Talk:2017 Bennelong by-election

Timing
Surely this article is premature. Alexander only has said he intends to resign. He must tender his resignation to the Speaker for it to be effective and not merely make a media announcement. He may not be eligible to stand as he has still not filed renunciation papers. To sum up, there really isn’t much to talk about. Enthusiast01 (talk) 23:44, 11 November 2017 (UTC)
 * I thought it might have been premature when I found it, but it's clearly time to have the article now that writs are issued, so it was at most 48 hours early. --Scott Davis Talk 04:29, 13 November 2017 (UTC)
 * Alexander is disqualified by operation of Constitution s 44(i); he was never (validly) elected.  His "resignation" was only his own recognition of that.  I guess this is why, a couple of days ago, Brandis referred in the Senate to a "purported resignation" of Parry. So this article was appropriate as soon as the evidence showed that Alexander was disqualified. Wikiain (talk) 21:56, 15 November 2017 (UTC)

Infobox
Do we really need the infobox on this page? It's almost never used for by-elections (there appears to only be one prior example where it was) and is going to get pretty unwieldy relative to the page. I suspect it will also probably not get fully populated as it's unlikely to be possible to get suitable free images of minor candidates, who may well never appear in person. It makes more sense for a general election. -- Rob.au (talk) 11:59, 22 November 2017 (UTC)
 * I'd say it's already unwieldy. Let's just have the photos of the only two candidates with a serious chance of winning, going by the 2016 result: Labor and Liberal.
 * Also, in the inbox "before_election =" appears as "Incumbent MP" and there is no incumbent. Wikiain (talk) 22:15, 22 November 2017 (UTC)
 * I've removed the infobox. I'm not sure on the best treatment of including photos of the lead candidates.  I'll think about it but anyone can jump in and do it. -- Rob.au (talk) 08:17, 23 November 2017 (UTC)
 * Removed again. Bloated and unnecessary for such a local matter. WWGB (talk) 11:32, 23 November 2017 (UTC)
 * I have also removed the misleading "info"box and replaced it with just the map and caption. I wonder if in fact the long caption should move to article body instead of crammed into the narrow space under the map. --Scott Davis Talk 10:33, 18 December 2017 (UTC)
 * Restored the original caption, which works just fine (though now includes a bonus Wikilink to the electorate's own article). This article does not need a long-winded description of the location of the electorate - it has it's own article where that can be found. -- Rob.au (talk) 15:04, 19 December 2017 (UTC)

