Dr Kevin Bonham
However, whatever this poll says or doesn’t say about the two parties, it says a great deal about the popularity of Paul Lennon. There is no way out of it — assuming Labor’s support level is actually more or less intact (as I believe that it is) then Lennon is remarkably unpopular among those who vote for his own party — his real approval rating among Labor voters could be as low as 50%. Actually, surveying of approval ratings for Tasmanian leaders is so low-profile that I have few results to compare Lennon’s overall 32% with, but I’ll assume it is not a career high. It certainly doesn’t compare with his predecessor, who once polled a staggering 71%.
EMRS Poll (Feb 2007): ALP 30 LIB 28 GRN 17 IND 3 UNDECIDED 22
Bonham Interpretation: ALP 46 (-3 since election) LIB 35 (+3) GRN 16.5 (=)
ANOTHER EMRS poll is out (http://www.emrs.com.au) and it is much the same as the last. Once again, Labor is shown as way down on its election result, with a miniscule lead over the Liberals, and the Greens are polling steadily, while a great mass of voters are claimed to be undecided.
It is little different from the same company’s poll of November 2006, and tellingly, quite similar to their poll of November 2005 (and we all know what happened just a few months after that!) A lot of silly season stuff has been said in the media about these EMRS polls, and in view of their obvious methodical problems, I will now be including a “Bonham Interpretation” figure to translate both EMRS and Morgan polls into what I think they really indicate. Of course, even these indications are prone to the usual vagaries of sample size and issues coming and going, and the long term trends are most important, but I’m hoping even rejigging one of these things will be an improvement on the endless stream of raw-figure fiction we are getting from the press.
My colleague Peter Tucker (http://oldtt.pixelkey.biz/index.php/weblog/comments/2448/) has beaten me to the punch with much of the necessary analysis of this particular EMRS poll. Readers may have noticed that Peter and I have quite similar opinions on many psephelogical issues, especially when compared with some of the views expressed in the comments section on this site. However, I’ve done some extra number-crunching, and come up with some further points that I think are worth making from these figures.
The first of these is that, as comment author “Lee” points out in response to Peter’s article, the Liberal primary vote continues to go nowhere much despite the supposed massive crash in committed Labor support. These are the last five readings for the Liberals from EMRS: 25, 28, 27, 29, 28 (election 31.8). As Peter points out, the 34.5% the Liberals polled in the last Morgan poll (Morgan has a very low undecided rate) is doubtless closer to the real level of support — but there is still no real evidence of the Liberal vote lifting into the upper 30s. The only positively encouraging sign for the Liberals is that their new leader is popular, including among Labor voters. Their party vote has still to show any serious signs of lifting.
The second of these concerns the undecided vote. As Peter notes, it’s hard to know what to make of the high EMRS figure for voters said to be undecided or supporting “independent” candidates. In this case the two combined total 25%. However, in this case, the inclusion of approval figures for leaders by voting intention gives us a very useful window into what the “undecided” voters really think. It shows that most of them do actually have opinions about the leaders, even if they don’t clearly indicate which party they prefer.
By subtracting the approvals by party from the overall approvals per candidate, the following are the approximate approval ratings for the party leaders among undecided and “independent” voters:
Lennon: Approve 24%, Disapprove 63%, Unsure 13%
Hodgman: Approve 43%, Disapprove 33%, Unsure 24%
Putt: Approve 42%, Disapprove 47%, Unsure 11%.
In general, the “undecided” voters are less likely to approve of each of the three leaders, and more likely to be unsure about them, than the committed voters. When those who are unsure on a particular leader are removed, the views of those uncommitted to a party on each of the three leaders are more negative than those of voters who did support a party. So there is probably a “sick of them all” factor among some of those recorded as undecided.
Furthermore, those “undecided” voters with an opinion on leadership have similar attitudes to the major party supporters concerning Peg Putt, but their attitudes to Paul Lennon resemble those of Liberal voters, while their attitudes to Will Hodgman resemble those of committed Labor voters! Since the attitudes of committed Labor voters towards Will Hodgman are actually quite positive (his approval rating among Labor voters exceeds Rene Hidding’s peak among all voters!), it doesn’t make sense to assume the “undecideds” include scores of Liberal voters who deeply loved Rene Hidding but are parking their vote while they make up their mind about young Will. These figures are far more consistent with them being mainly Labor voters who are tentative enough to get flagged, in many cases incorrectly, as “undecided” by EMRS’s methods, which is what I have suspected all along. More significantly, what this all hints at, that was not apparent previously, is that Labor voters who are soft in their indications of support also have a very negative view of Paul Lennon personally.
