Tuesday, 10 November 2009

Gary Morgan’s common sense

It is a measure of the importance of The Australian in the political life of this country that Newspoll has somehow assumed the position of being the opinion poll of importance for journalists. AC Nielsen can put out a picture of electoral support once a month in The Age and the Sydney Morning Herald; Galaxy can make its occasional sorties in to the News Limited tabloids; Morgan can consistently publish the result of its face-to-face and telephone surveys on its own website; the fortnightly surveys by Newspoll published in The Australiancontinue to be the ones that the media, particularly the electronic elements of it, take as being the real measure of what Australians think.
I am unaware of any evidence to support this pre-eminence given to the Newspoll view. My glance at recent election results compared with the final predictions of the pollsters suggests that there is no such evidence. As I said at the start of this little note, it is a measure of the importance of the national daily rather than the accuracy of the pollster that produces the influence.
It thus behoves commentators such as me to be wary of giving too much influence to Newspoll compared to others and that is exactly what I did a week ago when Newspoll came out with its finding that support for the Labor government had dropped seven percentage points in a fortnight. I wrote a week ago in Crikey :
I have studied goodness knows how many opinion polls over the years I have been involved in running election campaigns but I have never come across anything remotely like a seven percentage point change in support for a political party in the course of a fortnight. Something very strange has gone on and I simply cannot take this morning’s Newspoll result seriously. It will take another couple like it   —   and with all the other pollsters pointing in the same direction  —  for me to change my mind.
Since then we have had a Morgan and Nielsen poll showing no such dramatic change in support for Labor but the references to a dramatic drop in support have continued.
The popular conclusion of journalists has been that Labor’s handling of a boatload of 78 Sri Lankans found in Indonesian waters off the north-west coast of Australia is the reason for this unsubstantiated drop in support.
I say unsubstantiated because Nielsen asked questions about boat people policy at the same time as it found only a very slight decline in voting support for Labor. If Labor’s vote has not fallen, it is hard to argue that Labor is suffering because of its boat people policy!
Newspoll and The Australian, knowing that the major polling competitor AC Nielsen was about to publish its monthly survey of opinion, were obviously a little concerned that their revelations of the previous week showing a collapse in the Labor vote were about to be tested, chose to engage in a little damage control. They conducted a special poll asking people about Labor’s policies on boat people. That was all well and good but they dingoed out on also asking the normal questions about voting intentions.
Which brings me to Gary Morgan and his Morgan Poll, which, when I was a boy, had the same influence on journalists as Newspoll does today. He is clearly puzzled by why the Newspoll finding on voting intention is so different to his own company’s. I think his comments on these recent differences between pollsters are worth recording:
Last week, November 3, News Ltd’s poll (Newspoll) claimed that Coalition support had jumped 7% to 48% while ALP support had dropped 7% to 52% in two-party preferred terms.
Even at face value, such a drop in support was extremely unlikely.
On Thursday November 5, when the Roy Morgan Consumer Confidence Rating showed an increase in consumer confidence (up 2.5pts to 128.0), it was clear to those who follow polls and politics that the Newspoll was wrong  — a “rogue” poll. An increase in consumer confidence is almost invariably positive for the government.
On Friday November 6, the Morgan Poll showed two-party preferred support for the ALP virtually unchanged (up 0.5% to 61%) and then today’s Nielsen poll also showed support for the ALP virtually unchanged (down 1% to 56%).
The evidence was clear. Yet the publication of News Ltd’s poll (Newspoll) in the first place had already had a major impact. The evidence showing the “error” of Newspoll was literally ignored by media discussion (e.g. the Insiders on ABC TV and the impact of the “rogue” poll was allowed to run unabated).
Pollsters and those who publish the polls have a responsibility to report the facts and the truth.
Newspoll should have conducted another poll as soon as possible when it saw the dramatic change in their results  — and if different, released the data to correct the misconceptions caused by their “rogue” poll.
It is extremely worrying that today’s Newspoll on “boat people” clearly did include questions on “Political support,” but the results from the “Political support” question were not published.
A statistical analysis of the data reported on Australians’ attitudes to “boat people” issues  — specifically the breakdown by “political support”  — suggests the ALP vote in that poll was very strong. The percentage supporting each political party clearly should have been released.
Polls and their publishers should not seek to set the agenda by selectively releasing polling data.
Polls and their publishers are powerful but with that power comes responsibility.
Post a Comment