Wikipedia talk:WikiProject Elections and Referendums/Elections Manual of Style/Results tables

Invalid votes

How are invalid votes included in the computation of percentages? Like this:

Name	Votes	%
Jerry	15	33.33
Tom	13	28.89
Dick	9	20.00
Harry	4	8.89
Invalid	4	8.89
Total	45	100

or

Name	Votes	%
Jerry	15	36.59
Tom	13	31.71
Dick	9	21.95
Harry	4	9.76
Valid	41	91.11
Invalid	4	8.89
Total	45	100

–H T D 17:45, 7 December 2013 (UTC)[reply]

The results for the candidates should add up to 100% – this is how the vast majority of election results are produced. This is why I don't think having a % next to invalid/blank votes is particularly helpful, as it can be confusing. Number 5 7 20:00, 7 December 2013 (UTC)[reply]

I think I wasn't clear: some countries do not include invalid votes when computing a candidate's percentage; instead they compute it from total valid votes. Then the valid votes' percentage is computed from all votes, including invalid votes. For an example, see French presidential election, 2012. –H T D 09:37, 8 December 2013 (UTC)[reply]

Sorry, I don't think I was clear either! The vast majority of countries do not include invalid votes when computing candidates/parties' % (and for the few that do, international sources still tend to present them in the usual format - i.e. with invalid votes omitted from the total votes used to calculate %). The total valid votes should always be 100%, no? Number 5 7 12:58, 8 December 2013 (UTC)[reply]

Taking a look at the examples I cited at Talk:Honduran general election, 2013, (mostly) the valid votes are not at 100%; there are even cases where the valid votes is repeated, one "total votes for candidates" which is 100%, and another "valid votes" whose percentage depends on how many invalid votes there are; IMO this is an inefficient way of dealing with this. Of course this is moot when there is no data for invalid votes we can merge "valid votes" and total votes.

Again, how did literature deal with this? –H T D 10:37, 12 December 2013 (UTC)[reply]

Of all the sources mentioned below, only one (EISA) actually total their columns, and they total valid votes to 100% (the other sources put all the additional info in a separate section). Not having 100% in the total row for valid votes is rather confusing, as the sum of the figures above will be 100. In the second example above, the "Valid" row is inconsistent because 41 is the total of the above rows, but 91.11 is not (hope that makes sense as it's rather difficult to explain). Number 5 7 17:58, 12 December 2013 (UTC)[reply]

The examples given on this page do not list invalid votes; I don't know if they're even allocated for. On this BBC results table, the total number of valid and invalid votes aren't listed. If you'd tally all of the party's votes, it's 27,147,331, less than the turnout of 27,148,975 or a 99.99% valid vote rate. They didn't list valid and invalid votes separately; instead they showed the turnout at 61.2%.

If we're listing valid and invalid votes, unless demonstrated otherwise, I'm leaning on using the percent of the valid votes over the turnout. This is where the gray shades are useful. –H T D 18:39, 12 December 2013 (UTC)[reply]

Which page are you talking about? I also don't understand what you mean by the penultimate sentence - could you make up an example table? Number 5 7 19:53, 12 December 2013 (UTC)[reply]

This one. We've added links on various sources. As for an example table: {{Singaporean presidential election, 2011}}

On this example, "Total vote cast" is 100%. The results tables I added at Talk:Honduran general election, 2013 have some sort of visual cue separating the candidates votes and the other stats, and all but one had the valid and invalid votes' percentages computed from the total votes/turnout. –H T D 20:00, 12 December 2013 (UTC)[reply]

Hmmm. I still think the bottom section is slightly confusing as the valid votes total is the same as the rows above but the % isn't. For this reason I favour leaving out the % for invalid/blank votes. However, if it were to be done, I'd rather see it in the style below (the first is with no %). The problem with the summary rows at the bottom is also where to put seats in the case of a legislature election - if the rows are entitled "Total valid votes" and "Total votes", which row do we put the total seats into? Number 5 7 20:07, 12 December 2013 (UTC)[reply]

