How polling is wreck­ing democ­racy

As pub­lic faith in democ­racy col­lapses, the in­sti­tu­tion is fur­ther un­der­mined by sus­pect polling, gorm­less pol­i­tics and a me­dia de­pen­dent on both. Mike Sec­combe re­ports.

The Saturday Paper - - Front Page - MIKE SEC­COMBE is The Satur­day Paper’s na­tional correspondent.

There can be few tasks in the life of a po­lit­i­cal jour­nal­ist as te­diously de­mand­ing as the writ­ing up of opin­ion polls. Most of the time it in­volves the con­jur­ing of an il­lu­sion of news.

In the vast ma­jor­ity of those sur­veys of electoral sup­port, noth­ing re­ally changes. Maybe the Coali­tion or La­bor has gone up by a per­cent­age point, maybe the pre­ferred prime min­is­ter mea­sure has bounced a lit­tle one way or the other.

But any po­lit­i­cal jour­nal­ist worth their pay knows this is ir­rel­e­vant. Those who pro­duce these pub­lished polls – Ip­sos in the Fair­fax pa­pers, YouGov Galaxy, which does Newspoll for the Mur­doch pa­pers, Es­sen­tial in Guardian Aus­tralia, and var­i­ous other polling out­fits – al­ways ap­ply the caveat of a 2.5 to 3 per cent mar­gin of er­ror in the sam­pling. Any move­ment smaller than that is prob­a­bly sta­tis­ti­cal static. In this coun­try, ev­ery fort­night, it is also front-page news.

Sure, once in a very long while some­thing dra­matic hap­pens, as it did in Au­gust this year when the right-wing coup in the Lib­eral Party saw Mal­colm Turn­bull dumped from the prime min­is­ter­ship and Scott Mor­ri­son in­stalled in his place. Then the polls swung wildly. By Newspoll’s mea­sure, the two-party pre­ferred mar­gin went from two points (51-49) in favour of La­bor in its poll pub­lished on Au­gust 12, to 12 points (56-44) on Au­gust 26.

Now, that’s a story. But it was very much the ex­cep­tion. At best, most polls in the var­i­ous me­dia con­firm longer-term trends.

Yet they have be­come an un­healthy ob­ses­sion in Aus­tralian pol­i­tics. To an im­pov­er­ished me­dia, says Greens leader Richard Di Natale, they have be­come “a sub­sti­tute for jour­nal­ism”. The un­due em­pha­sis placed on them, says ALP pres­i­dent and for­mer trea­surer Wayne Swan, “de­val­ues pol­i­tics” and threat­ens to be­come a sub­sti­tute for lead­er­ship.

Polling re­sults have been a sig­nif­i­cant fac­tor in the demise of Aus­tralia’s past four prime min­is­ters. In two cases, Kevin Rudd’s and Tony Ab­bott’s, there were other ma­jor is­sues, too, re­lat­ing to their style of lead­er­ship. But the other two, Ju­lia Gil­lard and Mal­colm Turn­bull, were cut down purely and sim­ply be­cause of poor polling.

Used prop­erly, sur­veys of pub­lic opin­ion can be a use­ful tool. But they can also be very dan­ger­ous. As Di Natale says: “They have how been weaponised, by peo­ple across the po­lit­i­cal di­vide, even by peo­ple within the same party. They have be­come a big part of the prob­lem in our democ­racy.”

Polls come in dif­fer­ent kinds. Most fa­mil­iar are the reg­u­lar sur­veys that ap­pear in the print me­dia, telling us which party is ahead in the race to the next elec­tion. There also are is­sues polls that seek to mea­sure, and of­ten to in­flu­ence, opin­ion on one sub­ject or an­other. Both have their prob­lems.

Let’s start with the po­lit­i­cal polls, which are re­ported in a way that be­muses and alarms peo­ple who ac­tu­ally un­der­stand the uses and abuses of pub­lic opin­ion sam­pling.

“Newspoll comes out and shows La­bor’s gone up 1 per cent and the gov­ern­ment’s gone down 1 per cent, and

it’s to­tally within their mar­gin of er­ror and it means they got the same re­sult as last time,” says Pro­fes­sor An­drew Markus, of Monash Univer­sity.

“But it gen­er­ates head­lines and has ev­ery­one run­ning around Par­lia­ment House say­ing, ‘What are we go­ing to do?’

“It’s ex­tremely desta­bil­is­ing. It feeds the 24-hour news cy­cle, and peo­ple hang out for it. But it is not a qual­i­ta­tive mea­sure of what the gov­ern­ment is do­ing.”

