Talk:2014 Tasmanian state election

polling in the infobox?

should this be there for this particular page? EMRS always has a very large undecided vote which they include in their polls (they don't push their respondents for an initial answer) so it seems a bit deceiving to put them in the infobox. On polling terms, Labor is only down 3% from the election. When compared to the actual vote, they're down 18%. It's wrong and disengenuous to do this and put them in the top infobox for comparison, I think they should be kept in the polling section for this page. Timeshift (talk) 02:16, 13 March 2011 (UTC)

What happened with the polling? I've made corrections based on the source. I've only gone back a few polls but more would seem wrong and probably need fixing. Timeshift (talk) 04:30, 10 August 2014 (UTC)

On further looking, EMRS have different results from one publication to the next? Exhibit A. Exhibit B. Errr....... Timeshift (talk) 04:33, 10 August 2014 (UTC)

I think it comes down to which table you use in the EMRS polling. A quick look at the source reports and it looks to me like you're looking at the figures which combine voting intention and 'leaning' towards a party? The convention I've followed in the past was to only use the voting intention figures (excluding the 'leaning' of undecided voters)—these seem to be the "headline" figures that EMRS plot in the graph on the front page of their website. I agree these should not really be in the infobox. --Canley (talk) 07:27, 10 August 2014 (UTC)

Look deeper. Using the two links I gave, look at Sep 2013. One has an undecided figure of 12%, the other 17%. Or am I missing something? Timeshift (talk) 07:44, 10 August 2014 (UTC)

EMRS's methodology is to ask the respondent who they would vote for if an election was held tomorrow. The headline figure on the chart is the percentages of decided voters who expressed a clear preference for a party, excluding undecideds. The undecideds are noted (17% in September 2013), then they are asked if they are leaning towards a party (5% were leaning to a party and were allocated in the combined table to those parties [2% Labor, 2% Liberal, 1% Independent], the remaining 12% were still undecided and had no leaning). So the difference is in May 2014, EMRS changed their reporting method and show a year's historical data with the leaners allocated, however in September 2013 the leaners were still counted as undecided in the historical table. For consistency, I think the raw figures (pre-leaning allocation) as published in the February 2014 and previous reports should be used for the 2014 election article. For the 2018 article, I guess we'd have to go with the assigned leaning methodology if that's what they are using from now on, as I don't think they publish the raw undecided figure any more. For further reading: Kevin Bonham mentions the change here, and goes into some detail in the comments of a Poll Bludger article here. --Canley (talk) 11:46, 10 August 2014 (UTC)

Ohhh, they've changed it. Thankyou :) Timeshift (talk) 01:16, 11 August 2014 (UTC)

2PP estimates issues

I've removed the 2PP estimates for this article and also 2018 as this is not a valid 2PP estimation method for Tasmanian elections for the following reasons. Firstly, preferencing is semi-optional in Hare-Clark (many voters vote 1-5 or as it soon will be 1-7 for their chosen party and then stop) and as a result the preference flows between parties are always considerably weaker than they are in federal elections, and more like they are in NSW elections, with a substantial flow to exhaust. Secondly, how to vote cards are banned, which is likely to slightly reduce the propensity of Green preferences to flow to Labor. Thirdly, from time to time the "Laborial" dynamic in Tasmanian politics on resource issues seems to drive a weaker preference flow as well, but it varies a lot (eg in 2021 the flow of Greens preferences that didn't exhaust was very strongly to Labor). The issues are most starkly illustrated by the 2010 election. At the 2010 federal election Greens preferences in Tasmania split 79-21. At the state election however 34% of preferences leaving the Greens (some would have come from other candidates) that had an opportunity to flow between the major parties exhausted immediately, and of those that did flow to a major party the split was only 52.6-47.4 in Labor's favour. With the Greens polling over 20% in that election, applying federal flows causes a 2PP error of around 11% compared to the actual votes cast. It is possible - but of debatable value - to construct 2PP estimates for Hare-Clark using the actual preference distributions but federal elections are not a valid model. Therealsleepycat (talk) 07:21, 30 November 2023 (UTC)