Saturday, September 24, 2016

Field Guide To Opinion Pollsters: 45th Parliament Edition

This is an outdated guide version which has been superceded by the 46th Parliament Edition.

----------------------------------------------------------------------------------------------------------
Just before the 2013 election I posted a Field Guide to Opinion Pollsters, which has become one of the more enduringly accessed pieces on this site.  However, over time parts of its content have become dated or specific to that election, and with more and more pollsters emerging as others disappear, the thing has got too long.  I've decided therefore from now that I will post a new edition shortly into the life of each parliament, editing it through that parliament as the need arises.  Pollsters not expected to be active in the life of the current parliament will be removed, but the old edition text will remain on the previous page.

There are a lot of polls about in Australia these days.  But how do they all work, which ones have runs on the board and which ones can you trust the most? This article describes what is known about each pollster and its strengths and weaknesses and includes coverage of general polling issues.

The gold standard for success for an opinion pollster is seen to be that its polls at election time get the result as close to right as possible.  However, some pollsters are little-tested against actual elections, and getting a specific election right is a combination of skill and luck.  In elections where there is a swing on the last day or two of the campaign, a pollster that is actually not polling correctly may have its errors cancelled out by the swing, and hence record a lucky hit.  There is more to being a good pollster than just getting it right at election time - a good pollster should also provide useful data between elections and do so using well-designed questions that are easy to interpret.  And a pollster should also present their data in a way that makes sense and isn't misleading or confusing.



Some Common Themes

There are some general issues that affect a number of pollsters that I should go through before I move onto individual pollsters.  If you just want to look up a given pollster, scroll down, and then you can scroll back to this bit if you see something you want to look up; it might be here.

House Effect

The issue variously called lean, house effect, skew or bias refers to the tendency of a pollster to produce results that are better for one major party or other (or for specific minor parties) than what is likely to be the real situation.  The term "bias" is a poor one for this issue because it carries connotations of the pollster themselves liking one party more than the other or intending to assist one side, but there is no evidence that this is actually true of any major pollster in Australia.  The extent to which the house effects for each pollster are stable, or change in response to slight methods changes or political circumstances, is often a subject of debate.

Bouncing

The issue often referred to as bouncing, but more technically as overdispersal or underdispersal, refers to how much a poll tends to move about from sample to sample even if voting intention isn't changing very much.  A given poll has a maximum margin of error based on its sample size, meaning that in theory 95% of the poll's results (once adjusted for the pollster's house effect) will be within that margin of error of the true value, but most of them will be much closer to the true value than that.  As the sample size increases, the maximum margin of error decreases, but the decrease isn't proportional.  For instance, for a 50-50 result from a sample size of 1000, the margin of error is +/- 3.1%, but for a sample of 5000 it is about +/- 1.4%, meaning that national polls with sample sizes above a few thousand are usually not worth the effort of producing them.  In practice, some polls tend to vary from sample to sample by much more than would be randomly expected, and these polls are bouncy or overdispersed.  Some polls are very static (except or sometimes even when voting intention actually changes sharply), and these are underdispersed.

In theory underdispersal is nice, because a pollster wants to accurately reflect the national mood rather than releasing polls that are wrong by several points.  No one wants to issue a rogue poll that everyone then ignores.  But a poll that is underdispersed may in some cases be so because it is slow to pick up major shifts when they occur, or indeed doesn't pick them up fully at all.  There is also the problem that there is no way to make a poll under-disperse when using truly random sampling from the entire Australian population, so if a pollster's results are very steady the question must be asked: how are they doing it?  Is it really a pure and random poll, or is the pollster allowing data from other pollsters to influence the way they fine-tune assumptions that create the final outcome?  (The latter practice is known as herding.)  Other possibilities include that underdispersed pollsters are using tracking from their own poll or other modelling assumptions to chop rough edges off their results, or surveying the same respondents too often.

Mobile and Landline Phone Polls vs Online Polling

No major Australian pollster only polls landlines.

In the lead-up to the 2013 federal election it was widely argued that the rising proportion of mobile-phone-only households (which contain mostly young voters) meant that landline-only polling skewed to the Coalition.  Yet at that election there was no such skew, and not much difference in performance between landline-only phone polling and polls that called mobiles. The most accurate final poll at that election polled landlines only.  Partly this was because unrepresentativeness in landline-only polling can be overcome by scaling (see below) and partly this was because the political attributes of landline and non-landline households seem to not be as different as might be expected.  See Christian Kerr's report of Newspoll surveying.

The 2013 election, at least, supported the view that purely online-panel pollsters have bigger problems to contend with than landline-only pollsters (and again the Newspoll study above is relevant).  Online panel polling, whatever its recruitment method, may have biases that cannot be removed, because online respondents are people who like filling out surveys (often in return for rewards) and are comfortable with technology.  Not everyone is like that, and it is a difficult thing to predict by demographic attributes alone, and one that may skew political opinion.

