It’s not unusual for a politician, faced with an unfavorable poll result, to say something like “the only poll that counts is the one election day”. And we all know that poll questions can be slanted to give the desired result, and correspondingly dismissed by those who don’t like the pollster’s choice of question. But I can’t recall a previous occasion on which a pollster has dismissed the results of his own polling, particularly when we are talking about a margin of 72 to 9. Yet, if we are to believe the editorial in today’s Oz, that’s exactly what Sol Lebovic of Newspoll did when his polling produced a result unfavorable to his client’s campaing for lower taxes. (thanks to reader Jethro for pointingthis piece out).

As discussed in more detail below, when asked whether they preferred a tax cut or more spending on health and education, the answer was 72 to 9 in favour of more spending. The same poll found 50 per cent of respondents believing that the top marginal rate was too high. In an editorial with the astounding (in view of the data) heading, Cutting income tax is a political winner, the Oz calls this a conundrum and cites Lebovic as saying

voters may be giving the “socially acceptable” answer on what they want the Government to do with the surplus.

The Oz goes on to say

In other words, while their consciences may be uneasy about cutting back the welfare state, their gut instincts are telling them we should be cutting back taxes. Mr Howard and Mr Latham would be well advised to respect the voters’ instincts.

It’s not clear whether Lebovic would accept this gloss, which suggests that his poll results should be discarded whenever they disagree with someone’s gut instincts (note that the claims about the instincts of the voters are baseless – the Oz editorial writer is expressing their own gut instincts and those of the political elite).

Lebovic is of course, correct to say that survey respondents often give socially acceptable, rather than accurate, answers. For example, surveys find that far more households take National Geographic than National Inquirer, but circulation figures tell a different story. Unfortunately for the Oz, it’s not a good basis for a political campaign if the viewpoint expressed is so socially unacceptable that only 9 per cent of people will confess to holding it. Moreover, voting is itself an expressive act. Someone who would secretly like a tax cut but doesn’t want to admit can’t actually secure the tax cut for themselves by voting that way – they only get the cut if a majority agrees with it.

A second problem, which undermines the idea of a ‘conundrum’ is that of the shifting majority. 72 per cent of people favored more health and education spending while 50 per cent said the top rate is too high – the number giving both responses could be as small as 22 per cent.

More importantly, there’s no necessary contradiction here. People might support cuts in income tax but think that health and education spending are more important. They could resolve the implied problem for the budget either by favoring a higher deficit, or by making up the difference somewhere else, for example through a higher rate of GST ( a suggestion raised in the previous comments thread by James Farrell) or lower defence spending.

Finally, it’s worth observing that the gut instincts of the voter have been tested in a number of recent state elections. Kerry Chikarovski went to the voters offering a literal fistful of dollars and was buried under a Labor landslide. Jeff Kennett cut services and was defeated by a Labor Party widely viewed as unelectable.

The Oz is grasping at straws when it claims that the overwhelming rejection of its policy line by respondents to its own poll is some sort of pretence. If I were Sol Lebovic, I’d be asking for more respectful treatment of my results.

Reader Jack Strocchi, pointed out this Newspoll report in today’s Australian, concerning voter preferences on taxing and spending. Written by George Megalogenis, who usually gets things straight, it bears the marks of intervention designed to adjust the finding’s to the anti-tax line that has been running hard in the Oz editorial column for some time.

The problems start with the headline Top rate too high, say half of voters. It might be inferred that the other half say that the top rate is not too high. But despite the fact that the excessively high top rate is the central theme of the article, we never find out the distribution of the remaining 50 per cent between “about right”, “too low” and “Don’t Know”. Looking at the partial numbers, I’d estimate that the “Don’t Know’s” at no more than 10 per cent of respondents, implying that about 40 per cent of respondent’s rejected the view that the top rate is too high.

The really interesting news comes in the second paragraph. By the overwhelming margin of 72 per cent to 9 per cent, voters would prefer more spending on health and education to a tax cut. Even among those paying the top rate, most of whom think it is too high, the margin is 69 to 13. If the Australian wasn’t determined to push its opinions into the news pages, this would be the headline.

The implication is that, as regards taxing and spending, the electorate is overwhelmingly more social-democratic than the current government, and arguably more so than the current opposition. It’s no wonder that even mediocre Labor state governments have routinely crushed their opponents since the Howard government was elected.

There are some interesting framing issues here. The “top tax rate” question appears to be framed in a “free good” way – that is, respondents are asked whether the rate is too high, but the fact that a cut would have to be financed somehow is not explicit. By contrast, the tax cut vs spending question makes the trade-off clear.

What really interests me, but isn’t clear in the report is the sequence of the questions (the Newspoll site hasn’t yet been updated). The preference for spending over tax cuts would be even stronger if, as the Oz report tends to imply, the “top rate too high?” question was asked first. Conversely, if the tax cuts vs spending question had been asked first, the framing bias in the top rate question would be reduced.

UpdateIn the comments thread, Don Arthur advises that the paper-based version of the Oz story gives the numbers as 50 per cent too high, 34 per cent about right, 8 per cent too low. Bearing in mind the absence of any mention that reducing rates might mean giving up services, I don’t find this too surprising.

Further update 24/2Andrew Norton at Catallaxy has a post on the same topic, with the same title. and with much the same conclusion. The only difference is that he regrets the outcome and I don’t. A fine illustration of the positive-normative distinction.