Tuesday, December 29, 2015

Polling And Penalty Rates

(Note: this piece now has a follow-up.  See What Scientists Do)

Penalty rates have been on the political radar lately. A poll on the subject released by The Australia Institute on Sunday has attracted a fair amount of interest.  Many Coalition MPs support cuts to current penalty rates (which are required extra loadings on pay for certain occupations for weekend, evening or public holiday work) and the Labor Opposition is currently campaigning against such cuts.  This will probably be a significant philosophical divide between the parties at the 2016 election.

If we are to believe the poll's sponsor and reporting of the poll by the SMH yesterday, the government will face a massive backlash, including from its own voters, if Sunday penalty rates in the retail sector are reduced as recommended by the Productivity Commission.  The reality is that the views of Coalition supporters on the proposed change are rather less clear.

The TAI Poll

The poll, a robopoll conducted by ReachTEL on December 17, polled residents of Warringah (Tony Abbott), New England (Barnaby Joyce (Nat)), Dickson (Peter Dutton) and Page (Kevin Hogan (Nat)).  The poll results can be downloaded from the TAI website.

Leaving aside the views of non-Coalition parties for a moment, what I want to focus on is the views of Coalition voters.  In New England, 9.1% of Coalition voters said penalty rates should increase, 47.8% said stay the same, 21.2% said be reduced, 22% said abolish them.  In Dickson, it was 10.3%, 56.2%, 16.8%, 16.8%.  In Warringah, 10.8%, 43.6%, 19.3%, 26.3%.  In Page, 9.3, 46.7, 19.9, 24.1.  So in each of the electorates, more Coalition supporters said penalty rates should increase or stay the same than said they should be reduced or abolished.

It is always useful when looking at these polls to see whether they add up to 100% or not.  Adding up to much less than 100 can be a sign of an unpublished inclusion of undecided voters.  Adding up to 100, as in this case, can be a sign that an undecided option was redistributed or disallowed.

In this case, it turns out it was a forced-choice question.  Poll respondents had to give one of the four options or else hang up and be excluded from the survey, thus losing the response they had already given to other questions.  At this stage other results have been released for Warringah only, and the full list of questions in the order asked has not been released for any of these polls, but voting intention is asked first. While TAI is better at releasing poll data than many other lobby groups, ideally any poll that is reported should have the full ordering of questions and all results released.

This is the question as asked:

"The law currently requires some workers in the retail industry to receive a higher rate of pay, or penalty rate, on Sundays. Do you think penalty rates for Sundays should be increased, stay the same, be lowered or abolished?"

So, suppose someone really couldn't care less about this question but realises they have to pick one of these options to continue in the poll.  They might disconnect, but more likely they'll go for whatever sounds most like the neutral option.  "Yeah, um, I dunno, um, stay the same I guess" (presses button).  Suddenly their key press shows up as evidence of a blowback against the Coalition.  It's not; this neutral and possibly apathetic voter might well go either way when they hear more evidence about the issue in the debate.  On that basis, I wouldn't be at all confident that the appearance of Coalition voter opposition to change is real, given that the Coalition-voter margin of (increase or stay same) over (reduce or abolish) is not that large in three of the four electorates.  (My initial claim on Twitter that "most" stay-the-sames would be don't-knows seems unlikely to be true after looking at other polling on the issue, but enough would be don't-knows to make counting "stay the same" as opposition to change extremely dubious.)

The poll is also of limited practical use because it is likely a reduction in penalty rates would be sweetened by an across-the-board increase in minimum pay rates; otherwise, the poorest workers are being made worse off.  However, this option isn't canvassed.

A forced-choice issues question design is quite unusual.  Normally, robopolls force answers only on voting intention and basic political questions like preferred Prime Minister.  Anything else, at the least, incurs some risk of increasing the hang-up rate.  However, as concerns commissioned polls, it is extremely common to see design assumptions made that could be avoided but that happen to be beneficial to the case the sponsor wants the poll to make.