That is to say, while these polls in­di­cate whether a gov­ern­ment is pop­u­lar, they do not give any clue as to why.

Therein lies a prob­lem: when the num­bers don’t mean much, mean­ing must be at­tached to them. As the ex­am­ple of the re­cent Lib­eral coup shows, that can lead to dis­as­ter.

The cur­rent gov­ern­ment lost 38 con­sec­u­tive Newspolls un­der Turn­bull’s lead­er­ship. The mean­ing at­tached to this by the right wing of the gov­ern­ment was that the fault lay with the leader who, they ar­gued, had failed to ap­peal to the con­ser­va­tive “base”.

Only af­ter the lead­er­ship change has it be­come ap­par­ent to all but the most de­luded that the fault lay less with the leader than the party.

Politi­cians aren’t the only ones apt to in­ter­pret – or mis­in­ter­pret – the poll num­bers to suit their agen­das. So do all man­ner of peo­ple with vested in­ter­ests, in­clud­ing the news or­gan­i­sa­tions that pub­lish them.

The pri­mary agenda of the me­dia is com­mer­cial. They are sell­ing a prod­uct.

“When a me­dia out­let com­mis­sions a poll, they don’t do it as an in­tel­lec­tual pur­suit, they want to get a strong story out of it,” says in­de­pen­dent polling an­a­lyst An­drew Cat­saras.

Iron­i­cally, it tends to be the case that the more static the poll num­bers, the more re­porters are apt to push what Cat­saras calls “the bounds of rea­son­able­ness” in their ef­forts to find sig­nif­i­cance in the num­bers.

If, for ex­am­ple, the head­line num­ber – vot­ing in­ten­tion – hasn’t moved, it is likely that the story will em­pha­sise some other mea­sure, like the leader’s pop­u­lar­ity.

“But,” says Cat­saras, “the leader’s rat­ings don’t mean much, as they al­ready are fac­tored into the vot­ing in­ten­tion. The pri­mary vote is the most im­por­tant thing, and from there the cal­cu­la­tion of the twoparty pre­ferred vote, which by and large is pretty ac­cu­rate. But the other mea­sures [leader’s ap­proval, pre­ferred prime min­is­ter] are ba­si­cally win­dow-dress­ing.”

Apart from the fact the news­pa­pers “like to make it a beauty con­test”, he says, some also push a po­lit­i­cal agenda.

“If their party isn’t do­ing too well, they will em­pha­sise the pos­i­tive, the thing that suits them. The stuff you read in Fair­fax is far more mea­sured than what you read in News. News is al­most al­ways push­ing a po­lit­i­cal bar­row, which is the con­ser­va­tive cause.

“The clas­sic ex­am­ple was 2007, when La­bor was ab­so­lutely ham­mer­ing the Coali­tion un­der Howard in the twoparty pre­ferred vote, and The Aus­tralian

… grasped on to pre­ferred prime min­is­ter, and wrote a se­ries of sto­ries about how Howard was com­ing back. It was bull­shit. At that time, var­i­ous blog­gers in­clud­ing my­self and Greg Jeri­cho wrote about it.”

The Aus­tralian re­sponded in print with a vig­or­ous de­fence of its fo­cus on the prime min­is­ter’s per­for­mance, but come the elec­tion Howard’s gov­ern­ment not only lost, Howard him­self be­came only the sec­ond prime min­is­ter in his­tory to lose his seat.

Cat­saras stresses that the bias he sees is not the fault of the poll­sters.

“These high-qual­ity out­fits – Galaxy, Newspoll, Ip­sos, Es­sen­tial, ReachTEL – aren’t do­ing any­thing wrong. The prob­lem lies in the in­ter­pre­ta­tion of the find­ings,” he says.

And while the Mur­doch me­dia are the worst when it comes to spin­ning the data, he says, “their poll­ster, Newspoll, is still the best”.

David Briggs, manag­ing di­rec­tor of YouGov Galaxy, which now pro­duces Newspoll, thanks Cat­saras for his en­dorse­ment, but also strongly dis­agrees about the im­por­tance of the pre­ferred prime min­is­ter mea­sure.

“It is ab­so­lutely rel­e­vant,” he says. “When you look back over the past 30 years, you find the leader of the Op­po­si­tion has got to be either ahead of the prime min­is­ter, or very close, to win. There hasn’t been a time when, at the elec­tion, the ex­ist­ing prime min­is­ter has had a sig­nif­i­cant lead over the leader of the Op­po­si­tion and lost the elec­tion.”

