Talk:Opinion polling for the 2020 New Zealand general election

Naming convention for parties

Previous pages had the parties' names spelled out: National, Labour, etc. I don't think it's a step forward to start using three letter abbreviations (NAT, LAB). It also breaks a computer program I have that reads all the years' tables. Any objection to me changing the column headers to the actual parties' names? ReferenceHunter (talk) 12:04, 31 October 2018 (UTC)

Simplicity and allowing the columns' width to be smaller is a step forward on the current use of abbreviations. Specially when considering that the table is sortable; i.e. the width will increase noticeably if the whole names are spelled out. Impru20^talk 12:07, 31 October 2018 (UTC)

The Table looks off

Some Thing about the table looks weird. For one thing the bottom column really should not be there and there is something strange to the side of the table. here is my proposal: JDuggan101 (talk) 13:34, 28 October 2017 (UTC)

Individual polls

Poll	Date	Sample size	National	Labour	Green	NZ First	ACT	Māori	TOP	Lead
26 Oct 2017 – Jacinda Ardern is sworn in as Prime Minister of New Zealand.
Roy Morgan Research	2–15 Oct 2017	894	46	31	6.5	11	0.5	1.5	2	15
2017 general election	23 Sep 2017	–	44.4	36.9	7.2	6.3	0.5	1.2	2.4	7.5

Anything in particular? I think the table looks fine as is. Clesam11 (talk) 01:17, 29 October 2017 (UTC)

@Clesam11: Thanks for the response. If you don't know me, I like things to be constant so that's the reason why I like it to be the same as the 2017 one. Also I've added seat forecasts already, at least I had a "note" at the bottom, if that's fine. Typhoon2013 (talk) 10:57, 29 October 2017 (UTC)

Although actually, I believe it's better to add the 2017 election result column at the start/top rather than last/bottom.Typhoon2013 (talk) 11:02, 29 October 2017 (UTC)

@Typhoon2013: Having the election result at the top would break the reverse chronology. However, i'm not sure who decided to make the table reverse chronological, I just went along with it because I don't mind either way. If we are going to change that it's probably best to do it early while its still easy, so if anyone wants to start a discussion about changing that, now's the best time! Thanks for adding the seat projections by the way. Clesam11 (talk) 21:30, 29 October 2017 (UTC)

The suggestion we use reverse chronological order was made at Talk:Opinion polling for the New Zealand general election, 2017#OK, I know it's antipodean and many things are reversed down there..., but it's certainly open for debate here.-gadfium 07:37, 30 October 2017 (UTC)

@Gadfium: @Clesam11: Ok but it maybe just me, but I really like it in a chronological order/top to bottom. It's just better and that's the way it is and also the 2017 article is from top to bottom and why not that. I am also talking in a larger perspective and not just NZ referendums but also the US ones (if that is also from bottom to top, then there may be other ones too). Like is there a rule in this project or it is just how people started it? Typhoon2013 (talk) 09:21, 30 October 2017 (UTC)

Election results, 3 digits vs 4 digits

For simplicity, I believe the 3 digit numbers provided by the electoral commission (I.e. 44.4, 36.9, 7.2) for the election results would be better than the currently displayed 4 digit ones. It makes calculating things like the leads a lot easier, as well as fitting better with the 2-3 digit numbers we receive in most poll results. Any objections to this? Clesam11 (talk) 02:14, 29 October 2017 (UTC)

Agreed. 3-digits is easier to read and much simpler than 4. Typhoon2013 (talk) 10:58, 29 October 2017 (UTC)

Forecasts

I personally don't think that the "forecast" by Roy Morgan should be counted as such. From what I saw on their site, they haven't actually made a seat calculation; rather someone else has applied the Sainte-Lague formula to the poll numbers to get those seat numbers. In my opinion, only forecasts that take many/all polls into account should count, like what the NZ Herald had for the 2017 election, and sites such as Peter's Stats Stuff and NZ Election Prediction. None of these sites have yet started projections for the 2017-20 term, and probably won't for some time.