In the 2013-6 parliament, landline-only polling entirely disappeared from the federal scene.  All major pollsters now either call mobiles as well, or have included some other kind of surveying (such as online panel surveying) as part of their sampling mix.

Scaling

Getting a truly random sample of the Australian national population is difficult.  Some types of voters are simply much easier to contact than others.  One option is to keep contacting potential respondents until you get exactly the right demographic mix.  However this can introduce time delays and increase the costs of polling if you are using phone polling.  Another option is to "scale" the responses you have by applying corrections based on which group you have less of in your poll than others.  For instance, suppose that voters in age group A are 10% of the voting population but only 5% of your sample, while voters in age group B are 25% of the voting population but 30% of your sample.  A simple scaling method would then be to double the value of each response from group A and multiply the value of each response from group B by 25/30.   In practice, scaling is much more complicated and a given response might be scaled on many different criteria at once, some of which might increase its weighting and others of which might decrease it.

Scaling effectively increases the margin of error and bounciness of a poll, because any sampling error in a small group that is scaled up could be magnified in the overall total.  There is also a risk that if a demographic group is hard to poll, then the voters who can be polled within that group might not be a fair sample, and that any error caused by that might then be magnified.  For instance, young voters are hard to reach using landline polling, excepting those living with their parents.  But are young voters who live with their parents representative of all young voters?

Some areas of Australia are simply very difficult to poll accurately by any method.  The Northern Territory is one of them.  Inner city electorates are also hard to poll because of high rates of enrolment churn and non-enrolment in the electorate.

Internal and external

Many prominent pollsters conduct both "public polls" and "commissioned polls".  A public poll is a poll either conducted by the pollster themselves without anyone paying for it, or commissioned by a major media source, for which full details of results are usually publicly released.  Although there is the potential in theory for the party biases of media sources towards a party to result in them hiring a pollster to present results in a good light for that party, there is really no evidence that this happens in Australia.

Commissioned or internal polls are polls paid for by a political party or by a group with an interest in an issue (such as a lobby group or company).  Commissioned polls usually ask standard voting intention questions, but it is the choice of the client whether to release results, and it is common for internal polling to be only selectively released (an increasing problem with robo-polling reducing polling costs).  Often the full details of commissioned polls are not released.

Some companies produce excellent public polling while also accepting commissioned polls in which the questions are asked in a way more likely to get the result the client wants.  Often the client wants a poll that shows strong support for their cause so that they can then get more publicity for their cause and attempt to convince politicians that it is popular.

Just because a pollster does good public polling does not mean their commissioned polls should always be trusted.  As a general rule no commissioned poll reported in media should be taken all that seriously, whatever the pollster, without the full release of the verbatim wording of all the questions in the order asked, and an extensive breakdown of results.  Even with these things, the wording of the questions often turns out to be suspect. Even if there is nothing wrong with the commissioned poll at all, there is still the possibility of selective release of good polls while not releasing the bad ones.  Furthermore, the accuracy of internal polling is prone to morale bias: some parties could be more likely to hire companies that tend to tell them what they want to hear, even when it actually isn't true.

Upfront Exclusion

This term refers to the proportion of voters who are eliminated from results because they either cannot specify a preference, refuse to answer the question, or fail to complete the survey interview.  For most pollsters this proportion is slight to moderate (a few to sometimes 10%).  In theory if undecided voters had a tendency to shift to a particular party, this could make polls very inaccurate, but there is not much evidence that this issue has bitten in recent elections.  Generally, the higher the upfront exclusion rate, the more chance that those voters who do reply are not representative, but this seems to become a serious problem only with polls that upfront-exclude over 10%.

The Green Vote

Some pollsters have a recent track record of usually or always overestimating the Green vote compared to actual election results, especially when the Green vote is fairly high.  An especially stark example was the 2014 Victorian state election, in which all 17 polls published in the two months before the election had the party's vote too high, by up to eight points.  Part of the reason for this is that the Green vote is actually very soft; there may be other reasons.  Small and new pollsters, and pollsters with high undecided rates, are especially prone to this problem.   Polling of "others" and "independents" is often also inaccurate.  Smaller parties tend to be under-polled if they are not specifically named, while the category "independents" tends to over-perform in polling compared to election results.  Voters may offer "independent" as an ambit wish for a good high-profile independent candidate, but they won't vote for one if one isn't on the ballot.

Preferred Prime Minister

Preferred/Better Prime Minister or Premier polling questions are a bugbear of Australian poll commentary, which would probably be more informed if such questions did not exist.

