Daily Media Quotation
Power Of Polls In Way People Misread Them
October 6, 2006
by Dennis Shanahan - The Australian
I have been involved in the reporting of Newspoll surveys for almost 20 years and have written more newspaper reports and analysis on Newspoll surveys than anyone else on earth.
After months of an intensifying debate about polling and Labor's present standing, it's time to talk about a few home truths.
As an illustration of how polling, and in particular Newspoll, plays a role in the rise and fall of politicians and their parties, here are some historical examples.
In November 1994 a Newspoll survey showed Alexander Downer's support as Opposition leader falling for the second time and I suggested it was the beginning of a decline. An astute and avid poll reader came to my office and said he thought my report was unkind to Downer. Within four months that astute poll reader, John Howard, was leader of the Liberal Party and within another year was Prime Minister.
In mid-1998 I wrote a story beginning: "The Labor Party is in a position to win government after Coalition support slumped to the lowest on record." Within five months Howard was re-elected as Prime Minister after defeating Kim Beazley as Labor leader for the first time.
In late 2003 I wrote a news story suggesting that Simon Crean's supporters were saying that if the polling got worse he would step aside as Opposition leader in favour of Mark Latham. Crean took exception to the article and carpeted me. Much was being made by his critics of the negative differential in Newspoll - that is, the difference between people who were satisfied with him and those who were dissatisfied - then between 20 and 35 percentage points.
Within months Crean stepped aside in favour of Latham.
In the lead-up to the 2004 election, the ALP under Latham looked competitive and was reported as such. As well, Latham's personal support was way up on Crean and equal to Howard at one stage. Latham lost the October 2004 election with a second successive fall in Labor's primary vote.
All this shows that polls are not predictive. It also shows that polls can have a dire effect on political leaders through the influence they have on their parliamentary colleagues. This is the political poison of polling and what makes them so hotly contested.
These two simple propositions have been lost in the present debate. Indeed, the fight over polling is more like a drunken wedding reception between feuding religions: lots of prejudice and suspicion, name-calling, false premises, superficial and glib judgments, partisan positions, confused logic, wild air punches and pissing in the wind.
For more than a year my Friday counterpoint, Michael Costello, a former chief of staff to Beazley, has complained about Newspoll's presentation and reporting in The Australian; Labor MP and former national secretary Bob McMullan has made a critical contribution in parliament; former Bob Hawke staff member and ABC journalist Barrie Cassidy has complained that the polling analysis suggests Labor can't win the next election; my Canberra bureau colleague Matt Price has written in The Sunday Telegraph of Newspoll's ostensible misreporting; several conspiracies have appeared on websites; and, of course, The Australian and Newspoll chairman and founder Sol Lebovic have responded.
Yet, in all of this, the sense of air punches, self-interest and cross-purposes is palpable. Arguments are mounted without even differentiating between primary vote and second preferences; complaints are made on a hotchpotch average of different polls; leadership is deemed a key political point or dismissed as over-emphasised, and equated with the party's performance; polls are accused of making predictions or not; what people are saying in snapshot is not separated from how they will ultimately vote; and margins of error are claimed not be given sufficient weight, then complaints are made that Labor is not being given due credit for leads of less than the margin of error.
What it is all about at the moment is boosting Labor's image as a winner in the electorate, convincing Labor MPs that the mammoth task they face of winning more than 16 seats is not beyond them and shoring up Beazley's position as leader.
Newspoll has Labor on 42 per cent and the Coalition on 41 per cent on primary vote and 53 per cent to 47 per cent on second preferences, while Beazley's satisfaction and dissatisfaction ratings are 32 per cent and 51 per cent respectively. Howard's satisfaction and dissatisfaction ratings are 47per cent and 41 per cent and, on the question of preferred prime minister, Howard rates 52 per cent compared with Beazley's 26 per cent.
This is good for Labor; not so good for Beazley.
The catchcry of the critics is that Labor is competitive and can win.
Well, who has said it isn't and can't in the reporting of Newspoll? This argument is about covering the fact that both the party and the leader have to be competitive and one cannot succeed without the other performing well.
While latterly complaining of a lack of due recognition in The Australian of the ALP's competitive position in Newspoll, Costello has avoided talking about the net approval rating of Beazley. In early 2003 he pointed out that Crean's "net approval rating modestly improving from minus 17 per cent to minus 10 per cent" was hardly a political breakthrough. Beazley's net approval rating is minus 19 per cent.
Cassidy complained on ABC radio and The Insiders that the headlines didn't reflect Labor's lead and that more emphasis was being placed on the popularity of the leader. But as an old hand Cassidy knows that two-party-preferred figures putting Labor level with the Coalition are worth little if they are not built on a primary vote of least 40 per cent, something Labor has achieved consistently only in the past three Newspoll surveys.
This is where the emphasis on second preferences, a calculated figure based on flows at the previous election, and which Cassidy concedes people are not sure about until the election campaign anyway, can be misleading. Labor has failed many times, even with a two-party-preferred vote of more than 50 per cent, because the primary vote was below 40 per cent.
McMullan was far more forensic and careful in his analysis. He went out of his way to defend the veracity of the polls and their place in political debate. His argument, based on an average of polls, was that on two-party-preferred terms, Labor had been in front of the Coalition for most of the year, yet it was wasn't reflected in what people thought.
He complained about five-point swings in the Newspoll because 600,000 people could not possibly have switched party allegiance.
Well, he's right: the polls tell you only what people are thinking now and there appears to be a tendency to use the polls to give a message to the Government or Opposition. The poll results are facts, not prophecies, and give a window into how people are thinking on a given day.
But when they are continuously high or low, they send a message only the deluded or foolish ignore.