Party	Votes	%	Seats	+/–
Fooian National Party	300,001	30.0	31	+1
Fooian People's Party	299,999	30.0	29	–1
Fooian Conservative Party	200,001	20.0	19	–1
Fooian Labour Party	199,999	20.0	21	+1
Invalid/blank votes	15	–	–	–
Total	1,000,015	100	100	0
Registered voters/turnout	1,500,000	66.7	–	–
Source: Central Election Committee

Party	Votes	%	Seats	+/–
Fooian National Party	300,001	30.0	31	+1
Fooian People's Party	299,999	30.0	29	–1
Fooian Conservative Party	200,001	20.0	19	–1
Fooian Labour Party	199,999	20.0	21	+1
Total valid votes	1,000,000	100	100	0
Invalid/blank votes	150,000	13.0	–	–
Total	1,150,000	100	100	0
Registered voters/turnout	1,500,000	66.7	–	–
Source: Central Election Committee

I think the reader would have to assume that the percentages should add up to 100%. If there are separate for valid and invalid votes, the reader should assume that the percent figure for the valid votes and the invalid votes should add up to 100% (meaning that the valid votes should be less than 100%). Some results tables have identical figures for "Total" that is always 100%, and valid votes which is 100%-invalid votes% but I don't think that's an efficient way of doing this.

If anything, the only figure that should be 100% is the registered voters. Like this:

Candidate A: 52%
Candidate B: 48%
Valid votes: 100%-invalid votes%
Invalid votes: 100%-valid votes%
Total votes (or turnout): 100%-total votes%
Total voters (or turnout): 100%

See for example how the French did it.

On the question on where would the seat totals go, it's almost always at the very first summary row. These summary rows are important if people are looking for trends on valid and invalid vote percentages. –H T D 03:59, 13 December 2013 (UTC)[reply]

I've spent a few days mulling over how we can best present this. I still think having the sum of a column that adds up to 100% not being 100% would be confusing, but I can see your point re showing what proportion of votes cast were invalid. As a result, I've devised the table below. It has an unambiguous total row in which the totals are exactly those of the columns above. However, there is a separate section below detailing the breakdown of valid and invalid votes (this could be split further to separate blanks to appease other editors!). It also avoids repeating the totals of the seats column again.

Another major advantage of this would be clarity for elections in which voters can cast more than one vote. In the example below I have used Luxembourg's latest election, in which 218,000 voters cast 3.3 million votes. It allows us to differentiate between the total of votes and the total number of voters. We could perhaps use "ballots" instead of "votes" to clarify further.

This also seems to be more similar to how results are presented in academic literature, as Nohlen, Adam Carr, the IFES and the IPU (and the French link you provided above) separate the detail in the bottom section into a different table. Thoughts? Number 5 7 21:16, 17 December 2013 (UTC)[reply]

Party	Votes	%	Seats	+/–
Christian Social People's Party	1,103,636	33.68	23	–3
Luxembourg Socialist Workers' Party	664,586	20.28	13	0
Democratic Party	597,879	18.25	13	+4
The Greens	331,920	10.13	6	–1
Alternative Democratic Reform Party	217,683	6.64	3	–1
The Left	161,759	4.94	2	+1
Pirate Party Luxembourg	96,270	2.94	0	New
Communist Party of Luxembourg	53,669	1.64	0	0
Party for Full Democracy	49,290	1.50	0	New
Total	3,276,692	100	60	0

Valid votes	203,557	93.18
Invalid/blank votes	14,896	6.82
Votes cast	218,453	100
Registered voters/turnout	239,668	91.15
Source: Elections in Luxembourg

In this suggestion I think it's safe to omit the valid votes; the reader can deduce that the percentage of the valid votes is 100%-invalid votes%. In that way we can avoid redundancy. –H T D 05:22, 18 December 2013 (UTC)[reply]

Leaving the reader to deduce things isn't necessarily a good idea - I've seen several comments on talk pages of articles on elections in which voters can cast more than one ballot (e.g. here) or even tags added to articles (e.g. here - this from an editor who frequently edits election articles, and this despite a note under the table explaining the reasoning) asking why the total number of votes is more than the registered voters. I think the above makes it clear. Given that the two figures (total and valid votes) in this example are different, I don't see how it could be redundant. Number 5 7 07:57, 18 December 2013 (UTC)[reply]