Given that Australian politics is so presidential and that the personal approval/disapproval ratings of the Prime Minister are a driving indicator of change in the 2PP vote, it might be expected that a question about who should be Prime Minister would yield good information.  It frequently doesn't.  For whatever reason (and it seems to have something to do with the don't-know option), the preferred leader scores of most major pollsters flatter the incumbent.  For instance, in Newspoll, if the two parties are tied on the 2PP vote, and their leaders are tied on personal ratings, then the Prime Minister will typically lead the Opposition Leader by 16 points as Preferred Prime Minister.  This skewing leads to the media talking about fairly big PPM leads for incumbent PMs as evidence of dominance when they are not, or small PPM leads or deficits as evidence that the government still has an ace up its sleeve when in fact they are evidence of trouble.  See Why Preferred Prime Minister/Premier Scores are Rubbish.

The only pollsters that seem to avoid this are ReachTEL (see below) and Morgan SMS.

2PP Preferencing: Last Election vs Respondent Allocated

Most pollsters publish two-party preferred results that are based on the assumption that voters who do not vote for the major parties will distribute their preferences in the same way as at the last election.  Many pollsters who do this try to calculate the preference flow for the Greens (and often a few other named parties) separately from other parties, but some (eg Ipsos) use "overall preference flow" which assumes that the average flow from all non-major-party voters will stay the same (even if the proportion of them who vote for the Greens changes.)

Some pollsters, however, use respondent-allocated preferences, ie they ask the respondent how they will distribute their preferences.  One problem with this is that many voters will actually follow what their party's how-to-vote card says rather than decide for themselves. In any case this method has a history of strong Labor skew at federal elections and is generally less accurate.  At some state elections this method has shown a Coalition skew, at least among some pollsters.

The 2016 federal election reinforced the superiority of last-election preferences, following some recent cases (2013 federal, 2014 Victorian) where the truth was somewhere between the two.  In the 2015 Queensland election, last-election preferences proved very inaccurate and it's likely respondent-allocated preferences would have been more predictive for that election, and will be so for some other such elections with very large swings.  In the 2015 NSW state election the most conservative estimates of respondent-allocated preferences were accurate.  It seems that voter choice about preferencing makes more difference in optional-preferential voting (which now exists only in NSW and the NT) than compulsory, because voters can choose to exhaust their vote.

For a detailed technical discussion at federal level see Wonk Central: The Track Record Of Last-Election Preferences.

Single Polls vs Aggregates

No matter how good a pollster is, no single poll series will consistently give a perfect and reliable picture of voting intention.  Aggregating polling from multiple polls to get a picture of the whole range of voting intention is usually more reliable than assuming any one poll or poll series is accurate.  If you just have one poll saying 52-48, you do not know for sure the leading party is in front.  If you have five with an average of 51.5-48.5, all taken at the same time and without significant house effects, you have a much better idea that the leading party really is in front.

Many people make the mistake of saying that if all the polls are within their margin of error of 50-50 then the race is as good as a tie.  Generally, this isn't true.

Different poll aggregates will give slightly different values at any given time because of the nature of the different assumptions made by those running them.  Such issues as what weight to place on given polls based on their past record, how quickly to assume a poll has ceased to be relevant and what the hell to do about Essential are not easy and different modellers will try different assumptions, and then modify them when elections provide more data.

A list of active polling aggregators is given at the base of this article.

Poll Fraud and Fake Polls

Poll fraud occurs when someone (possibly a pollster) simply makes up numbers, which means it can produce a "poll" without needing to spend time or money surveying anyone.  Poll fraud can be detected by various methods, including results that fail non-randomness tests in their fine detail.  Poll fraud is a problem at times in the USA.  No poll fraud in commercial polling has been detected in Australia to my knowledge, but fake polls are now and then circulated.

Faked private polls seen in Australia have included a faked "ReachTEL" for the seat of Toowoomba North at the 2015 Queensland state election, a faked "Nationals internal" for the seat of New England in October 2017, a faked Labor internal for the 2018 Batman by-election, and a faked "ReachTEL" for the seat of Curtin at the 2019 election.

A satire website called The Bug Online, associated with the @drongojourno and @thebugonline twitter feeds, routinely produces fake Newspolls on what it believes, not always correctly, to be Newspoll nights.  Gullible Twitter users are sometimes fooled into believing these are real Newspolls.  Real Newspoll results do not appear via the Australian until 9:30 pm on Sunday nights, though sometimes results or hints are tweeted or broadcast before then by reliable or mostly reliable journalists. 

The pollsters

Newspoll
Method: Mixed online polling and landline robopoll
Preferences: Based on undisclosed mix of previous elections (including federal and state elections) and sometimes own judgement

Newspoll, house pollster for The Australian, is Australia's best-known polling brand and the one that most influences political debate, election betting market moves, and public comment about party standing.   