He does agree with Cat­saras, how­ever, that vot­ing in­ten­tion is the most im­por­tant num­ber. He also agrees with Cat­saras’s puz­zle­ment at the cur­rent state of the pre­ferred prime min­is­ter mea­sures across the var­i­ous polls, which con­sis­tently show Bill Shorten as un­pop­u­lar.

“It’s not as if peo­ple don’t know him af­ter five years. It’s a mys­tery why he hasn’t done bet­ter,” Briggs says.

Shorten has brought his for­merly frac­tured party to­gether, is tac­ti­cally smart, has put for­ward bold and mostly pop­u­lar pol­icy ideas and is strong on de­tail.

Why don’t peo­ple like him? Yes, he played a role in bring­ing down pre­vi­ous lead­ers of the party. But so did Mal­colm Turn­bull, and the pub­lic liked him. Yes, Shorten has strong trade union links, but Bob Hawke was leader of the Aus­tralian Coun­cil of Trade Unions be­fore he be­came La­bor leader – and also knifed Bill Hay­den, and was a for­mer wom­an­iser and heavy drinker – and the pub­lic loved him.

Off the record, one poll­ster con­fesses he could put it down to only one thing: “It’s the way he looks and sounds.”

The point is that mod­ern pol­i­tics and me­dia en­cour­age su­per­fi­cial judge­ments. Con­tem­po­rary politi­cians get a grab on TV or ra­dio for a cou­ple of sec­onds if they are lucky, and the re­sult is that it’s not what they say that im­presses but the way they look and sound while say­ing it. And once a neg­a­tive per­cep­tion takes hold, it is con­stantly re­in­forced in the re­port­ing of the polls.

It is not only the party polling that en­cour­ages su­per­fi­cial judge­ments. We also are del­uged with is­sues polls. In some re­spects, they are even more du­bi­ous than polls of vot­ing in­ten­tion. As An­drew Markus notes, poll­sters have had a long time to de­velop sam­pling and weight­ing for­mu­las for po­lit­i­cal polls, and to re­fine them by com­par­i­son with ac­tual elec­tion re­sults. That is less true of is­sues polls. The is­sues about which re­spon­dents are be­ing asked also are of­ten far more nu­anced than the sim­ple mat­ter of choos­ing one po­lit­i­cal party or the other.

For ex­am­ple, a re­cent Fair­fax–Ip­sos poll sur­veyed at­ti­tudes on cli­mate change and en­ergy se­cu­rity. It asked re­spon­dents whether they thought gov­ern­ment should pri­ori­tise (a) re­duc­ing house­hold power bills, (b) re­duc­ing car­bon emis­sions, or

(c) re­duc­ing the risk of black­outs.

Over­all, re­sponses were 47, 39 and 13, re­spec­tively. But where was the op­tion for peo­ple who un­der­stand the re­al­ity that re­new­able en­ergy is now cheaper, cleaner and, if done right, more re­li­able than power gen­er­ated by fos­sil fuel? Where was op­tion (d) “I be­lieve all three goals can be achieved si­mul­ta­ne­ously”?

When I put it to Ip­sos di­rec­tor Jes­sica El­good, who runs the polling pro­gram for Fair­fax, that the poll had pre­sented an es­sen­tially false choice, she con­ceded the dif­fi­culty of get­ting a mean­ing­ful re­sponse from a sin­gle ques­tion.

“When you have com­plex is­sues like that, you need either a far longer in­ter­view, where you can spend 20 or 30 min­utes tak­ing peo­ple through the var­i­ous el­e­ments of the topic, or a fo­cus group or work­shop dis­cussing the sub­tleties,” she said.

And while her or­gan­i­sa­tion, like other big polling out­fits, does such in­depth anal­y­sis of opin­ion for some clients, in­clud­ing gov­ern­ment de­part­ments,

“they are not the kind of sur­veys that gen­er­ally make it onto the front pages of news­pa­pers”.

She claimed, none­the­less, that even a sin­gle ques­tion pro­vided “in­sight”. But much de­pends on how the ques­tion is asked.

All too of­ten, says Wayne Swan, is­sues polls are com­mis­sioned by those with vested in­ter­ests, and the ques­tions asked are not de­signed to mea­sure opin­ion but to in­flu­ence it.

“If, for ex­am­ple, you put the ques­tion on cap­i­tal gains tax re­form by ask­ing peo­ple if they are in favour of re­form to take away huge tax breaks from peo­ple on high in­comes, they’ll say, ‘Yeah, we’re in favour of that,’” he says. “If you ask if they are in favour of tax re­form that will tank the prop­erty mar­ket, you’ll get a dif­fer­ent re­sult.”

