Inside Story

The war the bloggers won

Political bloggers brought a new rigour to interpreting the polls

Greg Jericho 23 August 2013 2886 words

The polls were right: election night 2007. Ruben Schade/Flickr

SOME people saw the arrival of political blogs in the early 2000s as the beginning of a war between the bloggers and the traditional media. If that’s true, then the bloggers have won, at least when it comes to reporting about polls.

To write about things differently from the traditional media, bloggers draw heavily on data and official reports. In fact, there is so much data around that it isn’t surprising that most of it goes unreported by journalists working for newspapers, television or radio. This provides ample space for those bloggers who enjoy spending their time mining numbers. They do it not because such things are newsworthy, but because they believe they provide a broader and more detailed picture than mainstream newspapers and broadcasters do.

But when it comes to straight politics, the traditional media have the advantage of location and access. They have staff based in Parliament House; they can get a minister on the phone; they get “off the record” murmurings. Some of the benefits of this level of access can be overrated, and many of those who blog about politics have their own contacts within party organisations. Some bloggers will have once worked for a party; others work within a union and are aware of deep, secret manoeuvrings.

Quite often political bloggers draw on the news reported by those in the traditional media. While this often results in media criticism rather than new information, media organisations’ increasing use of polls has opened up an opportunity for a group of people who have not only critiqued the media’s political coverage but also changed the way reporting is done. This group of election analysts and polling nerds even brought with them a name – psephologists.

The psephologists sprang from different places, but mostly from academia. Some were students of politics, others were statisticians. But while they seemed to be examining just a narrow field of the political debate, their influence and the ructions they caused media organisations here and in America were among the largest achieved by any branch of political blogging.

This happened for two reasons: analysis of polls didn’t need require any advantages of access; and their criticisms served to undermine a key business model of the media outlets who commissioned the polling.

Psephologist bloggers had their first significant impact in Australia in the lead-up to the 2007 election. Throughout that year, the fortnightly Newspoll numbers, like those of most other polls, were bad for the Liberal Party and John Howard. Labor had led the polls before Kevin Rudd took over from Kim Beazley in 2006, but with Rudd as leader the lead strengthened.

In the Australian, Dennis Shanahan, who was given chief responsibility for commenting on Newspoll’s findings, could see this as clearly as anyone. He had to see it, because to deny it would have been to suggest that Newspoll, part owned by the Australian, was wrong.

Early in the year, on his political blog on the paper’s website, Shanahan wrote that “John Howard can lose the next election.” But like a sports commentator anxious to keep the audience interested in a match that already seemed won, he kept seeing changes in the wind throughout the first half of 2007:

10 April: “The last Newspoll taken for the Australian, which is not included in the analysis for the first quarter, showed a drop for Labor and a slight cooling for Rudd personally.”

1 May: “All of a sudden, the Howard government is getting a scent of a chance, just a chance, that it can slow the juggernaut that has been Kevin Rudd since December.”

4 May: “Howard’s chances of winning are improving and Labor is going to need more than a quick fix.”

The analogy of a sports commentator fits nicely with how and why polls are covered in the media. Political journalists love nothing more than to portray themselves as dispassionate observers of events, and charges of bias always cut even the most prejudiced journalist to the quick. And polls seem to be unbiased. They are merely numbers, and so journalists are able to report on the political situation exactly as a race-caller would – letting us know who is ahead, who is coming up fast, who is dropping behind.

The other similarity with sports commentary is that news organisations like the Australian pay for polls. In essence, the ability to report on the poll (first, and with the authority of an owner) is much the same as a television station buying the rights to telecast football. Television sports commentators will continue to see chances of a comeback from a team not because they are deluded, but because they need to do all they can to keep the audience interested. If the margin gets too wide they will look for other things to focus on – will the full-forward kick ten goals, will the defeat break a record, will Gary Ablett Jr get forty possessions? It’s a commercially founded decision.

The same applies to the polls. The task is to persuade readers they are interesting and deserve to be on the front page. There’s not much point spending a few tens of thousands of dollars commissioning a poll if you’re going to bury it on page eight. The commentary needs to find something front-page-worthy even if the actual numbers are not.

For the Australian, Newspoll has taken on an even greater purpose – that of influence. Although it’s now just one of a number of polling organisations, Newspoll is treated as the most important among those involved in the theatre of federal politics. For this to continue, the Australian needs to continue to talk up its importance.

