Inside The Beltway: Political polling by telephone is more scientific than door to door, but the voters are turned off

'Twenty percent of this business is art, 80% is science,' says James Zogby, the most respected of the hardy quadrennial reviewers - the pollsters - who emerge every four years to tell journalists how Americans would vote that day.

'Twenty percent of this business is art, 80% is science,' says James Zogby, the most respected of the hardy quadrennial reviewers - the pollsters - who emerge every four years to tell journalists how Americans would vote that day.

'Twenty percent of this business is art, 80% is science,' says James Zogby, the most respected of the hardy quadrennial reviewers - the pollsters - who emerge every four years to tell journalists how Americans would vote that day.

The quotation inevitably reminds one of Sam Goldwyn, who once described gin rummy as a game that was '90% luck, and 10% what cards you draw,' and who went on to add 'anyone who would go to a psychiatrist should have his head examined.'

Today, the two - gambling and psychiatry - are not as unrelated as they may once have seemed. First, almost all polling, whether the most sophisticated political sampling or the crassest loaded commercial inquiry, is done by phone, using all the automated systems modern science has made available.

Thus making it at once easier and more difficult, true and false, and - above all - more accurate and more inaccurate - than back in the days when Dr. Gallup would send out (largely untrained) operatives armed with a questionnaire and a clipboard to ring pre-designated doorbells for a friendly interview and a cup of coffee.

Now, the telephone interrupts the dinner hour - by design - and before the caller can tell the answerer that she is a pollster, the conversation ends with a rude phrase, a simple 'not interested' or just 'no, thanks' and a click.

Thus have the intrusive telemarketers influenced politics. The number of calls by a pollster to which a response is given has declined by roughly half in ten years, from 65% percent to 35%. So the sample inevitably becomes diluted, the process lengthened, the temptation to go 'off message' or depart from the script increases, and the necessity to 'weight' the results becomes greater.

The original sample, one assumes, is reasonably accurate. Most of the pollsters are professional, and it's not too hard to classify voters once they've voted. But considering that less than half of eligible voters vote, and only one-third of the polling sample responds, the results must be called 'raw numbers' weighted by age, race, sex (what we now politely call 'gender'), ethnic origin and likelihood to vote, among other measurables.

This requires not only the application of statistical categories, but also some assumptions about geography, occupations, social class and, heaven knows, political issues.

Journalists, who don't like to tell their readers their usefulness is at an end - 'Candidate X is well ahead and will win easily' - seize on polls to tell you 'too close to call.' All concerned should look again to Sam Goldwyn, who also told us, 'An oral agreement isn't worth the paper it's printed on.'





Have you registered with us yet?

Register now to enjoy more articles and free email bulletins

Register
Already registered?
Sign in