Given that it is a proportional system, listing each company as a separate "forecast" is merely equivalent to listing prominently each company's latest result, but out of 120 seats not 100%. 122.58.23.26 (talk) 08:57, 31 October 2017 (UTC)

There is a note below the table that makes it clear that the seats were calculated using the electoral commission's MMP seat allocation calculator, so it's not really misleading. Either way, whether its an average across many polls or a poll on its own, it can still be wildly inaccurate to the final results. Whether we have seat projections or not, there's always a chance they'll be completely wrong, that's just the nature of projections and polls in general. I think they should stay. Clesam11 (talk) 02:39, 1 November 2017 (UTC)

Sortable table, solution to ordering debate

As we all know, there's been some debate above (and on the 2017 polling talk page) about anti-chronological ordering of polling. I've decided to test a solution which makes it possible to sort the polls in both chronological and anti-chronolgical order. The only real downside to this is that the way we previously displayed the dates (5–10 Nov, 10 Oct, etc) would cause it to sort the polls incorrectly. As a solution, with some inspiration from the Canadian polling page, I have made it so only the last day of the survey is displayed, as well as displaying the month names in full. This was the only way I could make the ordering work properly unfortunately. This solution may not be ideal for everyone, so please raise any concerns you have here. Clesam11 (talk) 09:04, 13 November 2017 (UTC)

Use "data-sort-value" for date ranges. Mélencron (talk) 13:42, 13 November 2017 (UTC)

Wow, nice one! All the drama around ordering can hopefully end now. Thanks! Clesam11 (talk) 22:08, 13 November 2017 (UTC)

"Lead" column misleading

Someone has added a "lead" column to the table which didn't exist on Opinion polling for the New Zealand general election, 2017. It counts the highest-polling single party's percentage lead over the second-highest-polling single party. This type of figure makes sense for first-past-the-post races such as the race for an individual seat, but it is misleading for indicating who's leading the race for a majority in Parliament, which is not determined by "highest result by a single entrant" like individual seat races, but by "whichever party or collection of parties can command a majority of the 120ish seats in Parliament". It's particularly misleading when NZ currently has the largest single party not part of the government for the first time since 1935 (or since 1984 if you're talking about most votes not most seats) - so it's exactly the wrong timing to introduce such a change if you're interested in accurately reflecting who is "winning" the race to be elected as the government (though it's exactly the right timing if you want to cast doubt on the legitimacy of the current government). It would make more sense to indicate whether the bloc of government parties or the bloc of opposition parties are leading, though even this would be misleading, firstly because NZ First could shift its allegiance, and secondly because it's seat majority that matters, not vote majority. The strength of various potential governing blocs is better indicated by the "Overall result (majority)" row on the Forecasts table. So I suggest we delete the new "lead" column. Elcalebo (talk) 01:25, 20 November 2017 (UTC)

What are your thoughts, @Clesam11: @Typhoon2013: @Lcmortensen: @Mélencron: @Sleepingstar: @Gadfium: ? I realise "lead" columns are common on similar polling pages (including in other MMP systems like Germany) but I still stand by my arguments above. Elcalebo (talk) 21:59, 26 November 2017 (UTC)

I completely see where you're coming from. However, I don't really view the lead column as problematic. Most readers are fully aware of New Zealand's MMP system, and those who aren't, definitely shouldn't be relying on a page exclusively for opinion polls to understand the ins and outs of any political system. We also have a seat projections/forecasts section which briefly describes our electoral system and how governments can be formed.

