One of Canberra’s leading opinion pollsters and social researchers was moaning to me this week about increasing telephone rudeness.
The principal of Datacol, Malcolm Mearns, said his callers can hardly get a word in edgeways.
His moan comes after a moan from the Labor Member for Fraser, Bob McMullan, about the way the media reports opinion polls – the way they jump to unsupported conclusions.
Mearns says genuine researchers and pollsters are victims of a vicious circle. As telemarketers get more insistent, ring more frequently and ring at more inconvenient times, householders have become more impatient and usually polite people will hang up immediately.
Researchers and pollsters would be able to explain and keep a respondent if they could just get 30 seconds, but even that is getting more difficult.
“It is hopeless if the householder answers the phone with sword drawn and pike ready at the sound of a stranger’s voice,” Mearns said.
When he started 20 years ago, telephone researching was a respectable part-time job, he said.
“Now it is low status,” he said. “It is very difficult to get good people.”
I think it is possible that this attitude is distorting the polls because they are excluding the impatient, the rude and the people with little time. Are such people more likely to be Coalition voters? Might this partly explain why Labor has been winning the vast majority of opinion polls since 1999 yet losing at the critical election time.
This brings us to McMullan’s complaint.
It is twofold. First, he says there is a perception that the Government will win the election, that Labor is not in the race. This colours the reportage of polling. He cited reporting of one of the few polls in which the Government was ahead (by a couple of percentage points) that, “The Howard Government has opened an election-winning lead . . . .” Yet, when Labor was in a similar position in the next poll, the reporting did not suggest a Labor win.
His second complaint is that reporters make too much of small changes, most of which are within the statistical margin of error of the poll. Usually, you can be 95 per cent confident that the poll is right within about 3 per cent. It means that if you did the same poll 100 times, in 95 cases the poll would be within 3 per cent of the true view of the whole population.
So, if a poll moves a few per cent from one fortnight to the next, it does not mean a thing. Yet reporters breathlessly conclude that the electorate has changed its view because of whatever the latest political shenanigan is: Telstra, visas for values, industrial relations and so on.
In fact, the vast majority of voters are switched off most of the time. Recent research by Professor Judith Brett and Anthony Moran, of La Trobe University, tracked the political apathy in depth doing the opposite of what pollsters do – not a snap shot couple of questions to a lot of people – say 600 – but a detailed tracking over time of the attitudes of 22 people. The constant apathy was abundant.
People are skeptical of polling mainly because they find the mathematics too difficult. How can 600 people accurately reflect the views of 15 million voters? Well they can within degrees of confidence.
Let’s take an evenly split population of four voters – two Coalition (C) and two Labor (L). You take a sample of two. Your sample can only be CC, LL, LC, CL. If your sample selected either of the first two permutations you would be wrong, out by 50 per cent in each case. If your sample was either of the last two you would be spot on. None the less, you are out half the time.
In this case you can be 100 per cent sure your sample reflects the total population plus or minus 50 per cent. Or you can be 50 per cent confident that your sample in fact reflects the population. That’s how margins of error and degrees of confidence work.
This column is too short (and its readers not fascinated enough with figures) to extrapolate this to a whole nation. But as the population gets bigger the percentage sample size needed to get a reasonably accurate view becomes smaller.
But only “reasonably” accurate – within a few percent. And therein lies the big difficulty for Australian pollsters: Australian elections are always so close. The difference between the two parties is rarely more than a few percent – the margin of error of the polls. Worse, a few percentage points difference can result in a very large difference in the number of seats won in our single-member system.
Overall, the polls get it right. They always say the parties are fairly even. Votes won in an election are always fairly even. But the party with a couple of percent more votes can win a landslide of seats.
So should we pack in polling. No. Perhaps more important than the question, “Who will you vote for?” is the question, “Why will you vote for them?” Politicians take notice of these polls. In fact, it is through polling that ordinary people can get a say and put a voice different from the vice of lobbyists and big interests who have easy access to ministers, the Opposition and the bureaucracy and have the money to produce artful documents to support their self-interested cases.
Do not hang up on pollsters and social researchers. Feel privileged that it is your chance to influence things.
And help is at hand. Legislation has been passed to set up a Do Not Call register. It will start early next year. You can put your phone number on it. It will be an offence for telemarketers (people selling stuff) to ring that number with fines of up to $1.1 million for corporations, including Australian corporations who engage overseas call centres to do their work for them.
Pollsters and researchers can’t wait. It may take some time, but with telemarketers turned off, people can return to ordinary civility on the phone and at least give a caller 30 seconds to explain that they are doing something useful – giving ordinary people a say and getting a better understanding of what the masses are thinking.
I’ll end here, because I hear the phone ringing.