Polls
Timeshift9, for the latest ReachTEL poll you have substituted for the SMH figures some different figures from The Poll Bludger. What is the latter and why should it be preferred to the SMH? Apologies for my brevity. Wikiain (talk) 07:24, 14 December 2017 (UTC) Wikiain (talk) 07:24, 14 December 2017 (UTC)
 * I've reverted to the original SMH report. If you still want to prefer The Poll Bludger, please justify. Wikiain (talk) 22:04, 14 December 2017 (UTC)
 * Because there is no undecided percentage in the other polls or the previous election result, making the pre-undecided figures meaningless as a comparison or an indicator. Furthermore, all polls where possible are added to wikipedia with the non-undecided figures. The primary votes after undecided is distributed is never a guesstimation, it is the second set of numbers provided by the polling firm. As per poll bludger: "... after allocating a forced response follow-up from the 2.4% who initially professed themselves undecided ..." - this confirms, if there was any doubt, that the sans-undecided primary vote totals are and can only be from the polling firm. If the question is over poll bludger itself, it is the most often used source throughout ozpol polling on wikipedia, with perhaps the exception of direct links (usually PDF or graphic) to the actual poll result table published by the polling firm itself, though public availability from one to the next is patchy at best. Furthermore, poll bludger sources are of undisputed better use as they simply report all known poll figures (poll bludger's raison d'être makes it the perfect polling source), which is the only thing we are referencing - as opposed to media outlet corporation references who scatter and bury often incomplete poll results in a potentially POV news media outlet article where instead, the main purpose is translating a poll for newspaper-like consumption for the masses, which does not in any way add (but muddies) to what is being referenced - making it the less desirable source to use. Furthermore, the ReachTEL polls in question are commissioned by SMH, making the SMH invalid as a reference as it is a WP:PRIMARY source from a subjective media outlet. Furthermore, poll bludger would have the sample size (of 819 rather than 864) correct as this is exactly what poll bludger lives and breathes on. The fact that SMH claims 864 for the 12 Dec poll means they confused it with the 16 Nov sample size of 864, because even within the same polling firm, the sample size from one poll to the next is never the exact same (or as close to never as one can be without simultaneously knowing every last ozpol polling result). To do so would go against all ozpol polling history. What is almost certainly the case (rather than the poll bludger randomly picking the number 819 from thin air and using that as the sample size) is the SMH got a trivial dry fact like the sample size confused with the Bennelong ReachTEL poll a month ago. Furthermore, this SMH article says 819, not 864, was the Bennelong by-election December ReachTEL sample size, further cementing what i've said. If the SMH provided an actual ReachTEL link to the full non-subjective poll results (also trivial and dry to a media outlet) then that would help, but there isn't one, and on the reachtel website and google I can't find one. In lieu and in summary, poll bludger is the only practical RS. So then... justify? Justified. And so comprehensively and convincingly, even by my standards, if I do say so myself :) Timeshift (talk) 23:52, 14 December 2017 (UTC)
 * (1) How does The Poll Bludger get better access to poll results than the commissioning organisation reports, if (and I agree) the actual results are not directly available? (2)  A "forced" response surely cannot have the same reliability as an initial voluntary response and so should not be used to modify the initial figure.  If I got a robocall from a pollster (as well as being annoyed by calls from parties) and pushed buttons to say I was undecided, and then the pollster phoned me to "force" a decision, I'd either refuse to answer or tell them just anything in order to get rid of them. (3) Your point about 819 or 864, though, sounds right. Wikiain (talk) 02:10, 15 December 2017 (UTC)
 * How does he constantly get new (and correct, so your concern/query is merely academic) polling up on the poll bludger website before it is available anywhere else online? I guess that's just part of the poll bludger mystique! Don't you know politics 101? The only thing more important than not revealing one's sources, is consistently being the first to go public with the breaking info, to the point that nobody (well, almost anyway...) cares about such academic irrelevancies of who the source of the source is! :) Timeshift (talk) 02:39, 15 December 2017 (UTC)
 * You seem to want people to believe that William Bowe the Poll Bludger gets inside information from both the Fairfax and the NewsCorp newsrooms, and/or from rival pollsters ReachTEL and Galaxy/Newspoll. Maybe he does, but to my mind these implausible implications—or, alternatively, your comment about "poll bludger mystique!"—raise, for me, a WP:RS problem.  Whether his figures are accurate is the very question. Wikiain (talk) 05:03, 15 December 2017 (UTC)
 * But if the shoe fits! And I never said he gets his information from newsrooms or pollsters. All I said was that somehow, he is usually the first to make it public, or at a minimum, have more complete polling figures than what is initially published in the media. We might not know his sources, but the remarkable consistency of his a) later proven accurate, and b) first-to-publish, polling data demonstrates poll bludger's primacy. After all, why else has it been used so extensively throughout ozpol wikipedia for well over a decade without fuss? It is the largest central polling repository, overall the first to publish, ever-accurate, and unlike media outlets who publish something inferior later (and with unhelpful-to-our-purpose subjective editorial opinions to boot), has no primarysource problems. I'm really not sure why you have a problem with it, particularly when you can't actually fault it. You haven't actually said it, but perhaps you don't feel that William Bowe isn't sufficiently qualified? Well he is a regular participant in federal and state election night ABC coverage, providing his psephological input as a doctoral candidate with the University of Western Australia’s Discipline of Political Science and International Relations, and has taught politics and government at both Edith Cowan University and UWA. So, can we move on please? Timeshift (talk) 06:20, 15 December 2017 (UTC)
 * But regarding 'undecided'... Firstly, the undecideds are not really the undecideds. Those who are truly undecided won't nominate one party/candidate above all others after simply being asked again, and that is the secondary (very small) undecided percentage which is usually not provided. Those who went undecided on the first question and then nominated a party/candidate on the second attempt, still chose one party/candidate above all others, and before election day to boot. They are clearly leaning in a particular direction, or, were playing the polling game. Anywho, no matter what contest in what jurisdiction in ozpol... when there are pre-'undecided' figures published and yet there are also post-'undecided' polling figures available, we have always used the sans-undecided figures for ozpol wikipedia - and in case you hadn't, you might notice a distinct lack of 'undecided' columns with 'undecided' percentages across all ozpol wikipedia polling tables. After all, if we used primary votes without including those from the substantial 'undecided-on-first-ask' group, what use would primary vote polling be when wanting to use it to see the size of the primary vote swings since the last election? Perhaps only of use to partisans who might quote the artificially small or artificially large swings depending on the situation. They would not be of much use at all, except possibly if viewed in isolation for some esoteric/useless 'bigger picture'. Timeshift (talk) 03:03, 15 December 2017 (UTC)

