Friday, July 12, 2013

New Not-A-Poll Added: Tas Labor Seats At The Federal Election

I've added this site's third ever Not-A-Poll.  The first was to obtain reader feedback on the use of jump breaks in articles (it went to preferences!) and the second was Best Tasmanian Premier of the Last 30 Years.  This one (in the sidebar on the right) asks interested readers to pick how many Tasmanian seats you think Labor will win at this year's federal election.  Once I know the election date it will be set to close at midnight the day before.  I believe that even after you have voted you can change your vote. At least, I can.  (Maybe you need to be logged in.)

Here's a quick form guide thus far.  (Or if it's really all too hard, just roll a six-sided die and subtract one; you'll have a one in six chance of getting it right!)




Bass (ALP 6.7%): A seesawing seat that has changed party five times in the last twenty years and has sometimes been won by extremely small margins.  The incumbent is first-termer Geoff Lyons, and the Liberal opponent is Brigadier Andrew Nikolic, who moved to the area two years ago.  Two ReachTEL polls with Julia Gillard as Prime Minister showed Labor being thrashed 60.3:39.7 and 61:39 in this seat but no polling has been seen since Kevin Rudd resumed the top job.  (Update Aug 29: ReachTEL polls continually show Labor trailing in this seat and it is now at odds of $9 to retain it.)

Braddon (ALP 7.5%): Incumbent Sid Sidebottom has held this seat since winning it from the Liberals in 1998, except for the 2004-7 term in which he was briefly unseated by Mark Baker on the back of a voter rebellion against Mark Latham's ill-disciplined Tasmanian forests policy.  The seat is harshly affected by Tasmania's current economic downturn and the June ReachTEL showed challenger (and ex-state MHA) Brett Whiteley with a massive 62.2% 2PP.  However, again, no post-Gillard polling has been seen. (Update Aug 29: ReachTEL polls continually show Labor trailing in this seat and Labor is now at odds of $6 to retain it)

Lyons (ALP 12.3%): This seat (formerly Wilmot) covers most of the state and has had only three members in the last 67 years.  Dick Adams won it in 1993 when Liberal Max Burr retired and his challenger is Eric Hutchinson, who also ran against him in the 2010 election. Despite the seat's large margin it is also affected by the dubious state of the Tasmanian economy and has been showing large swings at state level.  The June ReachTEL showed a whopping 21% swing to a 59-41 lead for the Liberals, but I don't think anyone really believes they were quite that far ahead, and most observers now think the seat will be held unless the wheels fall off the Rudd return.  Still one to keep an eye on. (Update Aug 29: Despite the massive swing required, ReachTEL polls continually show Labor trailing in Lyons, although some are treating them sceptically.  Betting markets are very closely split on this seat, with the Coalition moving into very marginal favouritism for the first time today.)

Franklin (ALP 10.8%): Another Labor pickup from the class of '93, this seat has been held by now Cabinet minister Julie Collins since 2007.  The ReachTEL showed her marginally trailing (49-51) behind Kingborough councillor and Brave Foundation director Bernadette Black, but with the change at the top and Collins' subsequent promotion since what danger there was should have subsided. (Update Aug 29: The most recent ReachTEL poll showed Labor very marginally ahead in this seat.  Betting markets have Labor favoured to retain it, but not heavily.)

Denison (Ind 1.2 vs ALP): An extremely safe Labor seat until the retirement of Duncan Kerr after 23 years in 2010, but at that point Labor dropped the ball by picking a dud candidate and running a relative non-campaign, and the seat was scooped on Green and Liberal preferences by left-leaning independent Andrew Wilkie.  The only credible polls conducted since (a couple of ReachTEL's) have shown Wilkie with a large primary lead over Labor's Jane Austin, the Liberals' recently announced (and hitherto unheard-of) Tanya Denison and the Greens' Anna Reynolds.  However, the how-to-vote card intentions of the trailing parties are unknown. (Update: It is now known that the Liberals will preference Wilkie, which is all you know and all you ever need to know.  ;)  )  (Update Aug 29: Labor has strangely preferenced the Liberals ahead of Wilkie in this seat, but on current polling it is unclear Labor really will finish third after preferences, and probable that even if they do, Wilkie's likely primary lead won't be closed down.)

Not-A-Poll is completely unscientific and highly stackable and intended mainly for fun. Feel free to post any predictions in comments too.

Update (21 July): With 100 votes received, Three has taken a narrow lead over Two.   Initially it was the other way around, but the lead changed some time in the last week.  Also firming has been Four (the status quo) which at one stage was last.