All commissioned polls should be treated with extreme caution and no media source should report the result of any commissioned poll without obtaining expert advice on its reliability.  However, journalists are indebted to those who give them easy stories in the form of polling data, and hence tend to go soft on its possible defects.

Other Polling On The Issue

A poll by Essential Research a few months back presents a contrast.  This poll asked:

"Do you approve or disapprove of the Productivity Commission recommendation to cut Sunday penalty rates to the same level as Saturday rates for workers in hospitality, entertainment and retail?"

Voters overall said thumbs-down to the proposal (32-54) but Coalition supporters, mildly, backed it (51-40).  I think the question design in Essential's poll is a little bit shaky too, since the information that the change is a "Productivity Commission recommendation" might skew the answers in favour of the proposed cut.  However, this at least provides some reason to suspect that some of those choosing "stay the same" in the TAI poll are actually indifferent.  To have as many as 32% of all respondents nationally endorse the cut (a figure barely matched in strongly Coalition seats in the ReachTEL poll) is another indication of this kind.

Another Essential poll has consistently shown vast support for some level of penalty rate pay, but abolition of such rates isn't really on the table.  A further Essential poll at the start of 2015 found voters opposed more broadly cutting "weekend and public holiday penalty rates for hospitality and retail workers" (23-68 including 37-53 opposition among Coalition voters).  This suggests the result changes when the cuts in question are limited to Sundays.  Voters tended to consider that it was more likely that businesses would simply make more money than that businesses would employ more workers.  (This is another question where more could be done since the possibility that businesses would employ the same workers for longer hours wasn't included, and the impact of the changes on consumers didn't even rate a mention.)

All up the Australia Institute's finding that voters generally don't like changes to Sunday penalty rates when these are proposed in isolation is consistent with what other polls have shown.  However their finding concerning Coalition voters specifically is not consistent with the limited other polling available.  This is also an issue on which there is not that much polling available, and what there is is mostly open to question.  So the debate about public opinion on penalty rates has a long way to run - as does the penalty rates debate itself.

See also Peter Brent ("Penalty rates debate: Turnbull, Shorten tread carefully")

(Personal NB: I have no vested interest in this debate and really no strong view on it.  I rarely even work on salary as opposed to contract at all, let alone in an industry where penalty rates matter.  I suppose I'm affected to the extent that it contributes to my home city, Hobart, going to sleep at about 2 pm every Sunday (which is a nuisance). On the other hand, over the rest of the weekend I'd rather be served by people who are satisfied with the work they're doing and the pay they're getting, rather than giving up more of the weekend than they'd like to make ends meet.

There's the argument (appealing to me) that if an employee wants to agree to work for a lower rate because they personally couldn't care less about Sundays they should be able to, but that then flows into the broader question of competition for work, which in turn flows into the question of treatment of those without it, and it all gets rather messy.  I like the idea of treating Saturdays and Sundays similarly (since the notion of imposing a religious day of rest on a secular society is irksome) but if that points to a cut in one, it might be seen as pointing to a rise on the other, or to some form of special compensation for those who work only or mainly on weekends.)

Added Jan 13 2016: I have also found two more penalty rates ReachTEL results.  In March 2015, 67.5% thought people who work on weekends and holidays should receive penalty rates, 18.6% said no, 13.9% undecided.  In August, the question was:

"Do you support or oppose the Productivity Commission's draft recommendation that Sunday penalty rates should be cut to Saturday rates in hospitality and non-essential services? "

This got the thumbs down, 33.3% to 53.2%, a result virtually the same as Essential's, but with a stronger partisan divide.


  1. Why are the questions posed by the pollsters apparently so poorly written. Is this a job they give to the work experience kid?

    1. For commissioned polls, the wording is often deliberately bad to try to skew the results. For non-commissioned polls, I think that sometimes they aim to inform voters in order to reduce the undecided rate, but underestimate the risk of adding bias in the process.

      Wording can make a massive difference. One recent US poll found it was possible to halve polled support among Democrats for a policy by saying that Donald Trump supported it.


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. Comments will be published under the name the email is sent from unless an alias is clearly requested and stated. 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.