Talk:1974 Luxembourg general election
 | This article is rated Stub-class on Wikipedia's content assessment scale. It is of interest to the following WikiProjects: | |||||||||||||||||
| ||||||||||||||||||
Move discussion in progress
[edit]There is a move discussion in progress on Talk:Luxembourgian general election, 2013 which affects this page. Please participate on that page and not in this talk page section. Thank you. —RMCD bot 09:16, 4 November 2013 (UTC)
About the results
[edit]Well, User:Number 57, I got a better source. Here: https://sip.gouvernement.lu/en/publications/bulletin/1974/BID_1974_2.html (It seem to be in french).
Some collaborator on the WP french page use it to create a page about this topic and it seems that you'll need to reverse it. At the same time, the change I did is about logic. 875K votes is more than 840K votes, than the % should be changed accordingly.
--InternauteLambda (talk) 14:18, 23 April 2019 (UTC)
- I was in the middle of posting this on your talk page to explain the percentages in the table, which are taken from the Nohlen & Stöver source (hence the reference in the note).
- The explanation is that voters could cast more votes in some constituencies that others. Therefore a percentage based on the total votes received is distorted in favour of the larger constituencies where voters could cast more votes. Instead they appear to have averaged the percentage across the constituencies to come up with a more representative figure.
- As an example, let's say a voter can cast 10 votes in constituency A and five in constituency B. In constituency A, three people cast all 10 votes for Party X and two for Party Y. In constituency B, three people cast all five votes for Party Y and two for Party X. Overall, the same number of people have voted for Party X and Party Y. However, in terms of the number of votes cast, Party X has received five more. The question is how to represent the percentages. Based on the overall votes cast, Party X has 53.3% of the vote. However, if the votes are weighted based on the potential number of votes cast (e.g. by dividing votes in constituency A by 10 and constituency B by 5), a more representative percentage can be obtained. See the table below as an example of how the different figures can be reached.
- Hope that makes sense? Cheers, Number 57 14:23, 23 April 2019 (UTC)
| Party | Constituency A | Constituency B | Overall | Weighted | ||||
|---|---|---|---|---|---|---|---|---|
| Votes | % | Votes | % | Votes | % | Votes | % | |
| Party X | 30 | 60% | 10 | 40% | 40 | 53.3% | 5 | 50% |
| Party Y | 20 | 40% | 15 | 60% | 35 | 46.7% | 5 | 50% |
- I get the concept. There's the number of valid votes underneath. But we're not talking about the same thing. I'm talking about the numbers and the % indicated. It's written that the first party got 840K votes and got 29.9% while the second one got 875K votes and 27% of the votes. Regardless of the fact that many votes could be cast by one voter, how could a party with more votes got less %, unless it's an error?
- Check this link out: https://fr.wikipedia.org/wiki/Élections_législatives_luxembourgeoises_de_1974#Résultats_nationaux. This is from the french WP page, and the source is the Luxembourg government. It seems that the % fit. Please check it and the lux. government link: https://sip.gouvernement.lu/en/publications/bulletin/1974/BID_1974_2.html.
- --InternauteLambda (talk) 14:36, 23 April 2019 (UTC)
- Based on my explanation above, percentages and total votes received are not linked. It would be possible for a party to receive more votes (because they were heavily voted for in a large constituency) but have a lower overall percentage. Taking the same example as above, let's say three people in constituency A vote for Party X and one for Party Y (30 votes vs 10). In constituency B, three people vote for Party Y and none for Party X (0 votes vs 15). Overall Party X has 30 votes and Party Y has 25. However, when the figures are weighted four people overall (57%) have voted for party Y and three (43%) for Party X. Number 57 14:43, 23 April 2019 (UTC)
- Well, here's the results from the document of the Luxembourg government https://sip.gouvernement.lu/en/publications/bulletin/1974/BID_1974_2.html:
- Luxembourg is divided in 4 constituencies. The document shows only the results for each district.
- I think that the results show what's happening.
- --InternauteLambda (talk) 15:35, 23 April 2019 (UTC)
- And for that 1984 election, that's another error. The % need to be switched and there's a document from the lux government showing that.
- --InternauteLambda (talk) 15:37, 23 April 2019 (UTC)
- I don't dispute your calculations are correct. However, this is not a case of either set of figures being wrong (the 1984 one is also not an error, it's the same correction applied as here) – it's a question of which set of numbers we want to go with. Either we can show the raw calculation as above (after which readers might ask how the CSV ended up with more seats than the LSAP), or we can show one that corrects the distortion caused by the different constituency sizes (the Nohlen & Stöver figures). Both are from sources that are deemed reliable. Number 57 15:44, 23 April 2019 (UTC)
- Well, having one more seat happens often when the total is the addition of regional votes and seats. I dont' think that's a problem. These results should be presented as it is. Plus the votes and % fit. And you could use the lux gov link (just indicate it's in french) and provide districts results to show the whole thing.