Of course, a lead column isn't at all necessary, but as you mentioned, it's certainly a common practice. I'm mostly in favour of it staying on the basis that it doesn't hurt anything, and because I like it – the latter obviously being subjective. We could always add a disclaimer above the table that mentions the lead column or briefly describes our electoral system, if that makes it less of an issue. Clesam11 (talk) 01:40, 27 November 2017 (UTC)

@Elcalebo: As per @Clesam11:, it is not "problematic". Yes, some other articles have a lead column while other do not. But imo, for this case, I do not see the case to add it because NZ uses MMP and any party can form a government (eg: NAT on lead, LAB likely to form gov). Plus, we have a seat table forecast at the bottom which explains more. But I really don't mind. Typhoon2013 (talk) 10:42, 27 November 2017 (UTC)

OK, I'll bow to the majority. Also, thinking on it further, it's not too much of a problem. Everyone (at least everyone in NZ) knows the govt is a Labour-led one despite National having the lead on the 2017 election result, so I suppose nobody will be confused. It will be interesting to see whether National stays ahead as the biggest single party (and, if so, whether its lead increases or decreases), and the Lead column will enable that. This isn't a very good measure of how well the government as a whole is doing vis-a-vis the opposition as a whole (because it doesn't include support partners), but the seat projection section can provide that info (albeit only the most recent info rather than tracking it over time). Elcalebo (talk) 23:10, 28 November 2017 (UTC)

Can we please discuss if the "lead" column is misleading, and not if it is "problematic". The inclusion of an analysis column, in a table that otherwise only contains data should be based on the merit/validity of the analysis. For an MMP system the concept of "lead" is misleading as it infers that there is a race with the winner determined by the highest number of votes. As illustrated by the last NZ election, this is not true. The column could be re-named, i.e. highest lead over second place, but the presence of this column is confusing at best and misleading at worst. Usefulmonuments (talk) 23:17, 10 October 2020 (UTC)

New format?

Hi @Clesam11: and saw you edits just before leaving this message. Sadly for me I oppose your new format, as it is a bit larger than the previous one and doesn't have the border lines. Although what are your thoughts about changing the format and/or making it better? Typhoon2013 (talk) 23:55, 23 November 2017 (UTC)

@Typhoon2013: That's fine, it is a bit of change in many ways and I can see why some people might not like it so much. Borders can be easily added if you'd like. I believe we can also change the colours of the borders as well if we wanted. Size/width can also be adjusted.

Option 1

Date	Polling organisation	Sample size	NAT	LAB	NZF	GRN	ACT	TOP	MRI	Lead
30 Oct – 12 Nov 2017	Roy Morgan Research	887	40.5	39.5	5	10	0.5	2	1.5	1
26 Oct 2017	Jacinda Ardern is sworn in as Prime Minister of New Zealand
2–15 Oct 2017	Roy Morgan Research	894	46	31	6.5	11	0.5	2	1.5	15
23 Sep 2017	2017 election result	N/A	44.4	36.9	7.2	6.3	0.5	2.4	1.2	7.6

Option 2

Date	Polling organisation	Sample size	NAT	LAB	NZF	GRN	ACT	TOP	MRI	Lead
30 Oct – 12 Nov 2017	Roy Morgan Research	887	40.5	39.5	5	10	0.5	2	1.5	1
26 Oct 2017	Jacinda Ardern is sworn in as Prime Minister of New Zealand
2–15 Oct 2017	Roy Morgan Research	894	46	31	6.5	11	0.5	2	1.5	15
23 Sep 2017	2017 election result	N/A	44.4	36.9	7.2	6.3	0.5	2.4	1.2	7.6

Maybe something like one of these? Clesam11 (talk) 00:30, 24 November 2017 (UTC)

@Clesam11: Much better, though I would like to add that I didn't mention earlier is that the colours for this is fainter than the previous, usual format. But for me, I always like things succinct/short should an article is most likely to be pretty big. We know for sure that from now until the next election, this table will grow even further so yeah. Typhoon2013 (talk) 09:28, 24 November 2017 (UTC)

Yeah, the fainter colours were only so I wouldn't have to vary between different colour text (to not clash with the background colours) and still have this format. I opted in for the same colours we use to shade the background of poll results because they work well with text appearing on top of them. When you say you'd like things short, are you suggesting abbreviated party names? For example (NAT, LAB, NZF, GRN, etc.) instead of the full party names? That could be useful if/when new parties need to be added, so I wouldn't oppose a change like that.