Between election campaigns it normally polls fortnightly, but sometimes the schedule is adjusted to respond to current events, to coincide with a new parliamentary week, or to avoid long weekends.  Also the contracted schedule is actually not quite fortnightly, so sometimes there is a three-week break for no obvious reason.   For more on Newspoll release gaps see How Often Are Federal Newspolls Released?  Currently the poll is typically released online on Sunday nights with in-print reporting on Monday, though in the past a Monday/Tuesday schedule was followed during many non-parliament weeks.

Until July 2015, Newspoll was a telephone pollster that dialled randomly selected numbers and only called landlines.  In July 2015 the brand was transferred away from the company previously running it (which was dissolved, with some key staff moving to start Omnipoll).  Now, Newspoll is operated by Galaxy (see below) and is a hybrid pollster using a combination of online panel polling (a la Essential) and robopolling (a la ReachTEL).  The robopolling is of landlines only (I think), but the online polling will reach respondents who do not have landlines.

The Newspoll brand has a long history, going back to late 1985, and has asked questions in a very stable form, making it an excellent poll for past data comparisons; these seem to have been not much affected by the 2015 methods change.  The brand has a predictive record at state and federal elections that is second to none, despite a fairly bad final 2PP figure in 2004 (as a result of a shortlived and incorrect 2PP estimate method).  The new Newspoll has performed very well at its early electoral tests, including stunningly accurate final polls in the 2016 federal elections, very accurate primary vote polling for the 2017 Queensland election, and the most accurate final polling of the difficult same-sex marriage postal survey.

Far too much attention is still paid to poll-to-poll moves in Newspoll without considering the pattern from other polls.  One behavioural change following the switch to Galaxy is that Newspoll has become much less bouncy.

From December 2017 shifts in the behaviour of Newspoll's 2PP relative to its primary votes became strongly apparent, as its 2PPs started to look a lot less like 2016 election preferences.  Galaxy started employing a different formula for One Nation preferences following the Queensland state election (with One Nation preferences splitting around 60-40 to Coalition), but this was not documented until it was suspected and then exposed by psephologists over a period of months.  In April 2019 Galaxy started applying an arbitrary 60-40 flow from United Australia as well

An often-discussed aspect of the old Newspoll was its upfront exclusion rates and I wrote a detailed article about that here.  Newspoll also attracts a massive volume of online conspiracy theories, most of them left-wing and virtually all of them groundless and daft.  Reading a full #Newspoll Twitter feed on a given poll night may cause permanent brain damage, and at least 95% of tweets that mention "Newspoll" and "Murdoch" together are rubbish.

YouGov Galaxy
Method: Mixed online polling and phone polling (including mobiles) for federal polls, robopolls for seat polls
Preferences:Based on undisclosed mix of previous elections (including federal and state elections)

Galaxy Research has been conducting federal polling since the 2004 federal election.  Galaxy's federal polling was formerly conducted mainly by random telephone surveying but its polls now use a mix of phone polling (I believe including mobile phones) and online panel polling.  Galaxy appears sporadically between elections and is the house pollster for a string of News Limited tabloids.  It is polling less frequently in its own name following its large deal to run the Newspoll brand.

Galaxy has a formidable predictive record and is an uncannily steady (underdispersed) poll.  Earlier in its career it appeared to produce slightly Coalition-leaning results between elections, but the lean would go away during the campaign.  There is a sharp contrast with Galaxy's specific issue/attribute questions, which (presumably at the behest of sponsoring media) frequently use murky and provocatively subjective language and are often difficult to make accurate sense of.  This has now spilled over to the Newspoll brand.

Galaxy sometimes uses other polling methods.  For instance it has been using automated phone polling (robopolling) in seat polls. At the 2016 election, these polls were notably underdispersed - they were not only much less variable than the actual results, but less variable than they would have been expected to be even if there was no difference between seats.

Galaxy was in my view the best pollster of the 2013 federal election campaign and lead-up, and the Galaxy/Newspoll stable shared this honour again for 2016.

In late 2017 Galaxy was acquired by YouGov, but this is not expected to affect its polling methods.

ReachTEL
Method: Robopolling (landlines and mobiles)
Preferences: Respondent

See at the bottom of this entry for uComms ReachTEL.

ReachTEL is the most commonly encountered "robopoll" and is now often used by Sky, Channel Seven and various newspapers.  It is by far the most commonly commissioned poll.  A robopoll conducts automatic phone calls to randomly selected landline and mobile phones, and respondents pick answers according to options stated by a recorded voice. Robopolls are cheap to run and can be conducted very quickly, but have the disadvantage that more voters will hang up on them immediately.  Therefore they require a lot of scaling, which in theory increases the chance of errors.  

