Saturday, March 1, 2014

Not A Poll: How Many Seats Will The Liberals Win In The State Election?

Readers may have noticed the lack of new articles over the past couple of weeks (though various old ones were updated, including frequent updates to the main guide page.)  The main reason was that I was away on fieldwork unrelated to elections and had not only limited internet connectivity, but also very limited spare time.  Anyway that's over now and I'll be online much of the time up to the state election, probably except for most of the long weekend.  There will be quite a few new state articles in the next two weeks (yes, that should include a certain other state!) and it's probably about time for another federal polling wrap soon.  Thanks very much to those who have so far donated money to support this site as without those donations I would not have been online over the last two weeks at all, and nor for much of the post-count.

Anyway, this thread just introduces my usual exercise in which you get to play psephologist by predicting an outcome.  In this case the challenge is to predict the number of Liberal seats that will be won at the state election.  Voting is on the sidebar.




In summary, the Liberals currently hold 10 seats (2 each in every electorate).  Polling has consistently showed them on track to pick up hefty swings at the expense of Labor and the Greens.  Based on seat breakdowns they have generally looked like gaining a seat in each of Bass, Braddon and Lyons. At times they have looked like gaining a seat in Franklin as well.  At the height of their polling, four-seat results in one or the other of the northern electorates looked possible but pollster house effects may have contributed to this.

It does seem the Liberal vote could be softening slightly, especially with the emergence of Palmer United as a competitor for the votes of those fed up with Labor and the Greens.  My own current view is that 13 seats is the mainline estimate, 14 for the slightly confident, 12 for those who are bearish about the Liberals' majority prospects, and anything else is adventurous to say the least.  But it will be interesting to see what readers think and whether they agree with me.  In very early voting, 12 has jumped into the lead; after 25 votes the combined majority options lead 12 by only one vote.

Anyway there is more than enough analysis of current polling up on this site in other articles for those who want to have a look at it.  This thread will track the progress of readers' estimates of how many seats the Liberals will win.  Updates will be posted every hundred votes or so. 

Not-A-Poll is unscientific, intended for fun and easily corruptible and stackable, and you can change your vote.  That said, its previous predictive adventures have been fairly accurate.

(PS: Yes, I know, it should say fewer than 10.  But once a poll has received a vote, the wording cannot be changed!)

Update 2 March: Well that was quick, 120 votes up already!

There is a pattern in Not-A-Polls that the earliest votes on the site tend to be more left-leaning than the later ones, and this is apparent today with 12 seats dropping from an early lead to third, with 14 the new leader.  Presently 68% of voters think the Liberals will win a majority, and the average expected Liberal seat tally is 13.03. There is some, but not that much, action outside the 12-14 band.

Update 3 March: With 172 votes in the bank, 13 moved into a tie for the lead with 14, on 52 votes each.  However since then there has been a surge for 14 which is the most popular option:

With more than 200 votes now in, 69% expect a Liberal majority, and the average expected Liberal seat tally is 13.00.  Outside the 12-14 band, there's a little more action below than above, with a flicker of interest in 15.

March 3: Not-A-Poll Stacked! It had to happen sooner or later, and it's not very technically difficult.  I was a little suspicious of the surge for 14 cf 13 in the above voting, but more suspicious when the most recent check revealed another flurry of votes that were almost totally for the Liberals:

Supposedly, of the last 53 votes, 11 have gone to 13, 21 to 14, 5 to 15 and 11 to more than 15, although prior to that the last option had less than 1% of the vote.  Also supposedly the total for non-Liberal majority options in these votes was 9.4% when previously it had been running at 31%.

Now an obvious innocent explanation for this would be that someone in the Liberal Party forwarded the link by email to a bunch of their comrades, resulting in a deluge of hits from different Liberals keen to express their view that the party would win.  And that would be fine.

But there's no evidence of that in my various site counters.  What there is, is evidence that a specific IP address (and yes, I have it recorded) has visited my site 87 times since just after 2 pm today. 85 of those visits were to the site front page from no referring link and two of them were to the graphics in this article.  Some of the visits were as little as ten seconds apart (with around 15 seconds a common break), though my guest took a break between 2:21 and 4:08 before resuming the volley.