What happens in single-winner or a group of single-winner elections where "valid votes" will be equal to "totals"? –H T D 02:56, 19 December 2013 (UTC)[reply]

I would still keep it both for consistency (especially if a template can be created), and as it would be useful to show the valid and invalid %s, whilst still keeping the figure for total percentage for the parties in the "Total" row above correct. Of course, in these cases there would be a single figure that is duplicated once, but as the figures are being used to show different things in different rows, I don't think it's a problem. It's certainly less confusing that having a total of a column that sums to 100% not being that figure! I'm happy to compromise on how data is arranged in the table (hence suggesting several alternatives above), but data integrity (i.e. a "Total" actually being the total of the rows above) is not something we should be messing around with. Number 5 7 16:56, 19 December 2013 (UTC)[reply]

I'd go with this, but I'd like the "Total" row labeled into "Total votes" or something. Simply labeling it as "total" would be confusing. –H T D 10:46, 30 December 2013 (UTC)[reply]

I've edited the above table to be "Total votes cast" instead. I was wondering whether the lower section should use "ballots" instead of "votes" to clarify the difference between the two (in local elections where I live we cast three votes on one ballot paper). What do you think? Number 5 7 15:24, 30 December 2013 (UTC)[reply]

Most uses are either "Total votes cast" or "Total votes". I'm of the impression that "ballot" can mean either the ballot paper, or the each of the questions on the ballot. We can also use "Turnout" in this one. –H T D 16:00, 30 December 2013 (UTC)[reply]

OK, so are we happy with the table above? Number 5 7 16:24, 30 December 2013 (UTC)[reply]

What happens if "Total votes cast" and "turnout" are different? –H T D 17:00, 30 December 2013 (UTC)[reply]

I'm not sure what you mean. How would that happen? Number 5 7 17:49, 30 December 2013 (UTC)[reply]

This happens in plurality-at-large voting, or in any election where a voter has more than one vote in a ballot question. –H T D 18:27, 30 December 2013 (UTC)[reply]

I still don't understand where the difference would be. "Total votes cast" is the number of people who have voted - i.e. the turnout. The actual total number of votes is in the first total row. The table above is an example of where voters have more than one vote. Number 5 7 18:35, 30 December 2013 (UTC)[reply]

The first total has to be clearly stated what total that is. Why not "Total votes", then use "Turnout"? –H T D 18:44, 30 December 2013 (UTC)[reply]

Because the first total row is not just total votes - it's also total seats. Perhaps the second total row should just be "Votes cast" (avoiding using another "total") - this is what IFES use in cases like Luxembourg. Number 5 7 18:56, 30 December 2013 (UTC)[reply]

Can we not use "Turnout" and have the actual turnout percentage to go along with it, with 100% going to total voters? Or that should be 100% so that the valid and invalid votes add up? –H T D 19:07, 30 December 2013 (UTC)[reply]

I think it would be easier to keep it as 100% so the two above add up. Have amended above - what do you think? Number 5 7 19:13, 30 December 2013 (UTC)[reply]

I won't oppose this as long someone comes up with a persuasive comment later. –H T D 13:23, 2 January 2014 (UTC)[reply]

Cool. I have created a template to display election results like this, as well as party colours as you requested below - see User:Number 57/Election template. Number 5 7 15:37, 2 January 2014 (UTC)[reply]

You know what, I'd ultimately favor whatever the editors want to use in the election results table. There are too many countries, elections and types, and it's impossible to make one, two, or how many templates that fits all. The only consistency we'd follow is that all election types from a single country must have a consistent format, if possible, and that no one should revert to whatever old format if it's been changed, as long as all other election types in a single country follow suit. –H T D 19:46, 11 January 2014 (UTC)[reply]

Naming of parties and colors

Some election results tables use the {{party name with colour}} template like Template:United Kingdom general election, 2010; this automatically adds a color bar and a pipelink to the table identical to the one use in the infobox. What happens to these tables? While the color bars might not be needed, some are useful if the results tables come with maps and charts. –H T D 18:00, 7 December 2013 (UTC)[reply]

I don't see a problem with adding an extra column with a colour bar if the colours for all parties are known. However, I think it should be avoided if not all party colours are known. I suspect there is a lot of original research going on with party colours as well (I've certainly seen editors making up colours for Israeli parties that existed only until the 1950s. Number 5 7 20:02, 7 December 2013 (UTC)[reply]

These should only be applied for parties whose colors are known; or of WP:RS are unanimous in assigning that particular color to a particular candidate/party. If they're unknown, just leave it blank (or transparent, in this case).

