Jump to content

Talk:Opinion polling for the 2008 New Zealand general election

Page contents not supported in other languages.
From Wikipedia, the free encyclopedia

Kudos for Kripto

[edit]

Onya Kripto, some good work here. --Lholden 22:39, 12 July 2007 (UTC)[reply]

The Kudos should be extended to all who have contributed to this page - we're getting cited in the blogosphere now too!! eg Dimpost, The Standard. Congratulations on a job well done. --Trevva (talk) 11:57, 27 October 2008 (UTC)[reply]

Well, mainly your excellent graph :-)--Lholden (talk) 05:44, 28 October 2008 (UTC)[reply]
Graphs are only pretty pictures - its the data they contain that counts! --Trevva (talk) 07:15, 28 October 2008 (UTC)[reply]

Poll dates

[edit]

Are the dates shown here supposed to be date released or date surveyed? Kelvinc 02:57, 10 October 2007 (UTC)[reply]

I've been using date released.-gadfium 05:24, 10 October 2007 (UTC)[reply]
Erm okay. Some of the older ones appear to be date surveyed (esp. the ones with a range). I would prefer (last) date surveyed, except that information is not always available, so for consistency it's probably best to go with date released. However, right now I think we kind of switched systems at some point, which needs to be fixed. Kelvinc 05:41, 12 October 2007 (UTC)[reply]
I'm sorry, you're right. The dates are (supposed to be) date surveyed. The date released generally appears as part of the reference.-gadfium 08:04, 12 October 2007 (UTC)[reply]
It's okay: I was somewhat confused as well. I've put in surveyed dates for TV3-TNS (the newest PDF shows older dates as well as some older results that were missing from the table), Roy Morgan, and One News (Colman says that they survey the Monday to Thursday before release so I just went with that), but similar information seems to be harder to find for the DigiPolls. Honestly, I think it's scandalous that polling firms don't release all their findings publicly within about a week of the initial release of the polls, as well as the questions that were used: I think that it casts their methodology into question. Kelvinc 03:53, 14 October 2007 (UTC)[reply]
All but one of the DigiPolls have dates surveyed, so there's only one DigiPoll and the Fairfax poll without the dates surveyed. I've added that fact, as well as how I got the One News dates, on a set of notes on the bottom. Kelvinc 05:27, 15 October 2007 (UTC)[reply]
Since different polls are conducted over different time periods, I think it would be appropriate to list the median dates in the polling period, rather than the end date of survey. For instance, if a Morgan poll is conducted from 1-13 October, the the listed date should be 7 October. For a Colmar poll conducted 1-5 October, the the median date would be 3 October. (Randomkiwi (talk) 11:12, 24 October 2008 (UTC))[reply]
The script I wrote to make the graphs uses the average date, so having a range doesn't effect it there, per se. The problem is that of tie in - it would be a big job to change it now! Maybe for the successor page, after the election? --Trevva (talk) 21:45, 25 October 2008 (UTC)[reply]

Fair use rationale for Image:Pollcap.png

[edit]

Image:Pollcap.png is being used on this article. I notice the image page specifies that the image is being used under fair use but there is no explanation or rationale as to why its use in this Wikipedia article constitutes fair use. In addition to the boilerplate fair use template, you must also write out on the image description page a specific explanation or rationale for why using this image in each article is consistent with fair use.

Please go to the image description page and edit it to include a fair use rationale. Using one of the templates at Wikipedia:Fair use rationale guideline is an easy way to insure that your image is in compliance with Wikipedia policy, but remember that you must complete the template. Do not simply insert a blank template on an image page.

If there is other fair use media, consider checking that you have specified the fair use rationale on the other images used on this page. Note that any fair use images uploaded after 4 May, 2006, and lacking such an explanation will be deleted one week after they have been uploaded, as described on criteria for speedy deletion. If you have any questions please ask them at the Media copyright questions page. Thank you.

BetacommandBot (talk) 20:16, 5 December 2007 (UTC)[reply]

Graphs needed?