It is of course entirely coincidental that the barrage (apparently originating from Victoria) started less than half an hour after I exposed a Liberal attack ad website on Labor as a clone of a Democrat attack ad website on Mitt Romney!  :)

After I filed this update there were another ten votes for 14 and another two votes for 15, and a similar number of hits from the same IP address. As such a heavy rate of stacking distorts the results so as to make them a useless indicator of reader opinion, I've removed the poll.  It was fun while it lasted, and for the purposes of testing how well it fared predictively, the 2 March update above is the official result.



8 comments:

  1. Kevin, would you like to comment on the poll commissioned by DIER on Tasmanians support for a 'forest peace deal'?

    ReplyDelete
    Replies
    1. I would like to see the exact wording of the poll questions together with any preamble that was included with the poll before doing so. I have not seen these anywhere online but have not yet checked the print edition.

      Delete
    2. Your right. There was 2 month delay after the DIER-funded poll was taken so it would fit into the election campaign. Asking people if they are sick of a 'war' is an easy way to get a positive result.

      Delete
  2. Unbelievable. Do people really have nothing better to do with their spare time? I'm a complete and utter political enthusiast who devotes much of my spare time on political related topics... And I wouldn't even do this for the side I'm supporting at any one time.

    It was no official poll, but a good insight into what more knowledgeable (about politics) think might occur. Geez.

    Considering we're, what, two weeks out now? What's the general feeling in Tasmania? Opinion heading back to Labor? Slipping further away? Or same old?

    ReplyDelete
    Replies
    1. I don't know if there is a "general feeling" or if it would be accurate (in 2006 there was a very strong "general feeling" there was going to be a hung parliament with only a few of us psephs flying the flag against); whatever the case there's been no fresh polling data for a few weeks. I'd expect we will see multiple polls over the final week.

      As for the not-a-poll stacking, when stuff like this occurs on this site, it is not necessarily politically motivated. It may come from people who hold grudges against me from other walks of life. The apparently fake "Mark Grewar" posts in comments to the Grewar article may well have come from a troll who I have repeatedly banned from chess forums, whose normal style just happens to resemble the real Grewar's loopier moments. A long time after the "Grewar" posts were deleted, that troll turned up and posted on it asking about the comments (I've since deleted his posts). However this particular exercise doesn't seem to be his work based on the IP data.

      Delete
    2. You need to take into consideration this is Tasmania using the complex Hare-Clarke proportional voting system to elect only 25 people. Kevin may agree this is a great explanation of it:
      http://www.abc.net.au/elections/tas/2006/guide/hareclark.htm
      The Libs are spending more than anyone else on TV advertising although I got an amusing PUP DVD in the mail today. I haven't looked at the Titanic 2 segment yet.
      It maybe a narrow Liberal win or another hung parliament with a different mix this time round.

      Delete
    3. In the north the general feeling is a narrow Liberal majority, with PUP a shot to pick up one in Braddon, and Labor picking up most of the rest.

      The idea of the Liberal's picking up the projected three in Franklin is considered dubious, regardless that the polls suggest that somewhat consistently. IF that occurs, then 14, if not 13.

      Delete
    4. There is a Mercury ReachTEL coming this weekend and I've already seen the results but can't disclose them; I'd expect the headline figures to come out publicly around midnight tonight and I'll have analysis up at 8:30am. I believe the Examiner have something coming which I'd assume to be the final EMRS.

      The Franklin thing which has been swaying in polling for some time (adjusting for Labor's advantage on leakage, ie the Libs need to beat Labor by somewhat more than a quota to win it) is a serious test of intuition vs data. How big does the modelled lead need to be before one says that it is happening?

      I have doubts about the strength of the Giddings ground campaign as a candidate. If Labor is reduced to one I'd not take for granted which one of the two will survive, at least not without seeing candidate-specific polling. Perhaps immaterial anyway as I can't see Giddings staying in parliament long if Labor is crushed.

      Delete