BUT in 2007 the bloggers were not interested in the hype; just the numbers. They not only used other polls to get an aggregated picture but also brought with them statistical analysis not seen in the standard reporting of polls – even such a simple thing as noting the “margin of error” of each poll. For such bloggers a 1 to 2 per cent shift in a poll was effectively no swing at all; for those in the media it was worth holding the front page for.

With the two-party-preferred polling remaining stubbornly in Labor’s favour as the year wore on, Shanahan increasingly focused on the “preferred prime minister” metric, which looked better for the Coalition. In the past, this shift in emphasis would have attracted little public criticism: a letter to the editor perhaps? Few political journalists are versed in the intricacies of statistical analysis, so virtually no one else in the press gallery was likely to question such reporting.

But by this time the Australian political blogosphere had already developed into quite a robust community. Among those blogging about polling were political science PhD students William Bowe and Peter Brent on their respective blogs, The Poll Bludger and Mumble, and econometrician Scott Steel, under the pseudonym Possum Comitatus, on his blog, Pollytics.

Others were also contributing elsewhere. Andrew Catsaras, a former lecturer in marketing, was posting on the now-defunct Ozforums under the pseudonym of Aristotle; expatriate Queenslander Simon Jackman, now a professor of politics at Stanford University, was providing regular analysis of the polls on his university blog; and psephologist Charles Richardson was writing for Crikey.

They were a pretty intellectually heavy group, and their common approach was to present analysis that went counter to all the rules followed by political journalists in the traditional media. Unburdened by the need to sell the polls, they instead presented detailed, sometimes quite academic analysis, which more often than not flew in the face of what commentators were asserting in the major papers. Among the most common point of conjecture was Dennis Shanahan’s focus on the preferred prime minister figures. Brent was among the most scathing about the worth of that measure in determining election results.

The Australian didn’t take kindly to the suggestion not only that their analysis was wrong, but also that the (generally small) fortnightly movements of Newspoll didn’t deserve the attention the Australian was giving them. Although these bloggers were mostly on the periphery of the media, read by an audience much smaller than any major newspaper’s, the editors of the Australian decided it was time to put things back in their proper place.

On 11 July that year, Shanahan used his blog to defend his credentials and explain his job under the title “Howard’s trend lifts him out of the trough.” In doing so, he confirmed the very criticisms that had been levelled at him by the bloggers:

My job is simply to tell people what the most interesting aspect of the latest Newspoll figures are and to put them into the perspective of reporting on those surveys for the last fifteen years or so.

For a start, there’s no interest in saying the latest polls haven’t changed – the interest, politically and journalistically, is where has the change occurred. After all, if there were no changes at all in any category there’d be no point in reporting the latest poll.

Shanahan, nicely and perhaps unwittingly, had explained the difference in mindset between mainstream political journalism and the bloggers. The Australian’s main purpose for running Newspolls is profit. It is a tool used to get people to buy the paper, purely, simply, completely. It hopes that the Newspoll will be viewed with a high level of seriousness in the political sphere and the Australian’s coverage of the Newspoll will thus play a role in the news of the day.

Shanahan’s defence also opened him up to other charges. His assertion that the Australian only reported things that change was rather disingenuous. The paper might always report on shifts in the polls, but the prominence various movements were being given within the paper could be starkly different. The bloggers took great delight in finding numerous examples.

Once again the Australian fought back, this time bringing in the CEO of Newspoll, Martin O’Shaughnessy, to add some intellectual heft to its argument. In the process, the field of battle shifted from Shanahan’s blog to the editorial page. Brent was phoned by the editor of the Australian, Chris Mitchell, who told him they were going to “go” both him and Charles Richardson in the next day’s editorial.

The editorial was titled “History a better guide than bias” and subtitled “Online prejudice no substitute for real work.” Amid this long defence of Shanahan’s analysis came perhaps the most priceless unintentionally humorous sentence written in a newspaper editorial. “Unlike Crikey,” intoned the paper, “we understand Newspoll because we own it.” The editorial also contained a word of advice for those blogging away: “The self-appointed experts online come instead from the extreme Left, populated as many sites are by sheltered academics and failed journalists who would not get a job on a real newspaper.”

So how did this battle end? In May 2010 the Australian hired Peter Brent to write a regular polling blog on its website.

And Brent is not the only one to find his way (by and large most psephologist bloggers are male) into the media. William Bowe and Scott Steel both moved their blogs onto Crikey’s website just prior to the 2007 election. Andrew Catsaras now provides a weekly report on the polls on the ABC’s Insiders program, and Simon Jackman writes a weekly blog post for Guardian Australia.