ReachTEL soon established itself as a reliable and accurate national and state-level pollster, being among the top few pollsters in a string of elections, including being the best statewide pollster at the 2015 NSW state election and the most accurate pollster of primary votes in Queensland in 2015.  However its tracking performance at state elections in 2017 was poor, showing the WA election as unduly close until its final poll and generally showing the LNP as winning the Queensland 2PP until very close to the end of the campaign.  ReachTEL bounced back at the Bennelong federal by-election where both its polls were much more accurate than all three of Newspoll/Galaxy's.  It was also the most accurate poll at the Tasmanian state election (where Newspoll/Galaxy did not take the field.)

ReachTEL's electorate-level public federal polling has sometimes been skewed to the Coalition.  It has also struggled (as have all pollsters, but perhaps more than some) in inner-city state electorates with high Green votes.

ReachTEL forces answers to some questions, often disallowing an undecided option and requiring the respondent to choose one option or the other.  This results in preferred Prime Minister figures that are often closer to the national two-party vote than those of other pollsters.  It also can produce worse ratings for the government on issues questions.  The suggestion is that there are many people who have slightly negative views of the government but will let it off with a neutral rating unless forced.  Forcing can cause voters to hang up but the company advises me that the percentage of hangups after the first question is very small.

ReachTEL leadership performance ratings use a middle option of "satisfactory" which seems to capture some mildly positive sentiment.  For this reason ReachTEL ratings when expressed in the form Good vs Poor seem harsher than those of other pollsters.  Lately the gap between ReachTEL and other ratings in this regard seems to be closing.

In 2016 ReachTEL switched to using respondent-allocated preferences for most of its polls.

A common annoyance with ReachTEL polls is that figures are published with initially "undecided" voters who give a response at the second attempt not redistributed, making the primary votes for all parties look too low. Other pollsters include these voters in the figures for their second-attempt response.  Especially when media sources don't publish a full version of the poll, this often makes finding out exactly what is going on with ReachTEL's primary votes very difficult.  Also, ReachTEL's use of the term "undecided" excludes "hard undecided" voters who have no idea what party they would choose - other pollsters call these the "undecided" voters (and most pollsters usually exclude them.)

In 2017 ReachTEL was found to be incorrectly distributing respondent-allocated preferences of Nationals voters in its federal polls, producing a skew to Labor in its respondent preferencing methods as most actual Nationals votes are never distributed.  This may have been cancelled out by a tendency of ReachTEL respondents to be otherwise overly generous to the Coalition in their preferencing (as seen in the Queensland election leadup).

From at least March 2017 reports emerged that ReachTEL polls were including a charity section in which the respondent could not complete the call without pressing a number, following which their details would be passed to a charity.  What happens to the poll data if the respondent hangs up at this point is unknown.

Cases of ReachTEL calling voters who do not live in the electorate being surveyed are often reported.  It appears this is a product of a small error rate but a very large number of total calls.

uComms ReachTEL
Method/Preferences: as for ReachTEL

uComms is a distinct company that uses ReachTEL technology to conduct polling, mostly seat polling.  It is often difficult to find out whether ReachTEL polls are polls by ReachTEL or polls by uComms using ReachTEL.  uComms has been little tested at elections so far but its Victorian statewide polling was no worse than others (not that anyone was all that close) while its Wentworth polling was no better.  In April 2019, an ABC report disclosed that two of the three shareholders of uComms are major unionists, resulting in the Nine papers (Age/SMH) announcing they would no longer use uComms, and several other clients expected to follow suit.

Ipsos
Method: Live phone polling (mixed landline and mobile)
Preferences: Last-election preferences (batched "overall preference flow") and respondent preferences are both released

Ipsos is a global polling brand with a good reputation.  Fairfax Ipsos started in late 2014 and is a live phone poll that samples both landlines and mobiles and operates on a similar scale and frequency to the former Fairfax Nielsen polls.   The poll's biggest issue is that it persistently has the Green primary vote much too high and the ALP primary too low; it is also prone to have Others on the high side.  It is also somewhat bouncier than other national polls, largely because of its smaller sample sizes (recently reduced from 1400 to 1200 as of late 2018).  At elections so far it tends to have performed well on the 2PP vote but not so well on the primaries. Prior to the start of the Fairfax polling, Ipsos conducted some other polls under the name Ipsos i-view.

Ipsos' leader ratings polling is noticeably more lenient than Newspoll's.  This especially applied in the case of former Prime Minister Turnbull.  Ipsos can have long gaps between releases, which does not stop Fairfax Nine journalists making a big deal of meaningless changes since the previous Ipsos in their coverage.

Ipsos uses a slightly different last-election preference method to other pollsters.  A single preference flow for all minor parties is found and the same flow is applied irrespective of changes in the (alleged) support for particular minor parties.  One media report suggests this method might have changed recently but this isn't verified yet.

