What’s this? Internecine warfare at News Limited over public opinion polls? It sure looks like it.

For decades, News Limited has relied chiefly on the polling company Rupert partly owns – Newspoll. The Australian uses it more or less exclusively, although the tabloids in the group have used other pollsters from time to time.

Newspoll has an outstanding reputation, meticulously built up and maintained by Sol Lebovic, a real straight-shooter and a thoughtful, careful pollster.

Over the years, his questions have been characterized by fairness, relevance, and comprehensibility. His sampling procedures and reporting of data have been transparent. His record for accuracy at election time has been first-rate.

However, the tabloids in the News Limited stable tend to prefer something a little racier, and have often commissioned one-off polls elsewhere. Their pollster of choice for the moment is Galaxy.

We had occasion to criticize one of Galaxy’s questions recently, but we have also defended the company against what we regarded as unfair criticism.

Now, however, Newspoll has taken a shot at Galaxy.

On Page 4 of The Australian of 10 July 2007, Martin O’Shannessy, chief executive of Newspoll, wrote a commentary to explain why a Newspoll question about the Howard Government’s intervention in the Northern Territory produced a result that seemed to contradict the results of a poll on the same topic by Galaxy published in News Limited tabloids.

The explanation was obvious: the polls had asked different questions.

However, O’Shannessy then criticized Galaxy for asking a question that he said was “set up in a way that revealed a pre-existing cynicism” about the Government’s action. “The result was that the predictable cynicism of the electorate was stimulated by the question.”

O’Shannessy went on to deliver a little homily: “As pollsters, we face a heavy responsibility to ask questions in a neutral way to avoid interpretations that obscure the truth, rather than uncover it”.

Here are the questions and headline results:

Galaxy’s: Do you think John Howard is addressing problems in Aboriginal communities because of the upcoming election or that he really cares?

Fifty-eight per cent said it was because of the election.

Newspoll’s: Thinking about Prime Minister John Howard’s recent intervention in the indigenous communities of the Northern Territory: do you approve or disapprove of the Prime Minister’s actions in relation to these communities?

Sixty-one per cent said they approved of the intervention.

Galaxy was measuring what people thought was Howard’s motive; Newspoll was measuring whether people supported the intervention or not.

There is no doubt that Newspoll’s approach is fair and relevant. You could argue that by using John Howard’s name and title, the question might have stimulated latent additional responses deriving from people’s feelings about him personally or the office he holds; arguably, the term “Government” might have been more neutral. But this is a fine point, and anyway Howard did very much personalize the Government’s response.

Galaxy’s question has its own legitimacy based upon the essence of the main opposing points in the public debate: either this was just an election stunt, or Howard did care about the children.

Two further points many be made. In both polls some 16 or 17 per cent of respondents answered “don’t know”. These are quite high proportions and suggest some degree of uncertainty or lack of knowledge in the community. Even more may have responded in this way if they had been offered the “don’t know” option explicitly.

We know that respondents are at times tempted to offer an answer to subjects that are new to them and about which they have not thought much, in order to get off the hook. In this situation the propensity for some to allow their support for Howard or their cynicism about politicians to influence their response must be allowed for.

We have found that offering an explicit “don’t know” is a useful expedient when an issue has just come to public prominence and many people may not have had time to think about it. In these circumstances, some respondents may not want to volunteer that they have no opinion, but may do so if offered the express opportunity.

O’Shannessy is making a separate but important point. Pollsters know the electorate is cynical about politicians. Indeed they are cynical about many other institutions: the media, the banks, the church, trade unions.

Knowing that, pollsters have a responsibility not to make the question too easy to answer in one particular way.

On balance, however, we think O’Shannessy is being a bit tough on Galaxy here. The Prime Minister’s motives had been widely canvassed in more or less exactly the terms used in the question. Thus there was little artifice in the question design. If such a question draws to the surface the latent cynicism of the electorate, so be it.

The weakness in the question was that it ignored a third possibility: that Howard was motivated both by genuine concern for the children and by electoral opportunism. What a cynic might call Tampa Syndrome.

To get round that, the question should have asked whether people felt Howard acted mainly because there was an election coming up or mainly because he really cared. It should also have allowed explicitly, as part of the question, the “don’t know” alternative.