[edit]

Lot of work went into this page - good effort all you editors. The page could do with a graph of the averaged results for ease of interpretation an the background colour of the notable events could do with being a little more subtle. -- Alan Liefting- (talk) - 21:24, 16 March 2008 (UTC)[reply]

There was a graph, but the editor who produced it left Wikipedia. I asked for a copy of the spreadsheet which produced it but too late, they were gone. The graph then became too out of date (and it had some display problems) and was removed from the article. It would be a substantial amount of work to recreate it, but worth while.-gadfium 22:29, 16 March 2008 (UTC)[reply]
I'll have a look at this during the weekend, unless someone beats me to it. -- Avenue (talk) 03:39, 17 March 2008 (UTC)[reply]
Done. -- Alan Liefting- (talk) - 20:15, 23 March 2008 (UTC)[reply]
Thanks, Alan. I had almost finished a graph too, showing smoothed party support curves; here it is. -- Avenue (talk) 02:24, 24 March 2008 (UTC)[reply]

The effort that goes into both graphs is impressive. What I'd like to see in the article draws from the strength of both; the detailed results and smoothed average of Avenue's graph, but the simplification by showing only the parties which have gained 5% support at some point since the last election in Alan's graph, and showing the support percentages every 5%, since the 5% threshold is so important.-gadfium 04:50, 24 March 2008 (UTC)[reply]

Why, thanks for that commendation gadfium! Yes both graphs have merit. The parties with the low polling end up in a clutter at the bottom. Avenue, perhaps you could rejig your graph with only those that gained the 5% threshold? -- Alan Liefting- (talk) - 05:51, 24 March 2008 (UTC)[reply]
Sure, I've replaced it with a simplified version. Let me know if there are any other improvements you'd like. -- Avenue (talk) 11:24, 24 March 2008 (UTC)[reply]
That's a great graph. I'll have to play with R and ggplot2 next time I need a graph for anything.-gadfium 18:19, 24 March 2008 (UTC)[reply]
Sure is a great graph. I have used it in the article in place of my simple version. -- Alan Liefting- (talk) - 20:27, 24 March 2008 (UTC)[reply]
Thanks. I'm still new to it, but I find ggplot2 quite an elegant plotting package. The documentation seems a bit spotty once you get beyond the basics, though. -- Avenue (talk) 12:22, 25 March 2008 (UTC)[reply]

Can we have a graph for the preferred Prime Minister too?-gadfium 21:56, 24 March 2008 (UTC)[reply]

Done. I've left out the confidence bands around the curves because they were quite wide and seemed too distracting, but other than that it's similar to the last one. -- Avenue (talk) 12:22, 25 March 2008 (UTC)[reply]
Thanks, very nice.-gadfium 18:09, 25 March 2008 (UTC)[reply]

Preferred PM for June Herald-DigiPoll

[edit]

I put in polling data for Morgan, Colmar-Brunton, Nielsen, and DigiPoll. Unfortunately, I was not able to find preferred PM information for the June Herald-DigiPoll on the Herald's site or elsewhere. Could someone please help? It may require looking at the print addition of the New Zealand Herald (it should be in the 28 June 2008 edition). -Rrius (talk) 12:51, 14 September 2008 (UTC)[reply]

Removal of %%%%%%%%

[edit]

I've stripped 'redundant' % symbols as a soup of 1100+ %%%%%%%%%%%%% make reading data difficult and context is understood. Fanx (talk) 16:47, 6 October 2008 (UTC)[reply]

speaking of comprehension, I find if you scroll down enough, it's sometimes hard to get a handle on whose numbers you're reading. Any ideas on how to fix that? plan 8 (talk) 18:29, 6 October 2008 (UTC)[reply]
fixed - by adding row of Party/Preferred Leader names at more frequent intervals. Fanx (talk) 20:09, 8 October 2008 (UTC)[reply]

New Summary Graph

[edit]

I've remade the main graph by parsing the html table directly in R, and then plotting it from there. It's not as pretty as the original, but its there anyway. Hope you like it! --Trevva (talk) 23:10, 3 October 2008 (UTC)[reply]