I'm going to trial a new table and see how things go. Most of these changes are aesthetic, so i'm fine with going back to the status quo if people aren't fond of the change. Clesam11 (talk) 10:16, 24 November 2017 (UTC)

@Clesam11: The new format looks really clean and tidy, great job. However, more of a pedantic nitpicky observation - just for consistency's sake between the polling tables and the forecast tables, would there be any chance you could consider making the top line of the headings of the polling tables "background:#EAECF0" as well? It would look really nice if the formatting of the tables were congruent throughout the article. (Albiet I do acknolwedge that poor @Limegreen: may have to, yet again, revisit that R code for the graph... so, ideally, it would be nice to find the right format for the table and just stick with it for the next 3 years ;) --Sleepingstar (talk) 20:34, 24 November 2017 (UTC)

I don't think the cells with the party names should have a background color – not accessible and also breaks sortability (for those who like to have fun with those columns like I do). Mélencron (talk) 21:04, 24 November 2017 (UTC)

@Sleepingstar: Colouring the banner was actually something I originally tried to do when tinkering around with this new format yesterday. Unfortunately, colouring the banners also results in breaking sortability – which is something a lot of people were keen on because of the change to reverse chronological ordering. Would be cool if we found a way to fix that.

When it comes to the graphs – thankfully, there won't be enough polls to really require one for a couple more months, so hopefully that's enough time to prepare! Clesam11 (talk) 22:53, 24 November 2017 (UTC)

@Clesam11: Wow and there is a difference now when you add borders, hehe. Now I'm keen to stand as where it is. Although you made a second format at the top there ^^, kind of combining both the new format and old format together, with the colours slightly bolder. I'm keen with either one, but I think I'm a bit leaning on the other one, just to make the colours clearer. :) Typhoon2013 (talk) 10:47, 25 November 2017 (UTC)

@Typhoon2013: I personally preferred the table with less borders as it matched better with the forecasts table, as well as being more aesthetically pleasing to look at (for me). However i'm happy to settle on this table. Although, i'd like to bring over the abbreviated party names again (NAT, LAB, NZF, etc.) – it allows for plenty of room to accommodate parties that may gain support over the election cycle. Clesam11 (talk) 11:29, 25 November 2017 (UTC)

@Clesam11: Well lol from my perspective I like the other one. At least they're both pretty similar. We better decide on something or a table format that would be good for the coming years and it will be pretty tough should we change the format by 2020, with many polls. Again, I'm fine with either one but I prefer the other one a little better. Typhoon2013 (talk) 10:38, 27 November 2017 (UTC)

Polling organisation v Pollster

@Lcmortensen: Uh, what does "Not good England" mean? Could you rephrase that please? Clesam11 (talk) 01:41, 24 November 2017 (UTC)

It just seems to be a bit slang to refer to "pollster". Besides, a "pollster" is a person, not an organisation, that conducts polls. Lcmortensen (mailbox) 01:44, 24 November 2017 (UTC)

Ah okay, thanks for clarifying that. Nothing particularly wrong with how it is now, it just seemed like it could be shorter. Clesam11 (talk) 03:47, 24 November 2017 (UTC)

Graph

@Clesam11 and Limegreen: Thanks Clesam for the graph, even though I did not expect it to be created until at least by the end of the year, but ok. Though I prefer the same format of the graph as previous ones as what Limegreen did. Did you use 'GraphPad Prism'? Because I prefer to use that, which is the same as the other graphs. Thanks. Typhoon2013 (talk) 03:55, 16 January 2018 (UTC)

@Typhoon2013: Yeah, I just figured a graph would be nice to have in the meantime. I definitely prefer Limegreen's graph that's generated in R with a smoother. Mine is created using Excel, but the options can be a bit limiting. There are options for smoothing in there, but it results in some pretty ugly trend lines due to a lack of anti-aliasing. Hopefully someone will get that R code working here too soon, but until then I thought we could settle for something else. Clesam11 (talk) 05:20, 16 January 2018 (UTC)

