Talk:Opinion polling for the 2016 Australian federal election

More frequent updates of graphical summaries?

Hi, thanks to whoever is doing them, but once a month now seems not frequent in an election campaign. Any chance of doing it more often? Tony (talk) 03:06, 28 April 2016 (UTC)[reply]

Not only are we approx less than a week from the election officially being called, BludgerTrack has Labor leading the ever-lagging trend for the first time. Definitely need fortnightly or even weekly updates from this point in, monthly is just not sufficient - as the current state of the graphs make it look nothing like BludgerTrack, but instead gives the very false impression that the Coalition are set to easily win the election. Timeshift (talk) 11:12, 29 April 2016 (UTC)[reply]

Two-party-preferred polling average by state

User:Timeshift9 deleted my attempt at explaining this table and somewhat huffily said "perhaps you might want to read Two-party-preferred vote, almost everyone on here completely understands this table." The table is ambiguous - even going to the source cited did not make it clear to me what those numbers are. What is it an "average" of? It doesn't seem to be an average of polls since the last election, even though some of the charts have a time axis from 2013. Is it only the last week's polls? By the fact that these numbers are updated weekly, that might be the case, but the table cited as reference doesn't say one way or the other. It could be a running average. It doesn't even give the date - I inferred that from the text in the link on the page.

If you are one of those "almost everyone" who "completely understands" the table, it would be helpful if you could confirm or correct my explanation.

Wikipedia is not a website for people "in the know" only - although in this case I wonder how many people are actually "in the know" - it's supposed to be comprehensible to general humans too. --PalaceGuard008 (Talk) 16:45, 24 May 2016 (UTC)[reply]

Anything not comprehensible doesn't last long. You're the only one so far to take issue with it. Not to mention... 'the source is ambiguous'? If you have trouble following the source then that's a reflection on only one person - you. You don't have to be familiar with pollbludger to understand the source link. Do you have trouble following one column to the next? And why would it be anything but swing since the last election? Everything we have on swings are since the last election, and the source only has the one state swing % column which is clearly beneath the since last election column, two clear indicators it's since last election. Either your comprehension is very poor, or for whatever reason you're making a mountain out of a molehill. Timeshift (talk) 16:55, 24 May 2016 (UTC)[reply]

User:Timeshift9, you are basically resisting an improvement to the article on the basis that you don't need the explanation. That may be so, but judging by your edits to the explanation that is because you have a confident but possibly incorrect understanding of the chart. The explanation, as amended by you, seem to indicate that it is an average of polls since the last election. That doesn't look right to me, but the source cited doesn't make this clear. In so confidently asserting your confident but possibly incorrect interpretation, you are accusing an editor of long standing (longer than you) of having "very poor comprehension". I don't think this is a sound basis for continuing this discussion. I'm going to alert other contributors on the main election page to this discussion and seek their views, but I will bow out here. --PalaceGuard008 (Talk) 09:53, 25 May 2016 (UTC)[reply]

Not even going to bother with you. Timeshift (talk) 11:36, 25 May 2016 (UTC)[reply]

I'm sorry what? What's your issue with my contributions? --PalaceGuard008 (Talk) 13:10, 25 May 2016 (UTC)[reply]

It's not an average, it's an aggregate. BludgerTrack does not average the polls, it assigns different weighting to different polls and eases older polls out of the sample. I'm not entirely convinced it's information we should be including, but if we are it should be described accurately. BludgerTrack's methodology is explained here, and if this information is to be included we must have a summary of how these numbers are arrived at. At the very least it should be below the raw data so as not to assign it undue importance. Additionally there are other aggregates run (by Kevin Bonham, Mark the Ballot, the Guardian and others), so we should not be acting as though BludgerTrack is the arbiter on this. I am inclined to think we should leave it out altogether as including it is rather confusing and gives the implication that it is some sort of poll, but I'm open to hearing options on how it could be included without being misleading and without undue weight. Frickeg (talk) 10:59, 25 May 2016 (UTC)[reply]

I recognise the concern Frickeg, but wouldn't the best way be adding the methodology link next to the source link and in the sentence prior to the table, add "as published by The Poll Bludger? Anything else is just interpretation, duplication and a big ugly unneeded block of text. Have another look now with that done. Timeshift (talk) 11:32, 25 May 2016 (UTC)[reply]