Uniquely among these bloggers, Jackman has also written for media outlets about a United States presidential election. In 2012 he wrote for the Huffington Post, using similar methods to predict the electoral outcomes to Nate Silver’s for the New York Times.

SILVER had come to prominence during the 2008 presidential race when, on his blog FiveThirtyEight, he correctly predicted forty-nine out of the fifty states and all thirty-five Senate races. To do this he didn’t just add up the polls; he brought high-level technical expertise to the game – including statistical tools such as Monte Carlo simulations. At that time he was still on the periphery, but Scott Steel recalls being in contact with him because Steel was using the same simulations to determine election results here in Australia.

And, like Peter Brent and the Australian, the New York Times hired Silver in 2010. By election year, 2012, his page was among the most visited on the paper’s website. At times during the final week of the campaign he was drawing around 70 per cent of all visits to the Times’s website for political news.

But Silver’s and Jackman’s approach of aggregating polls and trusting the numbers irked those reporters and commentators who valued more the importance of access. Neither Jackman nor Silver ventured out on the campaign trail to watch the candidate and talk to his staff to get the inside word on how they were travelling, or ask what their party polling was telling them.

During the 2012 campaign, a number of American journalists took the same approach as Shanahan had in 2007, and rebelled against these number jockeys. The old way of reporting was seen to be under attack. After all, if these bloggers were already 90 per cent sure of the outcome of the election, what was the point of following the candidates around on a cramped bus and eating takeaway for three months?

Most infamously, Peggy Noonan, writing for the Wall Street Journal on the day before the election, took aim at those who viewed the calculator as more valuable than the gut feeling of an experienced reporter:

Who knows what to make of the weighting of the polls and the assumptions as to who will vote? Who knows the depth and breadth of each party’s turnout efforts? Among the wisest words spoken this cycle were by John Dickerson of CBS News and Slate, who said, in a conversation the night before the last presidential debate, that he thought maybe the American people were quietly cooking something up, something we don’t know about.

I think they are and I think it’s this: a Romney win.

The next day Obama won exactly the number of states predicted by Silver and Jackman, and by other bloggers who used the appropriate quantitative methods.

Despite this success, things still jarred between Nate Silver’s approach and the way things worked at the New York Times. He was criticised during the campaign for making a bet with MSNBC talk-show host Joe Scarborough on the outcome of the election. The New York Times’s public editor, Margaret Sullivan, described the action as “inappropriate for a Times journalist, which is how Mr Silver is seen by the public even though he’s not a regular staff member.”

Silver recently left the Times and has taken his talents to ESPN, a global television network, where he will write on politics and on sports. This is not particularly surprising: before writing on polls he had devised a system for using baseball statistics to rate player and team performance.

IN AUSTRALIA, meanwhile, the war over how to analyse the polls is over. The bloggers have won and, rather than fight, journalists now use the knowledge gained to improve their reporting. In the second week of the current election campaign, for example, the ABC’s Sabra Lane provided this report on a poll:

On two-party-preferred basis, the Coalition is easily ahead at 52–48. That’s no difference from the last poll taken a week ago. Now Newspoll interviewed 1134 people – the margin of error for this particular poll is plus or minus three points – and most movements within this survey are within that margin of error.

The use of margin of error has now become almost ubiquitous. Some reports of polls will still leave such technicalities out, but more and more that’s a sign that the article is attempting to report on the poll in the old way – where small movements are exaggerated for effect by the journalist to sell the poll rather than provide sensible analysis. More importantly, such reporting is no longer repeated without question by other news organisations. The psephologists have given them some tools to question the old way.

Interestingly, the Australian Financial Review no longer tries to beat the bloggers at all and has, in effect, joined them. Rather than rely entirely on its own poll, data journalist Edmund Tadros has aggregated all publicly available polls with weighted averages and produced a much more nuanced examination of the movements.

As was the case with the bloggers, this approach effectively devalues the importance of individual polls, but it gives readers much greater understanding and a better picture of the real situation.

It didn’t stop there. In the second week of the campaign the paper’s Jason Murphy analysed the betting markets of every seat. And in a sign that the fight is well and truly over, he used a Monte Carlo simulation to determine the likely election result.

The bloggers won, but mostly because some journalists and news organisations were smart enough to realise it wasn’t a war at all, just a case of more knowledge being available. And like all smart journalists are taught to do, they used it. And no one tries to argue any more that people who don’t own the polls don’t know what they mean. •