@Clesam11: I mean it's alright really, especially there are only like 3 or 4 polls so far. We could change it soon when we have more polls, plus, the next election won't happen in two years. Limegreen can change it to the usual format when he sees this/or has the time but I'm glad you can help with graphs. I hope you can help even further. Moreover my GraphPad trial session had ended a while ago but I can help update the table. :) Typhoon2013 (talk) 07:13, 16 January 2018 (UTC)

Been a bit preoccupied, plus there have been remarkably few polls so far. I would be tempted to make a few tweaks to the last graph format, though my hope is that I could make the R script a bit easier to work with so that it doesn't just need one person to maintain it. Limegreen (talk) 01:21, 2 February 2018 (UTC)

Are Roy Morgan out?

Just to note that Roy Morgan seem to have stopped their monthly political polling. This is going to make this cycle super-sparse in terms of polls. It seems surprising that no-one is polling following the change of leader for National! (either that, or it isn't out yet). Limegreen (talk) 00:10, 21 March 2018 (UTC)

Yup, it seems Roy Morgan is out. This is definitely the biggest shortage of opinion polls I've ever seen. We currently have only two semi-regular public opinion polls - that's nuts! Here's hoping that more organisations commission polls. Clesam11 (talk) 07:33, 22 March 2018 (UTC)

Roy Morgan has begun federal polling in Australia again. It seems they didn't want to poll both Australia and NZ at the same time. More interestingly though is that Iceland, a country with a population lesser than Christchurch, has had more post-election polls following an election that took place a month later than NZ's election. Insane! Clesam11 (talk) 03:57, 7 April 2018 (UTC)

Prizes, Horizon Research and Reid Research

Our policy previously on Horizon Research has been that they encourage self-selection through offering prizes and vouchers upon signing up, and we therefor should exclude them. I have previously agreed with this stance, however recently I was browsing and noticed that Reid Research, the company that Newshub commissions for its polling, is participating in very similar behavior by offering prizes and vouchers upon singing up as well. As a result, the case for excluding Horizon polls has become a lot flimsier in my opinion.

Horizon still presumably selects a sample representative of the New Zealand electorate when conducting their polls, so at this point they're employing the same methods as Reid Research as far as I can tell. The only way to justify excluding Horizon at this point would be to also exclude Reid Research. This would put us down to only one public opinion poll in all of New Zealand. So, at this point i'd like to start including Horizon polls in this article as well, but if anyone has any further reason for why we should exclude them it'd be great to hear it! Clesam11 (talk) 10:31, 12 April 2018 (UTC)

I agree that Horizon research should be included. JDuggan101 (talk) 10:41, 12 April 2018 (UTC)

Panel polling does seem on the rise, and especially with Roy Morgan out of the game, there is relatively little other info to go on. --Limegreen (talk) 01:11, 17 May 2018 (UTC)

I concur with the above – as a side note, when Harris Interactive started polling in France in March 2011 (dropping a shock poll showing Le Pen leading and the right eliminated), they used a similar technique to recruit their initial panel by offering a prize of ~€7,000. To the best of my knowledge, all polls in France are carried out in this manner (compensation of less than €1 per respondent per poll). Mélencron (talk) 03:33, 17 May 2018 (UTC)

Morning, @Clesam11 (and @JDuggan101, @Limegreen, @Mélencron , and @Nixinova)

I've just been looking into Horizon's rogue polling numbers. They're way out of alignment with every other pollster (Reid, Kantar, Curia, Roy Morgan, Talbot Mills). They're using some very dodgy methodology. Source: https://www.horizonpoll.co.nz/page/668/labour-and-

1. Horizon are using an internal database and/or a self-selecting online poll NOT randomised calling like other pollsters "The onlne survey has 1,563 respondents."
2. Even crazier, Horizon are weighting polling results to the 2020 election results!! This is madness for a school student, let alone a research company. This is why they are severely underestimating the National Party vote. "Results are weighted on, age, personal income, party voted for in 2020, employment status, region and ethnicity."