The oddity in the Not-A-Poll so far is the level of voting for Five.  I believe this mainly represents Labor supporters voting for what they want to happen rather than what they actually think will happen. If so, then either Liberal supporters are less prone to voting by wishful thinking, or fewer Liberal supporters have so far read this poll.  (The latter is to be expected as the major venues through which this site gains hits all have a leftward skew.)

In the last week, I obtained ReachTEL state figures from their most recent sample, and these were far more Labor-friendly than the previous state Labor ReachTEL breakdown.  The samples are in both cases tiny, but combining the samples and adjusting for the apparent slight Coalition lead in ReachTEL's federal sampling, it seems the ReachTEL state samples (sample size 285) suggest a Labor 2PP of around 52.8% in the state (margin of error 5.8%).  At that level Bass and Braddon would be probable but far from certain Coalition pickups, assuming the state swing wasn't uniform.  With a uniform swing Braddon would be a tossup.

The Morgan Tasmanian samples since the Rudd switch thus far (sample size 544) point to a Labor 2PP of around 58.8% (MOE 4.1%) after adjustment for house effects - ie scarcely any swing and no seat losses to speak of.   The problem here is that I don't know what sort of results Morgan were getting before the Rudd switch, except for one result (59-41 in Labor's favour off a very small sample) that is very difficult to credit.  So it is possible that Morgan's internet sampling methods naturally lead to a very pro-Labor result in Tasmania and therefore that the new figures are exaggerated.  I'm reluctant to place much weight on them unless they are confirmed by other sources. 

We will need to wait for seat-based polling but based on what we have, I would not expect that Labor is a long way behind in Bass and Braddon anymore.  Also, it's possible based just on the ReachTEL samples (small as they are) that the Rudd bounce has actually been larger in Tasmania than nationwide, something that is a bit surprising if it's true.

Update August 7:  With 200 votes received, Three remains in the lead over Two.  Since the 100-vote update there have been hefty swings to None (+5%), One (+7%), small swings against Two and Three and a 5% swing against Five.  I suspect the latter appears because the same ALP loyalists who voted strongly for Lara Giddings early in the previous Not-A-Poll also voted early in this one (which is not saying everyone voting for Five fits that profile.)

Between the 100 and 200 vote profiles there was of course the Bass, Braddon and Lyons ReachTEL, returning narrow to moderate Liberal leads in all three of these seats from quite large samples; my discussion of it is here

Update August 15: With 259 votes received, Two has just caught Three.

With the addition of 59 votes, None has gained 9, One 8, Two 23, Three 14, Four 3 and Five 2.  That's in theory an average of exactly 2, compared to the average of 2.45 for the first 200 votes.  However, it's not that simple, because some people who have already voted are changing their votes.

The rush to Two could be down to anything, and not-a-poll is far from being immune to ballot-stuffing (the question being why anyone who wasn't extremely bored would bother.)  However, a possible contributor to the rush to Two today is a gaffe by Geoff Lyons (Bass) in which he made false statements diminishing the military service record of his opponent Andrew Nikolic. 

ABC report that the Coalition believes they are winning Bass and Braddon while the ALP strongly disputes this and believes it is saving the lot.  I have not seen either side's internal polling and hence cannot comment on it.

Update 19 August

Watching this is more fun than watching the betting markets!

Since the 200-vote update the swings are against three and up and to two and down, with the biggest swing from Three to Two, which have swapped places.  The average for votes post 200 is 1.88 (noting that this may be affected by people changing their vote), and the average overall is now 2.26.

Update 29 August


A ReachTEL poll with Labor trailing badly in all of Bass, Braddon and Lyons and barely ahead in Franklin

400 votes are in and Two now has a large lead, but in the last 98 votes received, there has been a bigger net gain for One (34) than Two (30).  There have also been more votes for None (16) than the former leader Three (12).  At the current rate One will very soon move into second place.  The overall average is 2.10 but the average of votes added since the last update is 1.57.  In the time it took me to write this, One received another vote and moved into a tie with Three.

There may be a glitch in the vote-changing function; it may not deduct your previous vote from its former category.  If you notice this and want to change your vote, don't let it stop you.

Final (Sep 7): It's been a fun exercise.  Over 500 votes were received, though I don't know how many of them were multiples:

Of the set of votes since the last update, five received one vote, four received eight, three received eight, two received 29, one received 50 and the Wipe-Out Gang scored 26.  That's an average of just 1.39.  But the average for votes for the whole campaign was 1.92, and Two's lead was too large for One to close it down.  So votes in the last three weeks have leaned towards one, but the vote overall has lag from those who voted in its early days. The vote for none in the last few weeks (22%) was not too different to the estimated chances of the wipeout on betting markets (which closed around $3.50).

Thankyou ever so much to all those who voted; we will do this again in the State election. 


No comments:

Post a Comment