes the
believability of polls and highlights the need to examine the
methodology used by pollsters before subscribing to their
results.

While a telephone poll might be effective in the US, where
telephone penetration is nearly 100 percent, it is of
questionable value in countries with less complete telephone
coverage. This is especially true in countries like India (and
Indonesia) where a sizable share of the large population lives
without access to the amenities of modern technology, like
telephones, the Internet or even television. With much of modern
polling depending on electronic contact, these people are apt to
be left out of the equation.

The problem in India was that the pollsters apparently
neglected to ask the over 300 million Indians living on less than
a dollar a day what they thought of the BJP's India Shining, a
government propaganda campaign aimed at extolling the country's
economic growth figures, while failing to deliver the poor from
their miserable existence.

These forgotten voters, mainly subsistence-level farmers, who
make up nearly one-third of the country's population, apparently
weren't sufficiently represented in the pollsters' samplings. The
result, according to analysts, was an overwhelming, unpredicted
rejection of the BJP by this disenchanted voting block.

It is just such intangible factors that can skew the results
of a poll and result in an embarrassing fiasco for pollsters.
And, Indonesia, with its far flung regions, diverse ethnicity,
and decidedly non- homogeneous, unpredictable mass, the margin
for such error is enormous.

There are no identifiable voting blocks like, for example in
America, with its urban, suburban, working class, Afro-American,
Hispanic, Northern, Southern segments that often vote as blocks
and can be reasonably predicted through careful sampling.

In fact, the results of the recent legislative elections, in
which no single party dominated, prove the degree of diversity
amongst the Indonesian voting populace. That is why pollsters in
Indonesia have to take exceptional care in deciding on their
sampling and, if possible, rely mainly on face-to-face interviews
in lieu of telephone contacts, which can only survey a certain
segment of the population.

Unfortunately, almost none of the local polling firms do this;
they mainly rely on telephone polling, which, while far cheaper
than face-to-face polls, at best gives an indication of how the
"haves" will vote and ignores the 70 percent of the economic
underclass that has no access to telephones.

Most troubling are the SMS polls run by nearly all of the
local TV stations. On any given night the results of these polls
are ballyhooed on one station or another as if they were true
readings of the public pulse.

The obvious flaw is that these polls only survey the small
minority of the population with cell phones and can be easily
manipulated by political parties that urge members to vote a
multitude of times.

While these types of polls are amusing and entertaining, they
are anything but scientific, a fact that the various TV stations
neglect to tell their viewers, many of whom apparently see the
results as indicative of actual political trends.

The greatest danger lies in the possibility of a "wag the dog"
effect of these polls, wherein fence-sitting voters, overwhelmed
by the predicted victory of one candidate or another, jump on the
"winning" bandwagon to avoid casting their vote for a "loser".
This creates a situation where the opinion pollsters, rather than
measuring public sentiment, actually influence it.

Most of the reputable polls to date have shown the front
runners in the Indonesian presidential sweepstakes to be Susilo
Bambang Yudhoyono and Amien Rais. While one has no reason to
doubt these results, it would be well to remember that both of
these candidates appeal primarily to educated, urban voters whose
decisions will probably be based on intellectual considerations.

At the same time, incumbent Megawati Soekarnoputri and the
Golkar candidate Wiranto have generally shown poorly in the polls
to date. But history, culture and common sense tell us that when
Indonesian voters go to the polls the driving force will be more
visceral than cerebral.

President Megawati's appeal to kampung voters who prefer the
status quo to change and the powerful Golkar machine behind
Wiranto will likely improve their numbers considerably when the
real votes are finally counted.

Opinion polling is certainly a legitimate instrument in
political campaigning, but can be a very imperfect science when a
one-size-fits-all approach is taken.

Those opinion polls that ignore the poor, the unemployed and
the isolated as well as disregarding cultural and historical
factors are exercises in deception and are likely to backfire. In
the end the only poll that really counts is the one where voters
go in the booth and punch their ballot.

The writer is Technical Advisor of PT Moores Rowland
Indonesia. The opinions expressed in this article are those of
the author and not necessarily the views of PT Moores Rowland
Indonesia.