Essential Report
Method: Online poll
Preferences: Last-election

Essential Report is a fortnightly (formerly weekly) online poll that was, for a long time, the house pollster for the Crikey website's subscriber newsletter.  It is now mainly published by the Guardian.  Essential's respondents are selected from a panel of around 100,000 voters, and about 1000 are polled each week, by sending out an email offering small shopping credit rewards for participation.  Until the end of 2017, Essential unusually used two-week rolling samples to counteract "bouncing" from poll to poll, but now each sample is an independent sample.

In its very early days Essential was a very bouncy and Labor-skewed poll, but it made changes at some stage in 2010 and delivered a good result at that year's election.  However, in the 2010-3 and 2013-6 federal terms the poll still had some problems.  It was too underdispersed (see Essential: Not Bouncy Enough), but in a way that seemed to cause it to become "stuck" and to respond slowly and incompletely to big changes in voting intention, as compared to other pollsters.  Quite why this was is not entirely clear - it could be to do with the composition of the panel or with repeat sampling issues within it (against which some precautions are taken).  Essential also sometimes displayed a very different trend pattern to other pollsters.  Its performance in the 2013 election leadup was idiosyncratic.  At the 2016 election it produced an impressive final-week poll but doubts remained about its tracking behaviour.  Since that election, my impression is that its poll-to-poll variability has become more natural.

Essential asks a very wide range of useful attribute and issue based questions that often help to drill down into the reasons why voters have specific attitudes, that in turn underlie their votes.  These are sometimes marred by high don't-know rates, which are an inescapable problem with online polling formats.  It is also unclear how representative Essential's panel is on issue questions - see the debate involving Essential and the Scanlon Foundation here.

YouGov / Fifty Acres
Method: Online polling
Preferences: Respondent, via a simulated ballot

Not to be confused with YouGov Galaxy.

In May 2017, Poll Bludger reported that YouGov, a major UK-based pollster, would be going into the field very shortly with a fortnightly online panel poll with a sample size of 1000.  YouGov has a reasonably good reputation for accuracy overseas but did not escape the 2015 UK election Poll Fail, and also had a 51% Remain vote in its final Brexit poll and Hillary Clinton with a four-point nationwide lead over Donald Trump (she actually won the popular vote by two).  Its 2017 UK election modelling was stunningly good though it also produced a separate poll that was no better than most.

YouGov's panel poll differs from Essential's in that it openly calls for panellists - in contrast there is no known way to deliberately get on Essential's panel.  Early YouGov results have had very strange respondent-allocated 2PP results that have been very favourable to the Coalition.  This may result partly from an unusual respondent preferencing method in which respondents fill out an imitation ballot paper (it is easy to envisage there being design issues with this method).  Another oddity is the inclusion of a "Christian parties" option, the vote for which is overstated and probably includes some hidden Coalition vote.

Since YouGov's acquisition of Galaxy, this polling series has ceased.

Morgan
Method: Varies, currently active versions include face-to-face, live phone poll and SMS poll
Preferences: Respondent

Roy Morgan Research is a polling house that traces its lineage back to Morgan Gallup polls conducted from the early 1940s.  The very experienced pollster was formerly the house pollster for The Bulletin magazine (which no longer exists), and suffered badly when it predicted a Labor win in 2001.  Now unattached to any specific media, Morgan is not as much discussed as other pollsters, but the lack of media attachment is not the only reason for that.  Morgan's polling is confusing and unreliable, often not sufficiently documented, and its reputation among poll-watchers has declined in recent years.

Various forms of Morgan polls are seen including the following:

* SMS only mobile phone polling (mainly used for state polls, also for some issues polling)
* Telephone polls (mainly used for leadership polling)
* Multi-mode polls (most recently a mixture of face-to-face surveying and SMS polling)

Other combinations of multi-mode polling have been seen in the past, and at one stage Morgan used to issue a lot of pure face-to-face polls, which skewed heavily to Labor.  Morgan also usually uses respondent-allocated preferencing, which can also create skew to the ALP.  The pollster has recently displayed severe skew to the Greens in its primary votes, and some of its local panels may be unrepresentative.  The small sample size of its state polls of the smaller states is another problem - Tasmanian samples are sometimes reported in the media, but with a sample size of around 300, why bother?

Morgan's multi-mode polls that include a face-to-face component have often skewed to Labor, but skewed to the Coalition for a while after Malcolm Turnbull first became Prime Minister.

Morgan polls seem to be very reactive to "news cycle" events.  SMS sampling (apparently drawn from a panel rather than random selection from the whole population) is probably too prone to "motivated response", with responses from voters who have strong views about the issues of the day being overrepresented in the results. My view is that SMS is a suspect polling method.

In the leadup to the 2016 election, Morgan issued many seat-by-seat results, frequently based on tiny sample sizes and often accompanied by unsound interpretation.  The pollster also stopped releasing national polling in the last month of the campaign, making it impossible to benchmark its performance for the future, and its future polling intentions are unclear.  Finally, in recent state elections Morgan's SMS polling (which appears to have now ceased) has been absurdly volatile.