Even when those ask­ing the ques­tion have the best of in­ten­tions, the fram­ing of is­sues is hard.

A strong il­lus­tra­tion of this point can be seen in at­tempts over re­cent years by the Lowy In­sti­tute to mea­sure pub­lic at­ti­tudes to Aus­tralia’s for­eign aid bud­get.

In 2017, the Lowy poll ques­tion told re­spon­dents that Aus­tralia in­vested $3.8 bil­lion in over­seas aid, and asked if they thought that was too lit­tle, too much or about right. Only 22 per cent thought we should spend more.

The gov­ern­ment seized on this re­sult to jus­tify mas­sive cuts it had made in the area.

This year, how­ever, Lowy posed the ques­tion an­other way, ask­ing peo­ple what pro­por­tion of the fed­eral bud­get they thought was spent on aid, and what pro­por­tion should be spent on it. On av­er­age, they guessed ac­tual spend­ing to be 14 per cent of the bud­get, and thought it should be re­duced to 10 per cent.

In re­al­ity, aid spend­ing makes up only 0.8 per cent of the bud­get.

What in­sight is gained from this? That Aus­tralians are stingy, or gen­er­ous, or just con­fused by large num­bers? El­good sug­gests the third.

Eco­nomic mat­ters in par­tic­u­lar do not lend them­selves to mean­ing­ful polling, as was ex­em­pli­fied in a re­cent Es­sen­tial poll on neg­a­tive gear­ing. The very long ques­tion first sought to ex­plain what neg­a­tive gear­ing was, then asked if re­spon­dents thought pro­posed changes would drive house prices up or down or have no ef­fect on them. The big­gest re­sponse was “don’t know”, fol­lowed by “no ef­fect”, which was likely a shy way of say­ing the same thing. Of the mi­nor­ity who had an opin­ion one way or the other, it is likely most were sim­ply re­flect­ing their pre-ex­ist­ing po­lit­i­cal al­le­giance.

But it’s not just eco­nom­ics. “A lot of is­sues are far too com­plex for the av­er­age voter,” says Briggs, “and that’s why we elect rep­re­sen­ta­tives to make those de­ci­sions for us.”

The con­cern is, how­ever, that as pub­lic trust in our elected rep­re­sen­ta­tives de­clines – in­deed, trust in in­sti­tu­tions and ex­per­tise in gen­eral – polls be­come part of an am­pli­fi­ca­tion sys­tem for ig­no­rance.

A 2017 Ip­sos re­port, “Per­ils of Per­cep­tion”, high­lighted just how wrong peo­ple are, across 38 coun­tries, about a wide va­ri­ety of is­sues.

Al­most ev­ery­where, for ex­am­ple, peo­ple were con­vinced the mur­der rate was go­ing up, whereas it had in­creased sig­nif­i­cantly in only two coun­tries – Mex­ico and Peru. In Aus­tralia, mur­ders de­clined 47 per cent be­tween 2000 and 2017, but only 17 per cent of re­spon­dents thought this was the case. Twice as many wrongly thought the op­po­site.

The re­port showed peo­ple greatly over­es­ti­mated the in­ci­dence of var­i­ous things they feared, from the num­ber of deaths due to ter­ror­ism to the crim­i­nal pro­cliv­i­ties of im­mi­grants. In the for­mer, the fig­ure is hugely down, glob­ally, in the 15 years af­ter Septem­ber 11, 2001, com­pared with the 15 years be­fore 2000, and in Aus­tralia down from seven to six. In the lat­ter, Aus­tralians thought mi­grants made up 40 per cent of the prison pop­u­la­tion, more than twice the ac­tual pro­por­tion.

The sub­head­ing of the Ip­sos re­port was “Things are not as bad as they seem”. But some me­dia and politi­cians find ad­van­tage in em­pha­sis­ing neg­a­tive per­cep­tions in­stead of data-based re­al­ity. Thus, for ex­am­ple, we con­tinue to see them run­ning law and or­der cam­paigns de­spite fall­ing crime rates.

Last month, the Fair­fax pa­pers pro­vided a clas­sic ex­am­ple of a di­vi­sive, poll-based beat-up. The sub­ject was the level of Mus­lim im­mi­gra­tion.

“A special Fair­fax–Ip­sos sur­vey finds only 14 per cent of vot­ers sup­port an in­crease in the num­ber of im­mi­grants from Mus­lim coun­tries while 35 per cent be­lieve the in­take should stay the same. But an­other 46 per cent be­lieve the in­take should be re­duced a lot or a lit­tle…” it read.