They are the laughing stock of the polling community because of the points above.

As such, they have to be removed from this page, right(?) and a note added why they are disqualifed.

Thoughts?

Geoff Geoffnz1 (talk) 18:57, 28 May 2023 (UTC)

Forecasts

Presently the forecasts section is mostly restating the latest polls in terms of seats instead of the percentages. That seems a bit pointless, and the word "forecast" implies a deeper analysis than just one poll. So I suggest that "forecasts" based on just one poll are removed, and only analyses based on multiple polls such as Stuff's poll of polls, and models such as that run by the Herald in 2017, and the model of Peter's Stats Stuff (see Opinion_polling_for_the_New_Zealand_general_election,_2017#Forecasts. I'll also disclose a COI as I run the NZ Election Prediction site that is mentioned alongside Peter's Stat's Stuff on that page as source number 151. AV85647 (talk) 02:55, 21 October 2018 (UTC)

I also wondered whether the forecasts presented should be within the range of perhaps the last 3 months, for example the Stuff-YouGov poll seems to be a bit outdated from December 2019, at a time when the mood of the nation in polling seems to differ signficantly from today. I also wondered whether we might be able to use other models, as well, such as those presented on https://politi.kiwi/? Sleepingstar (talk) 20:49, 26 September 2020 (UTC)

I disagree with the removals at Template: 2020 NZ election forecasts, even though they're old they're a useful supplement to the party vote percentages, as seat allocations are not immediately translatable to seats. We should be adding more, not less.  Nixinova   T   C   06:20, 27 September 2020 (UTC)

The old graphs

Because of the formatting changes to the table, and the lack of polling, I didn't really get round to adapting the script until tonight, but wonder if anyone objects going back to the previous style of graphs

I need to get rid of the standard errors, as they are not computing. And could go back to the more intense colours. Thoughts? Limegreen (talk) 23:52, 11 February 2019 (UTC)

There are also the new graphs, which I'm making and updating frequently:

The minor parties graph has the good point of being centered on the 5% threshold, while it's hard to see on the old graphs that NZF is struggling to reach that threshold. Thoughts? Julio974◆ (Talk-Contribs) 08:53, 9 April 2020 (UTC)

Thanks for your work! The new graph are definitively better. I think it's better if possible to have a single graph. The threshold could be shown by a slightly differently colored line like on the model used for the israeli elections. As for the parties with polling results so small they bungle together as a flat line around 1 %, I would tend to think these results are too meager in the first place to show them on the poll graph. Only ACT can really be kept. Just my two cents, though.--Aréat (talk) 15:22, 9 April 2020 (UTC)

Great work on the graphs. But, I think NZ First and the Greens should be removed from the minor parties graph as they received above 5% in the last election. Don't really see any point in showing them in two graphs. But that's just me! Clesam11 (talk) 04:23, 28 April 2020 (UTC)

Since all the small parties are in the upper graph as well, this would only make sense if all the parties are only shown in either graph. However, I don't mind the current state which makes the second graph a zoomed in version of the first, that is easy to interpret. The only thing that I would recommend is to remove National and Labour from the legend of the second graph to avoid confusion.. --Gbuvn (talk) 06:57, 28 April 2020 (UTC)

The distinction between the two graphs made sense for previous electoral cycles (and *may* still make sense for this electoral cycle), when the Greens and NZ First were both polling in double figures. They are now, much more clearly closer to the 5% threshold. Similarly, other minor parties (TOP, Maori, Conservative) were consistently higher than the ACT party. So I guess it is a question about whether people would want to compare across elections, as the previous style was consistent through 2017, 2014, 2011 and 2008. This isn't an argument against updating; there are definitely things I would change. Limegreen (talk) 18:31, 1 May 2020 (UTC)