- --InternauteLambda (talk) 15:58, 23 April 2019 (UTC)
- And I urge you to check the WP french page on the subject, to see how complete it is.--InternauteLambda (talk) 16:00, 23 April 2019 (UTC)
- Again, I don't dispute that the figures on fr.wiki are correct for what they are, or have a problem with using a French source. The question is whether we show raw or corrected percentages. I will have a look at the N&S book again this evening and see their rationale before committing one way or the other.
- As an aside, I just did the N&S calculation myself and got the same result. Number 57 16:08, 23 April 2019 (UTC)
- I don't dispute your calculations are correct. However, this is not a case of either set of figures being wrong (the 1984 one is also not an error, it's the same correction applied as here) – it's a question of which set of numbers we want to go with. Either we can show the raw calculation as above (after which readers might ask how the CSV ended up with more seats than the LSAP), or we can show one that corrects the distortion caused by the different constituency sizes (the Nohlen & Stöver figures). Both are from sources that are deemed reliable. Number 57 15:44, 23 April 2019 (UTC)
| Party | Modified votes[a] | Modified % |
|---|---|---|
| CSV | 49,258 | 29.9 |
| LSAP | 44,461 | 27.0 |
| DP | 38,394 | 23.3 |
| PSD | 16,648 | 10.1 |
| KPL | 14,438 | 8.8 |
| Liberal Party | 925 | 0.6 |
| RSP | 720 | 0.4 |
- ^ Total of (number of votes in constituency divided by number of seats)
- No, they're not the same. The LSAP got more votes than CSV. And the % you have are different than mine. As for the presentation, I've given you my answer: Raw!
- --InternauteLambda (talk) 16:58, 23 April 2019 (UTC)
- And you could check the english WP page of the lux elections of 1964. It ressembles what I try to do on this page.--InternauteLambda (talk) 17:17, 23 April 2019 (UTC)
- Are you being deliberately obtuse? The table is to show that how the percentages in the Nohlen source were calculated; the number of votes is not supposed to be the same as those in the results table. Also, I don't know why you're referencing the 1964 article, as the percentages there are also ones from Nohlen that have been recalculated to avoid the distortion (the Communist Party has a much higher number of votes than the Democratic Party, but a lower percentage).
- Anyway, I checked the Nohlen book, which states (on page 1254) that the calculation of the percentages is the same as done by STATEC, the national statistics agency, and that "a simple addition of the votes won by different parties would not lead to any usable figures as the number of votes the electors are entitled to choose varies considerably in the four electoral districts. The distortion caused by this unequal weight of the voters can finally be overcome by the division of the party votes by the respective district magnitude. The sum of the different quotients obtained per party divided by the overall sum of the quotients finally forms the national share of the vote of the party."
- I checked the STATEC website, and this is indeed how they calculate election result percentages. See this page where the results for all elections since 1999 are detailed. A relevant example is 1999:
- DP - Demokratesch Partei – 632 707 votes – 22,35% – 15 seats
- LSAP - d'Sozialisten – 695 718 votes – 22,29% – 13 seats
- Even though the LSAP have more votes, STATEC give the DP a higher vote share based on this calculation.
- So whilst I was originally planning to go with the raw votes percentage, the fact that STATEC are still publishing the results in this format, and the comment in the Nohlen books is swaying me towards going with the Nohlen/STATEC approach. Number 57 21:32, 23 April 2019 (UTC)
- No, I'm not obtuse. I just think of simplistic ways to present election results, where the number of votes corresponds to the % shown. Now, you're using a calculated way to present results using a method that seems complicated and it's only used by Luxembourg statistics? KIS, man! Keep it simple! Why going for calculations only for little seat quirks happening from time to time? The raw approach present things in the way they should be. Plus, the little indication, at first reading, shows that people could vote more than once, which explains why there's 3M votes for around 200K voters, so you'll not have to change it. But, that the number and the % that don't fit could bother many more than me. Especially if on the other pages, it presents the raw method.
- So, bottom line: Keep it simple! Do it raw!
- --InternauteLambda (talk) 23:23, 23 April 2019 (UTC)
- But this percentage calculation is not only used by STATEC – it is also used by the Nohlen book, IFES, Parties and Elections, Election Resources. Number 57 14:55, 24 April 2019 (UTC)
- Well, they complicate themselves by doing calculations that are not necessary. Let's present what's found on the governmental sources like it is. Keep it simple! Do it raw!
- --InternauteLambda (talk) 16:03, 24 April 2019 (UTC)
- STATEC is a government source, as is the Election Commission, both of which use the same calculation. Number 57 16:42, 24 April 2019 (UTC)
- They are non-official results. My source is also governmental but there's is the official results, presented raw! Bottom line: why doing more calculations on top of other calculations because there's minimal seat quirks? --InternauteLambda (talk) 17:47, 24 April 2019 (UTC)
- STATEC is a government source, as is the Election Commission, both of which use the same calculation. Number 57 16:42, 24 April 2019 (UTC)
- But this percentage calculation is not only used by STATEC – it is also used by the Nohlen book, IFES, Parties and Elections, Election Resources. Number 57 14:55, 24 April 2019 (UTC)