Talk:Opinion polling for the United Kingdom general election, 2015/Archives/2011/August

From Wikipedia, the free encyclopedia
Jump to: navigation, search


Just looking over the table of polls, it is beginning to look very cluttered and cramped, and could do with a bit of a clean up. I wanted to ask people's opinion on whether all of the columns are really needed - is there really a need for the '% Lead' column and the 'Seats Lead' column for example - neither are actually directly relevant to the final result and the reader can work this out from the information in the individual party results. Would it perhaps be better just to have a table that has the main results, and then the predicted majority column - after all, that is ultimately what decides an election. I'm thinking of something like this:

Date(s) Conducted Polling Organisation/Client Sample Size Conservative Labour Liberal Democrats Other Predicted
21–22 Dec YouGov/The Sun 1,695 41% (288) 42% (329) 8% (8) 10% (7) 4
15–16 Dec YouGov/The Sun 1,922 41% (292) 41% (325) 9% (8) 10% (7) Hung Parliament
Lab short 1 seat

Does anyone else have any suggestions to make the tables look less cluttered/cramped? Spiritofsussex (talk) 15:02, 5 January 2011 (UTC)

Quite. I stumbled across this page, and it's despairingly cluttered. The hung parliament box is particularly difficult to read.
On a side note, does anyone have any idea how those seat calculations (from the polls) were substantiated? I say this because the YouGov poll for 3-4 October in the table has an error (says "Lab short 9 seats" AND "Lab short 29 seats"), and I can't figure out from the PDF document how the seat numbers were figured out, as I can't see such figures in the document.
If anyone else is up for it, I'd be happy to work with someone else/other people on going over this page completely for an overhaul. —JeevanJones (talk) 19:18, 13 July 2011 (UTC)
I'd certainly be happy to help on an overhaul of the page - it is certainly in need of one. There are a number of basic things that could be done - including removing some unnecessary information such as '% Lead' which people can easily work out for themselves. Also, I would be quite tempted to remove all of the seat projections for a couple of reasons:
1. They are only projections, based wholly on the assumption that every constituency seat will follow an identical swing - something that is never going to happen; and
2. The forthcoming boundary changes are going to reduce the total seats to 600, thus rendering all of these projections absolutely worthless.
The article for the last parliament (Opinion polling for the United Kingdom general election, 2010) never gave any seat projections at all, and it only seems that a user started doing it on this page, and it became the norm.
In terms of wider, more intrinsic changes - a quick look at other opinion polling articles could provide a few ideas about better format for the results to be displayed in, but I think this would need to be thought about and discussed on this talk page for all users to have their say in before any big changes are made. Spiritofsussex (talk) 08:46, 15 July 2011 (UTC)
I agree, seat projections would be rendered fairly redundant pretty soon (not to mention they're highly debatable, as you say). What are your thoughts about different formats from previous articles? Looking at the article, I'd probably remove the predicted majority and seat prediction columns, and probably simply highlight the party with the highest % at each row. That being said, I suppose the background colour (ie on the lead/majority columns) can be useful for a quick glance at which party is ahead in the polls. I also wonder if it's worth taking the pollsters into account as there is a big discrepancy due to a number of reasons (weighting etc.) between say YouGov and ComRes which make them difficult to compare at times. —JeevanJones (talk) 10:35, 15 July 2011 (UTC)
I am free now, so I'm going to work on an overhaul of this page. If it's deemed inappropriate, feel free to revert any changes I make. —JeevanJones (talk) 11:43, 20 July 2011 (UTC)
Re: overhaul: Okay, I've overhauled the results table, streamlining and vastly simplifying the display of the data. Any extra information sought can easily be found through clicking through to the source of each polling report. I have removed the projected majority/seats column for a number of reasons:
  1. They are only projections, therefore quite suspect (anyone with experience in UK polling predictions will know well how imprecise they can be).
  2. There is to be a major boundary change and reduction in seats (from 650 to 600), therefore rendering any seat allocation predictions fairly worthless for comparison.
  3. Percentages of the vote are superior as a measure of polling. Indeed, very few media outlets use seat projections as a reliable measure of popular support of political parties, but rather the percentage of support.
Regretfully, I had to remove the visual graphic due to it no longer being applicable to the new table of results. If anyone could find the original image without notations, it would be much appreciated.
I'm open to recommendations for better ways of showing which party is in the lead in the polls, other than through making the figures bold, as in retrospect it may not be as clear as I would perhaps have liked.
As I've said before, feel free to bring back features I have taken out, my overriding intent was one of accessible simplicity, per WP:BOLD. —JeevanJones (talk) 14:19, 20 July 2011 (UTC)

I'm profoundly unconvinced by the removal of the 'predicted majority' column. While theoretical, this information is useful and is not contained in the linked to polls. — Preceding unsigned comment added by (talk) 12:13, 21 July 2011 (UTC)

The reason for removing the predicted majority column has been discussed at length above. I have today expanded the article a bit - the main changes I've made are:
1. Added a section on the methodology of each polling company - this will alleviate the worries expressed above about the questions over different companies' methods - people can judge for themselves by reading the brief summary of different companies' ways of working.
2. I've taken some inspiration from the United States polling articles and added shading to the boxes of the party in the lead, which will provide a clear picture of which party is in the lead with just a quick glance.
3. I've reinstated the listing of political events in the table, as I think these are often helpful to explaining movements in the polls - e.g. listing the election of a new party leader goes a long way to explaining why a party has a sudden spike in the polls.
The page is looking a lot better than it did a few weeks ago, and if anyone has anymore ideas about how to improve it further, please feel free to contribute. Spiritofsussex (talk) 10:48, 26 July 2011 (UTC)
Wow, just wow. I must say, excellent work, Spiritofsussex. I was thinking the article really needed some emphasis like that, and the background shading works well. I only wonder if there could be a good way of showing ties in the polls with further background shading (it may just be me looking at it oddly, though)? Your other two points make good sense. Nice job with the methodology. For the casual reader, the events could definitely help explain the change in trends. All in all, the page looks a lot better and easier to read. —JeevanJones (talk) 10:54, 2 August 2011 (UTC)