by Wendy Ogbourne
I’m not sure whether the gathering of data for Opinion Polls and Market Research can reasonably be called a Science, an Art, or just plain guesswork. But there’s no doubt it has become an integral part of modern life.
There are actually distinct differences between them. Opinion Polls, mainly used to ascertain political views and voting intentions, aim to provide a snapshot of public opinion at one point of time.
Market Research surveys, on the other hand, are used to discover and predict people’s past, present and future behaviour, largely in the realms of leisure activities, consumption, travel and brand loyalty.
There are many reputable polling organizations, which are governed by their own country’s Code of Conduct – Roy Morgan and A.C Nielsen in Australia, Gallup in US. Most major newspapers also run their own polls, such as Newspoll by the “Australian” newspaper, Galaxy by the “Daily Telegraph”, or the “Guardian” in UK. Smaller, special interest groups conduct private polls from time to time.
Interviewing is done in a variety of ways – traditionally face-to-face, but increasingly by phone, email or online. The time taken to complete all the questions can vary from a few minutes, to almost an hour. Interviewers are coached in the need to present the questions in a totally unbiased and consistent form, to eliminate bias. Many years ago, I was employed as an interviewer for the Australian Bureau of Statistics, visiting a set number of homes each month, to gain information about employment. The rules were strict.
How do these polling organizations select the people they interview? There are several methods used, depending on the issue to be polled, and the size of the sample required. It’s been estimated that with a poll sample of 1000 people, there is likely to be a 3% margin of error, which is considered to be an acceptable margin for most purposes. To reduce that to a 1% margin, you would need to sample 10,000 people, which is obviously more time-consuming and more costly for the company conducting the survey.
“Random” sampling is done by random selection of phone numbers, email addresses or house addresses from the electoral roll.
“Quota” sampling sets quotas, of age or gender for example, from which respondents must be chosen.
A third method could be called “passers-by” sampling, where an interviewer stands in a shopping mall, and accosts shoppers for their opinion on a specific issue.
It’s clear that whatever method is used, it is always open to error. To a large extent, people self-select themselves as respondents, as many do not wish to take part in the survey, or give untrue answers, for whatever reason. It’s a reasonable criticism that only a certain type of person will give a complete stranger (even if he or she does wear an official badge), the details of his marital or financial situation or his political affiliation. This seems to be even more true for internet polling. Only people who enjoy doing polls online will be doing so. It’s also true that the way a question is worded, and even the order the questions are presented can affect people’s answers. We all know what is meant by a ”loaded” question. The truly reputable polling organizations go to great lengths to avoid this.
So how come opinion polls so often get it wrong? The truth is, even if you accept polling as a science, it’s not an exact science and never can be, as long as human beings are involved. When politicians say, “The only poll that counts is the one on polling day”, they’re right. As polling heats up in the weeks and days leading up to an election, polls follow each other thick and fast. People can and do change their minds rapidly in those last few days, as final events unfold. And being human, as we all are, we can be influenced by the poll result we read the day before. Two forces can come into play here – the “band-wagon effect”, where one party seems to be so far ahead, that we decide they must be the best one if everyone else thinks so; and the “underdog effect”, where people vote for the candidate or party who seems to be losing. Do these two effects cancel each other out? Who knows?
There’s plenty more to be read about all this, simply by Googling “Opinion Polls” or “Market Research”, and Wikipedia is always a good place to start.
© Wendy Ogbourne, 2010
This entry was posted by wendy on Monday, February 1st, 2010 at 12:00 am and is filed under Science Stuff.
You can be notified of any comments on this entry through the RSS 2.0 feed.
You can leave a comment in the box at the bottom of the post.
Print This Post
After you click the print button, a new page will appear.
Click the link at the bottom right of the page to print it.