As a general rule, Morgan polls should be treated with a lot of caution at all times.

Lonergan Research

Lonergan is another robopollster that has fairly recently moved into public polling (and has also done a few internal polls for the Greens and other left-leaning entities).

Lonergan had a poor 2013 campaign with its seat polls showing a massive Coalition skew and a commissioned mobile-phone-only poll proving very inaccurate (perhaps because its sample size was too small).  Its final 2016 federal poll, however, was quite accurate despite being taken nearly two months before the election.  Its NSW 2014 state election polls showed skew to the Coalition and results from its commissioned seat polls have been mixed.  A Lonergan landline-only seat poll of the Batman by-election, however, scored a bullseye, but it may have been lucky as it was taken before public attacks on the Greens candidates by disgruntled Greens members.

Lonergan initially attracted criticism for scaling results to voter reports of how they voted at previous elections,. Some voters may not report their voting behaviour at previous elections accurately, and may over-report voting for the winner, as a result of which polling becomes skewed towards the other side.  I am not aware of the poll still employing this method.

JWS Research

JWS Research is another robopollster.  It conducted a massive marginal-seat poll at the 2010 federal election with indifferent predictive results on a seat basis (but an excellent overall 2PP estimate) and a similar exercise at the 2010 Victorian state election with excellent results.  In the 2010-3 term it was notable for a string of aggregated marginal seat mega-polls, including some jointly commissioned by AFR and a Liberal Party linked strategic advice/lobbying firm called ECG Advisory Solutions.  These polls were often blighted by the release of clearly unsound seat projections based on them, but that is not the fault of the data.  JWS also conducted many local-level seat polls at the 2013 campaign.  Electorate-level polls released by JWS during the 2013 campaign showed a strong general lean to the Coalition.  It is likely that the series of aggregated marginal polls experienced the same issue.

In the 2013-6 cycle JWS kept a lower profile, but it releases very thorough and useful issues polling every four months in an omnibus called True Issues.

MediaReach

MediaReach is another IRV pollster (robopollster) that is reported as being owned by a firm with several years' experience in the field.  It has done, for example, state polling of WA and the NT and an electorate poll of Mackellar.  At both the Northern Territory election and the election for the seat of Solomon, MediaReach overestimated the large swing against the CLP by about five points.

MediaReach was heavily used in the 2018 Tasmanian election campaign by the Liberal Party with numerous results released to the media.  These polls performed very well, effectively tracking the collapse of the Jacqui Lambie Network vote and that the Liberals were on course for a comfortable victory.  There may have been a self-fulfilling prophecy aspect in their success.

Dynata (formerly Research Now)

Dynata is an online panel pollster similar to Essential.  It has produced a fair amount of mostly commissioned issues polling, with relatively little testable polling of vote shares.  In 2016 it conducted Senate polling for the Australia Institute which overestimated the vote for every party named (especially the Coalition and Greens) and hugely underestimated the vote for unnamed parties. (Senate polling by any pollster is likely to encounter this problem to some degree.)

Community Engagement

Community Engagement produced one national commissioned poll and some commissioned seat polls at the 2016 election.  Documentation on its early polls was so inadequate that it is not even clear what kind of pollster it was.  Early results were not accurate.  The pollster has been active in commissioned polling since, including for the ALP in Western Australia in 2018, where it was described as a robo-poll.  The report in the West Australian also showed the pollster copying ReachTEL's unfortunate habit of including "undecided" in a headline figure, though whether it prodded respondents and discarded the unproddables, as ReachTEL does, is unknown.

WAOP

Western Australian Opinion Polls (WAOP) is a very long-active but relatively little-known WA pollster. It has polled commissioned polls on a range of issues, usually with modest sample sizes, and recently released commissioned voting intention polling for the federal seat of Pearce.  Its use of decimal results means its results may be mistaken for ReachTEL's.  There is very little information online regarding WAOP's methods and no available basis yet for testing its accuracy.  In 2012 one of its polls was described as a "telephone poll".  A 2015 poll used mainly "voice broadcast" polling (ie robopolling) with a small live telephone sample.

EMRS

EMRS is a Tasmanian pollster that has surveyed state and federal voting intention in Tasmania since the late 1990s, and sometimes does commissioned voting polls inside and outside the state.  It is best known for quarterly state voting intention polling.  Currently its state polls use a mix of landline and mobile polling while some of its other polling uses a mixture of online and phone polling.  I believe that even its state phone polling is at least partly a panel poll and not truly random.

EMRS can be a difficult pollster to make sense of because its undecided rates are often much higher than for other pollsters, and this often applies even after the initial prodding of unsure voters to say which party they are leaning to.  (Antony Green's partial defence of the company's high historic undecided rates here was refuted here).  At past state elections the pollster has tended to overestimate the Green vote and underestimate the incumbent government of the time's vote by, on average, a few points each time.

