It is well known among businessmen, that the reaction of the public toward a given product or service, can be fairly accurately estimated by making samples inquiries among the relevant customers. This device is being applied to the science of government. In the United States, promising strides have been made by public opinion polls like the American Institute of Public Opinion, Crossley and ’Fortune Magazine.’
Exhibit 7.1
Polling Techniques
7.1a Polling Techniques
The polling technique was extended to elections
by collecting what were known as straw votes from small portions
of the electorate. For example in 1936, the Literary Digest attempted
to collect straw votes by sending ballots through the mail
to people whose names were taken randomly either from telephone
directories or from lists of automobile owners. This resulted in
gross miscalculation because the sample did not include people who
did not own telephones or cars. Some of them did not even return
the ballot.
There are several polls that try to make election
predictions, of which the best known is the one conducted by George
Gallup, for the American Institute of Public Opinion. The accuracy
of the results of the poll depends on securing a sample representative
of the total universe, that is the whole group whose
opinion is being sought. If each unit in a universe is equally included,
then even a small sample can give accurate results. Beyond a certain
point, if there is an increase in the size of the sample, the chances
of a reduced number of sampling errors are greater. (A sampling
error is the range existing between divisions in the sample and
those of the universe.)
Drawing the sample at random would be one way of
developing a representative sample. However this kind of random
sampling is impossible for most political surveys. Instead census
tracts are used, giving the number of residences and their locations.
Quota sampling is a less complicated sample, though it is
less reliable. In this type of poll, an attempt is made to secure
a sample reflecting the variables among the population that can
affect opinion.
7.1b Avoiding biased samples
In many cases conclusions may happen to be based
on biased samples. For example a reporter may write that students
are becoming more conservative, after interviews with a dozen students
on the Yale campus. In general the success of opinion polls depends
on the accuracy of the sample rather than on the number of persons
interviewed. It is important for the sample to represent a cross
section. The error in the polls should not exceed three percent.
The neutral wording of the questions is also important, as pointed
questions can elicit pointed replies. Voters may also be interviewed
after they have cast their votes. This is known as exit polls,
which are also an aid to the candidate to estimate the respective
areas of his weakness and strength and to adjust his campaign techniques
accordingly.
It cannot be denied that in spite of the difficulties
involved in polling, it is such a useful device that it is used
by various organizations. Parties conduct polls during elections
to find out their strong and weak points. In fact, polls are an
essential art of the American politics. The news media carry out
surveys and even the polling organizations are involved in this
task. Polls are also run by interest groups to support their claims
that the people are in favor of or in opposition to a certain policy.