Saturday, March 15, 2014

State election night arrangements and viewing tips

This is the usual post to give details of election night arrangments for this site.


From c.6:30-11 pm I will be doing live coverage at (link above). Based on last election, I expect the link to be an interactive one where I can answer questions and will answer as many as I can, as well as providing constant updates and calls on who is in and out.  Please be aware that keeping on top of the figures and preparing comments is a busy process and so I may sometimes be slow.

I respectfully ask journalists (and anyone else unless they have scrutineering info) not to call me during this time.  If you read my coverage you may find questions answered there.  I may be able to do interviews late tonight and certainly tomorrow.  Scrutineering info from those who have my number (text preferred) is welcome at any time.

There will be a late night wrap on this site that will be posted sometime after midnight.  There will then be dedicated post-count threads set up to follow the progress of post-counting in each of the electorates.  

Here are some things to bear in mind when watching the coverage (whether online, at the TEO site, or on the various TV coverage):

* There will be exit polls shortly after 6 pm.  Southern Cross will have one and I believe Sky may too.  The raw vote percentage results for these things are often pretty good.  Turning these figures into seat totals is a tricky exercise so seat total projections from them should be treated cautiously.

* If the result is as decisive as it looks like being, we will know the result relatively early in the night.  But don't expect to know it based on the earliest booths.  These tend to be unrepresentative not only in electorate terms but also in swing terms.  For instance there will be vast swings against Labor in small timber booths in Lyons, as there were in the federal election, but through the night the swing in Lyons on an electorate basis will drop.

* There are substantial shifts from the final on-the-night figures to the final primary figures as postal and other votes come in.  This makes the calling of close seats difficult and has led to some seats being incorrectly written off in the past.

* While we should know the result tonight we are unlikely to know the final seat total in every electorate - on average about two electorates will have seat totals that are still up in the air.  We will also probably not know the fate of several candidates for many days because there will be close inter-party battles even when the party totals for the electorate are resolved.

* The final distribution of preferences does not take place until the expiry of the 10-day wait for postal votes to arrive.   In the past most of it is done in about two days (making Mar 25-6 the likely critical days) but the greater number of candidates may slow the count somewhat.  I'll be on remote fieldwork again and uncontactable by phone when this is going on, but will update at least daily and expect to take a day off fieldwork to work on the cutup if results look close.

* Remember that Hare-Clark is about candidates and not just parties.  So while it is normally possible to tell which parties will win from the quota totals, sometimes it isn't that easy.  For instance a scenario I've explored for Braddon involves Labor with as little as 1.3 quotas beating the Greens and PUP who may have half a quota each.  Preferences also play a role, although when a party's last remaining candidate is elected/eliminated a lot of its votes exhaust because of voters voting 1-5 and stopping.  Finally on that matter, leakage is important.  Parties lose votes to leakage when surpluses are distributed or when candidates with substantial numbers of votes are excluded.  It is a very complex matter to call Hare-Clark results on the night and sometimes even after!

I've removed the election-forecast graphic from the sidebar so that it is not confused with results from actual counting.  For the record this was the final version:

(The final Liberal win chance estimate was 97%.)

The aggregate changed from 14-6-4-1 after the Newspoll showed a bad result for PUP.  However there are so many uncertainties in this that the total will probably be something other than the one given; it will be surprising if every electorate goes as stated.  In particular, for Braddon there are four significant scenarios, each of which in my view has a less than 50% chance of happening.  And for Denison and Franklin, I assess the chance of the stated scenario at greater than 50%, but there is still a significant chance of something else.  It's notable also that some alternative results (like 15-6-4 for instance) have an increased chance because they have two alternative pathways by which they can occur.  Even with all the polling data we have had, Hare-Clark is always a difficult system to project.

I am especially unsure about Braddon because of the question of the reliability (or otherwise) of polling for Palmer United.  The apparent non-naming of PUP in Newspoll means their vote may be underestimated in my aggregate, but even if it is they are no guarantee to get over.

The other way in which my aggregate could well be off is if the Newspoll was actually completely accurate in capturing a late swing to the Liberals, especially coming off Palmer and the Greens.  Then there is the possibility of the Hare-Clark equivalent of a landslide, even up to (say) 16-5-4.

I hope people will enjoy my coverage tonight and I hope I'll be able to successfully highlight some of the tricks of the Hare-Clark system and how they affect tonight's results.

No comments:

Post a Comment