Thank you for updating the opinion poll graph. I've been meaning for some time to learn an appropriate statistical package so I could do so, but you've saved me the trouble. I haven't seen such a graph in any other source; this is something which makes Wikipedia a particularly valuable source for election data.-gadfium 05:30, 4 October 2008 (UTC)[reply]

You're welcome - I felt the same way, and being overseas (Denmark), wanted to know how it all stands at the moment. I've tried to post the source code with the document, but its fairly complex and completely hoses the wiki text - if you have any suggestions how to get around this, I'd love to hear them. It works by reading the HTML from the webpage and parsing in, so the idea was that it should be easy for anyone to run. --Trevva (talk) 23:48, 4 October 2008 (UTC)[reply]

If anyone has any suggestions about what else they'd like to see in the graph, I'm open to them - just post them here. I haven't made a "Preferred Prime Minister" one yet, but its relatively straightforward if anyone has a burning desire... --Trevva (talk) 19:18, 6 October 2008 (UTC)[reply]

The graph looks good - very well presented. The final percentages is a nice touch. It would be good to do the preferred PM graph as well. The data is there, it would be nice to see the changes graphically. There seems to be an interest in it since the media actually bother to get the polling data. The interest in the article has peaked at 134 hits per day this month and peaked at 194 last month. There will be an increasing interest as the election approaches. -- Alan Liefting (talk) - 08:51, 8 October 2008 (UTC)[reply]
I bet I make up a good portion of those! -Rrius (talk) 08:53, 8 October 2008 (UTC)[reply]
After keeping an eye on the stats for a new page that I was regularly returning to it seems the traffic shows unique user hits. -- Alan Liefting (talk) - 11:45, 8 October 2008 (UTC)[reply]
I was attempting to be funny. -Rrius (talk) 11:54, 8 October 2008 (UTC)[reply]

Rrius - you got a smile from me, anyway. I've just put up the Preferred Prime Minister graph - it turned out to be three edits, and voila! I'm a little rushed at the moment, so don't have the chance to tweak it fully. I removed the confidence intervals, because they were pretty large and obscured the view (and also aren't as relevant here). Comments, suggestions? --Trevva (talk) 19:09, 8 October 2008 (UTC)[reply]

Has anyone else noticed that the Herald Digipoll puts Helen Clark more than 10% above the level of support she gets in all the other polls? Seem to be some methodological issues there! --Trevva (talk) 19:35, 8 October 2008 (UTC)[reply]

Yes, a bit of exploratory data analysis seems to show that the poll results vary significantly between the polling companies.Mrfebruary (talk) 04:18, 13 October 2008 (UTC)[reply]

Green Party Support analysed by Polling Company, NZ polls from Oct '05 to Oct '08.
NZ Labour Party Support analysed by Polling Company, NZ polls from Oct '05 to Oct '08.
NZ National Party Support analysed by Polling Company, NZ polls from Oct '05 to Oct '08.


I was able to improve the bias analysis quite a bit after some thought. Rather than just look at the distribution of values, I de-trended the data by calculating the biases for each company ie the difference between the value fitted by the smoother, and the actual estimate from the poll. I was then able to test for significant differences in the bias between Labour and National (being the main two parties of interest). I conclude that two polling agencies (Colmar-Brunton, Nielsen) show a statistically bias towards National in relation to Labour at the 95% level. TNS shows a statistically significant bias in the reverse direction i.e. towards Labour in relation to National. --Trevva (talk) 13:01, 19 October 2008 (UTC)[reply]

Biases in the party support polls by indiviudal polling company. Companies whoose name is marked with an asterisk show a statistically significant bias towards either Labour or National at the 95% level (ie the absolute bias in one party is significantly greater (or less) than the other). Bias is defined in absolute terms as the (p-s), where p is the value estimated by the individual poll, and s is the "mean" value estimated using the Loess smoother taking into account all polls. No polls show significant deviations from zero at the 95% level.