The take­out mes­sage seemed clear: al­most half of vot­ers were con­cerned about bring­ing more Mus­lims into the coun­try. But it was mis­lead­ing by virtue of omis­sion, as Michael Pas­coe pointed out in an ex­co­ri­at­ing piece for The New Daily.

An­other Fair­fax–Ip­sos poll, pub­lished only the pre­vi­ous month and writ­ten up by the same re­porter, found 45 per cent of re­spon­dents wanted all im­mi­gra­tion cut either a lot or a lit­tle.

As Pas­coe noted, the dif­fer­ence be­tween the two re­sults – one per­cent­age point – was within the mar­gin of er­ror. He flayed the “once-ven­er­a­ble Fair­fax mast­heads” for hav­ing “ven­tured a lit­tle fur­ther down the road of click­bait and pop­ulism, join­ing the Mur­doch news­pa­pers with a sec­tar­ian beat-up”.

An­drew Markus, who writes the an­nual Scan­lon Foun­da­tion “Map­ping So­cial Co­he­sion” re­port, poses the ob­vi­ous ques­tion: “What, then, were those peo­ple re­spond­ing to?”

The Scan­lon sur­vey, re­leased last week, asked 77 ques­tions, in­clud­ing a de­tailed se­ries on im­mi­gra­tion, us­ing a very large sam­ple of re­spon­dents and a mix­ture of sur­vey tech­niques – 1500 phone in­ter­views and 2260 on­line sur­veys.

Says Markus: “We found a height­ened neg­a­tiv­ity to­wards im­mi­gra­tion. But it was about per­ceived over­crowd­ing, per­ceived lack of gov­ern­ment plan­ning, per­ceived im­pact on house prices, and not to do with the ideal of im­mi­gra­tion for Aus­tralia, not to do with mul­ti­cul­tur­al­ism, not to do with cul­tural di­ver­sity.”

Cer­tainly, Scan­lon found neg­a­tive at­ti­tudes to­wards Mus­lims – about 25 per cent in the phone polls and 40 per cent in the self-ad­min­is­tered on­line polls. Still, only 18 per cent of phone re­spon­dents and 29 per cent on­line favoured a re­li­giously dis­crim­i­na­tory im­mi­gra­tion pol­icy.

That is, about half the num­ber cited in the Fair­fax beat-up. And the neg­a­tive views were heav­ily con­cen­trated among con­ser­va­tive vot­ers: 26-35 for Coali­tion sup­port­ers and 56-59 One Na­tion.

But that is not the real point here. The point is that polls only mea­sure per­cep­tions, and are only use­ful to the ex­tent that they (a) fairly frame the is­sue, ( b) re­flect in­formed opin­ion, and (c) are ac­cu­rately and soberly re­ported.

Much of the time, the pub­lished polls fail on one or more of these grounds. Of­ten, it is all three.

At best, polling gives a use­ful in­sight into the broad ori­en­ta­tion of the pop­u­lar opin­ion, which is of­ten un­in­formed, even prej­u­diced.

The test for the me­dia, says Richard Di Natale, is whether they sim­ply record that opin­ion or take the harder op­tion of fer­ret­ing out the facts. For politi­cians, he says, the test is whether they seek “to lead and shape the na­tional de­bate, or whether they sit back and be pas­sive ac­tors in agen­das run by other peo­ple”.

“You can either test opin­ion and re­spond to it, or you can shape and lead it. That takes a bit of vi­sion and courage, but that’s what lead­er­ship is.”

Ac­cord­ing to Wayne Swan, you can­not gov­ern ef­fec­tively by ref­er­ence to opin­ion polls. “If you do that,” he says, “then you are pro­ceed­ing by look­ing in the rear-vi­sion mir­ror. So­cial progress is only ever achieved by per­suad­ing peo­ple to move their opin­ions over time. Opin­ion polls can be a guide for ac­tion but should not be the de­ter­min­ing fac­tor. If you re­lied just on opin­ion polls, none of the great re­forms would ever have been made.”

Di Natale and Swan are both right. Poll-driven pop­ulism is un­der­min­ing our trust in pol­i­tics, even in democ­racy it­self. Faith in gov­ern­ment is at record lows.

The irony is, we know this be­cause all the polls mea­sure it, even as they con­trib­ute

• to it.




Prime Min­is­ter Scott Mor­ri­son, Speaker Tony Smith and Op­po­si­tion Leader Bill Shorten pose with the house of rep­re­sen­ta­tives this month.

Newspapers in English

Newspapers from Australia

© PressReader. All rights reserved.