What's there now is better, but this is one place where I really think we need that text to explain the information. We can't expect our readers to know what an aggregate is, and we can't make them go to an external link to interpret information we're presenting. If we want state-and-territory specific information (and I think that would be a good idea), wouldn't it be better to present the raw data and so avoid all this complication? I know some of the polls at least provide that breakdown. Frickeg (talk) 13:22, 25 May 2016 (UTC)[reply]

By explaining the methodology you're creating WP:OR. We seem to trust readers to know what an aggregate is in the above section don't we? But luckily we have an aggregate blue link to explain it. Timeshift (talk) 13:24, 25 May 2016 (UTC)[reply]

What? No we're not. We'd be sourcing it, obviously. But I still think the raw data would be better. Frickeg (talk) 20:38, 25 May 2016 (UTC)[reply]

State-based figures

Timeshift, wow, this is a big gyration. Have they changed the methodology in the territories? Tony (talk) 03:03, 6 June 2016 (UTC)[reply]

Suggested changes

If we're going to have state-based figures, can we please have the actual figures and not the PB aggregations? We have a whole page for this stuff now, there's no need to truncate the information. Also, with individual seat polls we really need the full primary results. I realise not all polls will publish that, but not all will publish 2PP results either, as this Cowper one shows, and the "assumption" made at the moment is totally unacceptable. Frickeg (talk) 12:54, 15 June 2016 (UTC)[reply]

Please no, not the primary votes, it's just too much. If someone wants to know they can follow the refs/google it. Seat margins calculated via 2pp (and sigh, 2cp) is the headline, particularly for seat-level polls, not the often patchy and widely MoE-varied primary vote. I was already put off by the widening of the graph from alp/lnp to all 2CP votes. I feared this when I added the first half dozen alp/lnp in a nice small graph that added valuable information in an article about opinion polls for the election. Timeshift (talk) 02:31, 16 June 2016 (UTC)[reply]

But in this Cowper one, someone ORed the 2PP margin by adding up the Ind/ALP/Grn primaries, which is not remotely OK but was their only option as the poll did not publish a 2CP. Why is it too much? Why should people have to click through to a ref to get the full information? We can easily narrow the graph in other ways - by getting rid of the useless "Seat classification" column, to start with. Frickeg (talk) 03:44, 16 June 2016 (UTC)[reply]

I agree with Timeshift that 2PP should be the dominant form; and with Frickeg that OR in 2PP figures should probably not appear (aren't there hidden assumptions?). Frickeg, any chance of a graph update? Tony (talk) 05:04, 16 June 2016 (UTC)[reply]

Well, not without adding in primary votes ...? I don't suppose there's a way to make columns collapsible? Frickeg (talk) 08:17, 16 June 2016 (UTC)[reply]

Poll authenticity

There have been a number of party commissioned polls but the poll of Adelaide published recently has been called into question. ReachTEL was said to have conducted it but it has denied its involvement, so it is now just a leaked poll of uncertain origin. Should it be removed or is the low sample size reason enough to leave it? I argue it should be removed until it is authenticated https://blogs.crikey.com.au/pollbludger/2016/06/21/electorate-polling-round/ --Z3nertr8p3r 01:25, 23 June 2016 (UTC) — Preceding unsigned comment added by Xenoworker91 (talk • contribs)

Poll Bludger appears to have asserted it was a ReachTEL poll "The poll was conducted for the ALP by ReachTEL" and then retracted that "UPDATE: And sure enough, ReachTEL denies it was their poll" yet the article they link to does not refer to ReachTEL. The source article says it is internal Labor polling. As far as seat-level internal polling goes, we've been adding them all. If you don't know how to view paywalled articles, this is what you do: go to news.google.com.au and in the search box paste this part of the URL: labor-frontbencher-kate-ellis-headed-for-defeat-to-liberals-in-adelaide-alp-poll-shows/news-story (change as needed for other articles) and search and click on the resulting provided link which will open in full. It is a very low sample size, but we've stated it so it should speak for itself. Finally, let's be clear here - I wouldn't read too much in to this poll in particular (I wouldn't read too much in to any single seat-level poll) as reading between the lines it's clear Labor released this poll and it's result for strategic reasons. The real result is likely Labor only just leading but way too close for comfort. They're attempting to ensure soft Labor voters aren't complacent and potentially vote Liberal thinking they can do so without consequence. If Labor wanted to release a credible poll they would have polled far more than the campaign low of 364. But this is all officially WP:OR and is not up to us to pick and choose which polls should and shouldn't be added for various reasons and rationales. Timeshift (talk) 01:39, 23 June 2016 (UTC)[reply]