Sources

[edit]

Could I please make a request to everyone/anyone who updates the actual poll data only to cite Curiablog as a source of last resort - its an intermediary source, and I think its more appropriate to cite the original source whenever its available. --Trevva (talk) 07:49, 24 October 2008 (UTC)[reply]

If you want to track down each bit of information in the source, have at it. For instance, I have found the site to be accurate and about the only place to find a complete written listing for TV3. Using video of TV3 coverage seems less suitable to me. I was using it as a place holder for the digipoll. I have found taht you often need to wait for the Herald's second article on the poll to get all of the information we use in the table. -Rrius (talk) 08:03, 24 October 2008 (UTC)[reply]
Yeah, I spent ages trying to get more information about the TV3 polls, and simply can't find anything at all - Curiablog seems an appropriate reference in those cases. I didn't realise that the Herald takes so long to get the data up either. Keep up the good work. --Trevva (talk) 08:38, 24 October 2008 (UTC)[reply]
With the Herald, the issue is that they sometimes post their analysis article first, which is incomplete. I'm just not patient enough to wait until I can find the complete results article. -Rrius (talk) 06:30, 26 October 2008 (UTC)[reply]

Alternative Split Graph Design

[edit]

Which would you all rather see? The current graph design, or the following --Trevva (talk) 19:44, 28 October 2008 (UTC)[reply]

Alternative summmary figure design, showing more information about the minor parties. Note the split in the scales
I prefer the current one. It show the true scale of the four parties with respect to each other. The proposed one does not add a lot of extra info and has a cluttered look especially around the later polling dates. Perhaps the rolling average of ALL parties should be added to the current graph rather than just those that are currently shown. -- Alan Liefting (talk) - 20:08, 28 October 2008 (UTC)[reply]
I thought that splitting it would allow me to put ACT & Maori Party on as well, but as you say, it gets pretty cluttered with all those points. Not sure what the right approach here is. Can you explain your suggestion a little more? The way I interpret it would be that for the smaller parties, I just put the smoothing line and nothing else (ie no final estimate, no points, no confidence intervals). Is that right?--Trevva (talk) 00:06, 29 October 2008 (UTC)[reply]
While both graphs are excellent, I would rather keep with the current graph. It's what viewers are used to and it has already received lots of praise. Also, in terms of Alan's second point, I suspect a smoothed line representing ACT, Maori Party and NZF would overlap with the Greens data points, also ending up a bit cluttered. Mrfebruary (talk) 10:35, 29 October 2008 (UTC)[reply]
It would be useful to have the averaged-out numbers for the other parties somewhere in the article. -Conniption (talk) 16:56, 1 November 2008 (UTC)[reply]

Outcome

[edit]

So, how did the model do in the end? Here's the comparative table from the raw result this evening (ie without specials):

Party Modelled Value (%) Confidence Interval Party Vote
National 46.4 2.5 45.45
Labour 33.9 2.1 33.77
Greens 8.7 1.2 6.43
NZ First 3.4 1.0 4.21
Act NZ 2.9 0.5 3.72
Maori Party 2.6 0.6 2.24

So it looks like although it did a pretty good job with the major parties, the model overestimated the Greens and underestimated NZ First and ACT. --Trevva (talk) 17:59, 8 November 2008 (UTC)[reply]

[edit]

Hello fellow Wikipedians,

I have just added archive links to one external link on Opinion polling for the New Zealand general election, 2008. Please take a moment to review my edit. If necessary, add {{cbignore}} after the link to keep me from modifying it. Alternatively, you can add {{nobots|deny=InternetArchiveBot}} to keep me off the page altogether. I made the following changes:

When you have finished reviewing my changes, please set the checked parameter below to true to let others know.

This message was posted before February 2018. After February 2018, "External links modified" talk page sections are no longer generated or monitored by InternetArchiveBot. No special action is required regarding these talk page notices, other than regular verification using the archive tool instructions below. Editors have permission to delete these "External links modified" talk page sections if they want to de-clutter talk pages, but see the RfC before doing mass systematic removals. This message is updated dynamically through the template {{source check}} (last update: 5 June 2024).

  • If you have discovered URLs which were erroneously considered dead by the bot, you can report them with this tool.
  • If you found an error with any archives or the URLs themselves, you can fix them with this tool.

Cheers.—cyberbot IITalk to my owner:Online 06:08, 21 January 2016 (UTC)[reply]