So the 'classic' style of graph is back (without consulatation, sorry). I'm really happy to make modifications, but it seems obvious (from eg today) that having the person making the graphs is following the New Zealand political news cycle. I saw the news this morning, and made the graph. Limegreen (talk) 09:22, 18 May 2020 (UTC)

Do you think it is worth shifting NZF and Green to the lower graph with ACT and the others? It is hard to gain an accurate understanding of their numbers now they are both so low. I understand this may change again, which could be fixed by shifting the axes a bit. Thoughts? (Also sidenote: is there any way you can export graphs in a slightly higher resolution? Not urgent but it would look a bit more aesthetic). Thanks, WBPchur ^{💬●✒️●💛} 10:05, 21 May 2020 (UTC)

I've reproduced the new graph style in ggplot, so I can update the charts as well as soon as a new poll is released (whilst keeping the style somewhat consistent). --Canley (talk) 00:20, 27 July 2020 (UTC)

Preferred prime minister

The news these days is all about how Bridges' PPM polling is lower, than say Collins. Suggest we add her to the PPM list? — Preceding unsigned comment added by 114.23.233.228 (talk) 09:51, 18 February 2019 (UTC)

New pollster: Stuff/YouGov

There's previously been an opinion that online-only panel polls shouldn't be included here. On the other hand, internationally, YouGov have a pretty good reputation as a pollster, so I'd be pretty inclined to include them. Any thoughts Limegreen (talk) 13:09, 29 November 2019 (UTC)

Fully support including them as well. There is a disclaimer above the tables explaining that the way pollsters collect their data varies. Clesam11 (talk) 04:20, 28 April 2020 (UTC)

Timeline of events

I just want to point this out since I believe there is a small edit war on regarding the timeline of events in the opinion polls chart from COVID-19 updates. This edit summary made by Impru20 is something I agree, and so should a lot of users too. It is better to summarise all of this into one line instead of making a huge thread regarding COVID-19 updates since this is the wrong article for it. Thanks and stay safe. :) Typhoon2013 (talk) 09:28, 27 April 2020 (UTC)

Just to note it is confirmed that Level 2 will be out in a few days so that should be worth mentioning that restrictions have eased, and have done so. Also I want to mention that I changed the date to 26 March instead of the 25th because it was just a one minute difference (11:59pm) for the official announcement. Typhoon2013 (talk) 04:57, 11 May 2020 (UTC)

UMR Polling

Recently new numbers from UMR polls have come out. Is it worth adding these to this page? Following are links to sources containing the polling: NZ Herald, Star News, The Guardian and The BFD. These numbers are technically leaked, but I believe they should still be published here with comments regarding their potential lack of reliability and accuracy. All the sources do corroborate. Cheers, WBPchur ^{💬●✒️●💛} 08:09, 1 May 2020 (UTC).

Just found where they were added only a minute later. My apologies - was confused by graphs :) WBPchur ^{💬●✒️●💛} 08:13, 1 May 2020 (UTC)

@WBPchur: Already added. However I do recall in the sources stating that National did not leak their own (Curia) polls since February. However someone added a Twitter source of another leaked Curia poll on April in the table. Should we leave the April Curia poll in or take it out? Typhoon2013 (talk) 01:56, 2 May 2020 (UTC)

There was a consensus in one of the previous electoral cycles that because they tend to be selectively leaked, they wouldn't be in the main table (which produces the graphs). That is how they ended up in their own section. Limegreen (talk) 23:48, 11 May 2020 (UTC)

Stats table is missing "Don't know" or "Refused to answer" responses

In Special:PermanentLink/958833652, by Geoffnz1, the user added a section that they obviously meant to post here. I have copied this section’s heading from that edit; that edit’s content is below:

This article has an error in the tables. It fails to include the number of voters who were "Don't know" or "Refused to answer". For the latest 1 News poll this was a massive 16%, so must be shown in the results table.