I have to say the justifications for using the Poll Bludger figures are entirely unconvincing from a Wikipedia perspective, and the “primary source” argument makes little sense at all, it seems counter-intuitive to me. I get the bit about media being potentially partisan (though even this perspective departs from being neutral in the Wikipedia sense), but as the party that commissioned the poll, they are by definition the most reliable source failing the polling company releasing data themselves. The point with Wikipedia is that sourcing trumps accuracy. We may feel that the Poll Bludger is “right a lot of the time”, but with no knowledge of where they are getting the data, in the event of a discrepancy we are forced to err towards the party that commissioned it as they are the only known holder of the original data. I can’t comment on how that source has been used elsewhere on Wikipedia, only on this case. The argument for using “forced responses” from undecideds is also troubling, especially with a robopoll. Rob.au (talk) 03:40, 17 December 2017 (UTC)

Alexander official resignation date 11 or 13?
Division of Bennelong says Alexander resigned 11 November, Bennelong by-election, 2017 says 13th. SMH ref of 17 Nov says "resigned from Parliament over the weekend" (11-12th), ABC ref of 11 Nov says both "has resigned" and "Mr Alexander said he spoke to Mr Turnbull late on Friday and told him he intended to stand down" (the Friday was 10th). John Alexander (Australian politician) has 11th (prose x2 and infobox). 2017 Australian parliamentary eligibility crisis Timeline says 11 Nov he "announces his intention to resign" and on 13th "Alexander officially resigns", and its Overview has 13 Nov. His bio at aph members has gone. I cannot find the 'former members' page there. So is Monday 13th likely the date he officially handed resignation to Speaker and therefore the date we should use for all pages? JennyOz (talk) 03:27, 15 December 2017 (UTC)
 * Though he announced his intention to resign on 11 November and was official on 13 November, it officially took effect as of 11 November: John Alexander's official parliamentary biography states "Resigned 11.11.17"... so sayeth the almighty aph.gov.au! Timeshift (talk) 04:23, 15 December 2017 (UTC)
 * Ah thank you so much almighty sleuth! I'll have to become more familiar with aph site:) JennyOz (talk) 05:16, 15 December 2017 (UTC)

Results and sourcing
Pointed edit comments while making fairly superfluous edits to references are not really the place to discuss the issue of a claim that overstates an issue and is premature. The counting is incomplete, as is the receipt of postal votes. The numbers are still subject to change. All the votes counted on election night - mostly by casual workers employed for the day - are all recounted at the divisional office, and the tallies are rechecked. The numbers move. They do every time, and this is normal.

At present Alexander's primary vote is at 45.05%. We are apparently comparing it to a previous result of John Howard in 2007 of 45.49%, a difference of 0.44% of the vote. It appears likely that Alexander's primary vote will continue to go up, based on current postals still to be processed, and an allowance for further valid postals still to be received as is normal (based on the return rate in 2016). If everything continues exactly in line with current counting (including rejected and empty envelopes and informality rates) and what would be the normal continued receipt of postals at this point relative to the likely final number, the difference can be expected to come sufficiently close (probably less than 0.2%) as to be within the margins of entirely normal movement of results through this stage of the count as the votes are recounted and rechecked. This process will continue through into 2018 even if Alexander is declared elected in the mean time (that can be done as soon as it is impossible for any other outcome, even if the count is not finalised).