The raw party support levels expressed in EMRS polls do not mean very much, because many voters who would be flagged as supporting a party by other polling methods are getting included in the “undecided” basket. I have mentioned one reason for assuming these are mainly dithering Labor voters above. Another is direct comparisons between EMRS and Morgan. Comparing the last five EMRS polls with the last five Morgan polls, I have noticed that the Morgan poll on average returns a total figure for the three parties that is 13.1 points higher than the raw-figure EMRS poll. Removing the typical 3% undecided from the Morgan poll, this difference falls to 10.3 points. Significantly, if the differences between the raw party figures (undecideds not reallocated for either poll) is looked at, nearly all of the difference is Labor — of the 10.3 points I mentioned, 8.8 points (85%) Labor, 1.2 points (12%) Liberal, 0.3 points (3%) Greens. Indeed, the recent Morgan poll showing 34.5% for the Liberals was the only case where either the Liberals or the Greens got a significantly better Morgan result than an adjacent EMRS. The only EMRS poll with a three-party total equivalent to the most comparable Morgan one (undecided subtracted) was taken just prior to the election when people’s minds were actually on politics. In a serious statistical no-no, this most unrepresentative of EMRS polls is being used to claim the Government has dropped fourteen points since the election. If that EMRS poll and its nearest Morgan counterpoint are removed from my own comparison, then the remaining four polls from each firm show a 13.2 point difference (Morgan minus EMRS), of which 11 points (83%) is Labor, 1.7 points (13%) Liberal, 0.5 points (4%).
Under these and other relevant circumstances, all kinds of adjustments could be used to try to translate the EMRS poll into something vaguely meaningful. For instance, suppose we add those voters shown by EMRS as “leaning” a given way to their party’s tally, and then distribute the remaining undecideds in line with the 83%/13%/4% ratio above. This gives results of Labor 44.8%, Liberal 33.7, Greens 18.5. Applying the special Minor Party Poll Fudge Factor ™ (two points off the Greens and one point onto each major party) gives us Labor 45.8, Liberal 34.7, Greens 16.5, a swing of just three points from Labor to Liberal since the last election, and very similar to the kinds of realistic Labor leads that Peter and I were speculating about in discussing his column on the Morgan Poll. (http://oldtt.pixelkey.biz/index.php/weblog/comments/best-liberal-poll-result-in-four-years/) As a real indicator of the comparative state of the parties, the EMRS poll raw figures are almost useless, as their “undecided” rate is way too high, and there is every reason to believe that “undecided” category breaks very unevenly by party and includes many voters who would be picked up as supporters of a party by other methods. EMRS has some serious survey method issues to address if its between-election polls are to be as useful as those it takes directly before an election when people are actually paying attention to politics. Otherwise we will someday see “LIBERALS HIT POLL LEAD” blared across newspaper front pages when they are actually still several points behind both on Morgan and in reality. Given the problems with interpreting EMRS polls, I actually think that at this stage, Labor has not that much to worry about. An eleven point lead isn’t typically enough to win outright in Tasmania, but not much more than that will quite often do the trick.
However, whatever this poll says or doesn’t say about the two parties, it says a great deal about the popularity of Paul Lennon. There is no way out of it — assuming Labor’s support level is actually more or less intact (as I believe that it is) then Lennon is remarkably unpopular among those who vote for his own party — his real approval rating among Labor voters could be as low as 50%. Actually, surveying of approval ratings for Tasmanian leaders is so low-profile that I have few results to compare Lennon’s overall 32% with, but I’ll assume it is not a career high. It certainly doesn’t compare with his predecessor, who once polled a staggering 71%.
My own view is that Lennon, since re-election, has not been clear about what kind of style he should adopt. Unsure whether to be a new Jim Bacon or a new Robin Gray, he is copping the rough reputation of the latter without the same reputation for barging through and delivering results. A current major example is the pulp mill, about which there seems to be nearly universal cynicism, with opponents of the project believing it will be corruptly approved and environmentally calamitous, while supporters expect it to be slowly and painfully strangled by procedural bungling and red tape. While paying lip service to major developments that generally never happen is a well-established sport in Tasmanian politics, Lennon has invested too much enthusiasm (and too many public resources) to get away completely unscathed should the project not proceed to fruition.
The loss of an ambitious Deputy who seemed cut from the same cloth has been a curse in disguise for the Premier, who instead of having a single obvious successor who posed no threat to him (why change like for like?), could now face eventual challenges from a range of Labor colleagues with quite different styles and appeals to his own. The risk to Lennon’s leadership has been overstated greatly (given that his party is probably really travelling not far below majority-retaining level whatever the raw EMRS figures say otherwise), and given that parties rarely dump a leader who they can still win with, but if he is still this unpopular in the electorate a year or so from now, it could get messy …
Kevin Bonham invites readers to write their own fake Bonham bio based on fictitious concoctions of misunderstood scrapings from Google – he notes that some contributors have already started!