A commissioned EMRS poll of the 2018 Hobart City Council mayoral race was very accurate, despite the race being a voluntary vote and much harder to poll than a state or federal seat poll.

Voter Choice

Voter Choice is a scaled opt-in panel project (like a smaller scale Vote Compass in that way) that produces some polling and also some comments based on qualitative research.   Its polling is untested at elections other than Wentworth (at which it was very inaccurate, especially concerning the Liberal-Labor 2PP result), although it released predictions for the Super Saturday elections that had some basis in very small sample unpublished polling exercises.

Voter Choice has openly documented that it adjusted results of its Wentworth polling to introduce new weightings because the numbers "looked wrong" (in what way and why isn't stated) and that after the new weightings were added the pollster "liked" the results.  This is apparently because of qualitative data but why is not made clear.  The Wentworth poll used a simulated ballot of major contenders to distribute preferences (an advanced form of respondent preferencing.)

Voter Choice claims and reports should be treated with enormous caution as the founder of Voter Choice sometimes makes very strange claims online, including a tweet that said Labor "could not win" the 2019 election if its 2PP vote was 52.6%.

Metapoll

Metapoll is an online pollster sometimes published in the Guardian in the leadup to the 2016 election.  It is also the author of a deluxe polling aggregate that initially included its own unpublished data, though this was later removed except in the area of preferencing.  The predictive performance of Metapoll's aggregate did not live up to the hype and the poll hasn't resurfaced since the election.

AMR Australia
See previous edition.

Nielsen (No longer active)
See previous edition.

Others

Others will be added here as I come across them or on request.

Online or TV News "Polls": They're Useless!

Ah, but what about those polls on newspaper websites or Yahoo that give you the option of voting for a stance on some hot-button issue?  What about those TV news polls that ask you to call a number for yes or a different number for no?

The short answer is that these are not real polls.  They are unscaled opt-ins and they are not scientifically valid as evidence of voter intentions.  For the first thing, as regularly noted in the fine print, they only reflect the views of those who choose to participate.  If a media source tends to be read more by right-wing voters, then its opt-in polls will tend to be voted on more by right-wing voters.

Secondly, opt-ins suffer from "motivated response".  People who care deeply about an issue will vote on them, but people who really don't have a strong view (but might answer a question put in a real poll that they've agreed to spend time on) will probably not bother.

Thirdly opt-ins are prone to co-ordinated stacking.  Activist networks will send messages by email or social media telling people there is a media poll they can vote in, and this will often lead to votes being cast from way outside the area to which the poll relates.  Opt-ins are easily flooded by this method, producing very skewed results.

Finally, opt-ins are often prone to deliberate multiple voting by single voters, either by people with strong views on an issue who want to manipulate the outcome or by people who want to ruin them just because the results are taken far too seriously.  There are ways to try to stop it, but some of them work better than others. (See in this regard the brilliant work of Ubermotive and also see the guide to how to stop it here.)

It is especially unfortunate that the ABC's Lateline employs "polls" of this kind.  They should know better.

I hope this guide is useful; feedback is very welcome.

Poll Quality Reviews

The following pieces on this site have compared the performance of different polls at a specific election:

Victoria 2018: Final Lower House Results, Poll Performance and 2PP Pendulum
2017 Queensland: Final Results and Polling Accuracy
How Accurate Was The Same Sex Marriage Polling?
2017 WA Washup
2016 Federal Election: Best And Worst Pollsters
New South Wales 2015
Queensland 2015
Victoria 2014
2013 Federal Election: Best And Worst Pollsters 

Polling aggregators

* My own, in the sidebar of this site (methods post here).  This is a relatively quick model, aggregating 2PP results using published 2PPs and primaries, and designed for fast updating as new polls come out.  It includes adjustments for accuracy and house effect.

* Bludgertrack.  This is the best known aggregator.  It incorporates state-level polling data to predict seat tallies and recorded an extremely accurate seat and 2PP projection at the 2013 federal election.  It derives its 2PP figures from adjusted primary figures rather than aggregating released 2PPs.

* Mark The Ballot.  A Bayesian aggregator of primary vote shares and 2PP.

Andrew Catsaras formerly did the Poll of Polls segment on ABC's Insiders and now and then posts his aggregate, which provides a monthly rounded 2PP figure and now primary estimates.

Several other aggregators operated during the 2016 election cycle and links and comments on them will be added if they resurface.  New aggregators may also be added.

No comments:

Post a Comment

The comment system is unreliable. If you cannot submit comments you can email me a comment (via email link in profile) - email must be entitled: Comment for publication, followed by the name of the article you wish to comment on. Comments are accepted in full or not at all. If you submit a comment which is not accepted within a few days you can also email me and I will check if it has been received.