However, I'm a new Wikipedia user and don't want to mess it up. I also don't have the time to do it.

Cheers

Geoff — Preceding unsigned comment added by Brianjd (talk • contribs) 07:02, 27 May 2020 (UTC)

Some polling companies don't report their don't know/refused to answer, and the results are presented as a proportion of decided voters. Ie, they subtract the whatever percent undecided, and present the remaining to total to 100. Coming from a scientific background, this seems like an odd approach, but it is how it has always been, seemingly, and seems to work OK. Perhaps the people that "Don't know" don't vote, and the people that refuse aren't too systematically different.

I wouldn't be averse to seeing it reported, but altering the table needs to be done carefully, as otherwise it will break the graphing scritps. Limegreen (talk) 13:19, 15 July 2020 (UTC)

Thanks for sharing this, @Brianjd, and replying, @Limegreen.

I'm back analysing all the polling data for www.theFacts.nz, so have a renewed interest in the data on this page.

"Don't know" and "Refused" tell an important part of the voting story as they include a proportion of swing voters. As such, I believe this woule be a valuable addition.

Who is in charge in making these decisions on this page? Geoffnz1 (talk) 18:49, 28 May 2023 (UTC)

Yabble

@Gwuby: has added a poll taken by Yabble to the preferred Prime Minister section. We haven't used any polls by Yabble in the past, and I suspect that their polls do not adhering to the standards of the more established polling organisations. It looks to me from a quick view of their website that the people who vote in their polls are self-selected. If this is correct, I think we should remove this poll from the article.-gadfium 21:30, 15 July 2020 (UTC)

I've rung their 0800 number and asked. While I was told that they can ensure that the sample represents a cross section of society, they are not in a position to confirm what they have done in this particular case "as it depends on the brief of the client", and they will not reveal what the brief was. So unless we hear via Newshup or The Project that their brief was for a poll that is comparable to the standards of the more established polling organisations, I suggest that we skip this one. I shall remove it. Happy for further discussion to occur here. Schwede66 23:12, 15 July 2020 (UTC)

Include Advance NZ?

They've now polled 1%, the first minor party not currently included to have done so as far as I'm aware.--Pokelova (talk) 06:38, 22 September 2020 (UTC)

If they're included in any poll results then yes they can be; I didn't see them in the 1 News poll.  Nixinova   T   C   21:20, 22 September 2020 (UTC)

Yep. If poll results get reported, they can be included. Schwede66 22:25, 22 September 2020 (UTC)

Add a new section for seat allocations

The contents of Template: 2020 NZ election forecasts as been nuked by people (at split discussions at #Forecasts 2 and Template talk:2020 NZ election forecasts) who don't want there to be too many "forecasts". I strongly disagree with this nuking as I won't care about the seat forecasts directly before the election, it would be more interesting to see the changes throughout the parliamentary term. This would be a useful supplement to the party vote table.  Nixinova   T   C   07:17, 28 September 2020 (UTC)

Table order

The tables are only listed in referse chronological order before the election takes place. Once the election is done, it's changed to basic chronological order. See 2017, 2014 etc. This wasn't "unilateral", it's a thing that happens to these opinion polling pages.  Nixinova   T   C   20:12, 17 October 2020 (UTC)

You're incorrect on that one. NZ polling pages used to list in chronological order, even prior to elections. Those other pages haven't been reversed primarily due to no one having gotten around to changing it. I recall it being agreed that because other pages generally list polls in reverse chronological order that NZ pages should follow suit - for consistency across Wikipedia. But I'm sure we'd be happy to concede to the majority if they voice their preference for chronological ordering here. Clesam11 (talk) 04:27, 18 October 2020 (UTC)

We should have the latest polls on top. Onetwothreeip (talk) 04:55, 18 October 2020 (UTC)

Colmar Brunton 404

Half the 1 News poll links 404 (everything before mid 2019, I think).  Nixinova   T   C   07:41, 17 September 2021 (UTC)