For this article to make such a prominent and significant claim of "lowest primary percentage ever" when the very status of the claim is based on these factors is highly problematic. I respect that other editors have edited a lot of election pages, but "we always do it this way" just does not address the discrepancy from WP policy when this claim is obviously premature at this stage of the count. Why exactly should this claim not be removed for the time being and restored if and only if the final numbers bare it out? Even then, to maintain WP:NPOV, the claim should note the small margin with which it became the lowest primary, if that does actually remain the case when the results are final. I just don't understand why this problematic claim is being forced into the article at this stage. -- Rob.au (talk) 14:18, 20 December 2017 (UTC)


 * The AEC results ref has only 1,092 remaining envelopes/votes to be counted of which 664 are postals and 428 are declaration and provisional. The current postal Lib primary, based on the 11,457 postals counted, is 51.75% compared to the overall Lib primary currently at 45.05%, with the yet to be included declaration and provisional which are weaker for the Libs. To get the current 45.05% Lib primary above the 2007-low of 45.49% at this very late stage, the Libs would need to get primary votes from least 75% of all remaining 1,092 votes, which is impossible - but to keep the unfamiliar happy, from the start I did include "so far". This record low Liberal primary vote in Bennelong's 68-year history is being compared to the next lowest primary for comparison's and context's sake, which is the 2007 result where the Liberal primary vote was 45.5 percent (and with 13 candidates running) which until this by-election had been the lowest. Furthermore, i'm not sure how being the last sentence in the entire article can in any way be viewed as "prominent". Particularly as it says "so far", and furthermore as the closing of the gap is impossible with only 1,092 envelopes/votes remaining of which 664 are postal, your concerns ring hollow. Timeshift (talk) 14:33, 20 December 2017 (UTC)


 * Your summary is incorrect. Based on previous elections, you can expect another 1160 postal votes to still be returned (and they can be accepted if received by the AEC up until 29 December, with the certificate signed before the close of polls on 16 December).  If you assume these continue to be accepted and rejected at the same rate, and contain the same percentages of first preference votes and informal votes, it will drive the gap down to about 0.2%.  While I note your uncivil reference to me as "unfamiliar", you've ignored the point I made that votes are being recounted and the numbers move as they are checked by permanent staff of the AEC, as observed by more experienced scrutineers in contrast to the election-night count.  This is completely normal process, and they can and do move by greater margins than this apparently record result.  Making this claim on Wikipedia is demonstrably premature. -- Rob.au (talk) 14:46, 20 December 2017 (UTC)


 * A week after a by-election, the vast majority of postals have already come in. It is not premature at all, however stating "so far" rather than claiming it to already be the case, a week after the by-election, should settle that for unfamiliar people. Closing the gap remains impossible. How many by-election results have you followed that make you think otherwise? Timeshift (talk) 14:58, 20 December 2017 (UTC)


 * I don't see the incomplete nature of the count as the main issue here. We need a reliable source identifying the information as significant - that does NOT include our own articles or Psephos or anything else, which would make it WP:OR. I dislike these kinds of seat-specific records in any case (redistributions make the whole thing of questionable merit), but we absolutely need a source saying "this was the lowest primary vote ever in Bennelong" before we even think about adding it. Frickeg (talk) 21:09, 20 December 2017 (UTC)


 * Assuming the 2017 by-election Liberal primary vote remains the lowest Liberal primary vote in Bennelong's 68-year history, why do we need another source to confirm a second time? We already know without any doubt that before 2017, the lowest Bennelong Liberal primary vote was 45.49% in 2007, as per source used to create the 68 years of results at Electoral results for the Division of Bennelong and the other 149 divisions. This isn't one of those strange, esoteric HangingCurve additions to a Division of X article where something is apparently the third highest something on a full moon with one leg held up - I agree with you on that. However, if the final results confirm it, the lowest primary vote in its 68 year history (for the party who held it for all but one term) is black and white - certainly the further-est thing from those questionable contorted electorate article claims. UPDATE: Ok, googled and found a Herald Sun ref: "In Bennelong on Saturday, Alexander went backwards by 6 per cent. His primary vote is the Liberals’ lowest since the seat was created in 1949 — even lower than when John Howard lost Bennelong in 2007". Happy? I am, thank you for the instigation - now that it is cemented we can all move on. Timeshift (talk) 21:20, 20 December 2017 (UTC)


 * I am happy with that source, yes. But I do think it is a good idea, as a rule, to require sources before adding these kinds of records. I still don't like them, but with a source there's a reason to put them in the article. Frickeg (talk) 05:48, 21 December 2017 (UTC)