If we're using colors, I'd suggest using {{party name with colour}} as this would be uniform for all results tables and we wouldn't have to change every single table if something has to be changed. Another issue is that it gives the short name of the party, such as "Conservative" instead of the longer one, such as "Conservative Party". –H T D 09:37, 8 December 2013 (UTC)[reply]

In addition to colors on parties that no longer exist, if contemporary sources did not assign colors to a party, but current sources are unanimous in assigning a particular color to that party, the color used by current sources can be safely used here. What's wrong is somebody completely making up colors. –H T D 14:51, 13 December 2013 (UTC)[reply]

Percent sign in results tables

Several tables that do not use Number 57's formatting do not omit the percent sign in election tables. What should be done about this? –H T D 18:00, 7 December 2013 (UTC)[reply]

I think having the % sign in every cell is unnecessary when it's a % column. Number 5 7 20:02, 7 December 2013 (UTC)[reply]

I wasn't pushing either way, just that some results tables do include the percent sign in percent columns. See for example the French example I linked above. Do election results tables elsewhere omit the percent sign? –H T D 09:37, 8 December 2013 (UTC)[reply]

The two main academic sources I used (Dieter Nohlen books and Adam Carr) both omit the % sign in individual rows, as do EISA, the IPU. The only other main source I have found is the IFES, but they do not present the results in tabular format. Number 5 7 12:53, 8 December 2013 (UTC)[reply]

How about news and media websites and official sources? BBC does omit the percent signs. This one omits it too. For presidential ones, this omitted it too. –H T D 10:37, 12 December 2013 (UTC)[reply]

Given the fact that it appears that the % sign is omitted most of the time in other sources, plus the extra clutter it creates, I think the case for excluding it is fairly clear cut! Number 5 7 17:47, 12 December 2013 (UTC)[reply]

I'd insist on using the percent sign but since it's omitted I'd submit on this one. –H T D

Graphs in results tables

Several results tables (particularly East Asian ones) include a graph along with the percent figure. This can be useful as graphs are handy and illustrate stats more readily than digits. What will happen on results tables that use those? –H T D 18:00, 7 December 2013 (UTC)[reply]

I don't think they should be used. In elections with a small number of candidates they might be useful, but in elections with 40+ parties and where a significant number are getting under 1%, the additional info would not be clear or useful. If we are going to be consistent either way, I would prefer to not include them. Number 5 7 17:54, 12 December 2013 (UTC)[reply]

Symbols in results tables

Quite a few results tables, mostly those which aren't using Number 57's formatting, include symbols. While these may not be useful for a lot of people, these are like the political colors used elsewhere, although this may not be easily used in things such as maps and charts. What will happen on results tables that use those? –H T D 18:00, 7 December 2013 (UTC)[reply]

What symbols would these be? If they're party logos, I don't believe they can be used in tables unless they are not copyrighted. Number 5 7 20:03, 7 December 2013 (UTC)[reply]

The only examples I saw these were Singaporean elections. They skirted around the WP:FU by making "simpler" version of party logos. I'm for removing these, FWIW. –H T D 09:37, 8 December 2013 (UTC)[reply]

Darker shades in results tables

Several tables that do not use Number 57's formatting have a darker shade for row/s that denote totals. These are useful in illustrating to the reader which stats are for a particular candidate and which aren't. What will happen on results tables that use those? –H T D 18:00, 7 December 2013 (UTC)[reply]

Well, we have to decide whether to use the shading or not and then apply it consistently. Number 5 7 20:03, 7 December 2013 (UTC)[reply]

I would favor a shade that is identical to the table header's color. This is important in distinguishing which stats are candidates' votes and which aren't. –H T D 09:37, 8 December 2013 (UTC)[reply]