Talk:Geographical distribution of Italian speakers

Issues with the current version

@FrankCesco26: the formatting and sorting of the version you keep reverting to are inconsistent, which is self-evidently undesirable. Countries should be sorted A-Z (so that the order won't need to be changed as the data is updated), and every country should take one line of wikicode (to make the code more readable), unless you can suggest a reason why it should be otherwise. There are also a bunch of non-formatting issues:

when the population figure is derived from a percentage, this should be stated in a separate note, with an appropriate source;

the figure itself, when calculated instead of stated in a source, should be rounded, to avoid giving a phony impression of accuracy (ie, 57,491,000 for Italy, and not 57,490,841);

same applies to percentages, rounding to one decimal place is perfectly fine, we're not splitting atoms here.

And for the love of all things holy, name your references. <ref name=":0" is not a good name to use.--eh bien mon prince (talk) 14:19, 9 November 2018 (UTC)

@Underlying lk:Ok that's perfectly fine, I've not named the reference by myself, but the visual editor does it automatically. In my opinion the less rounded percentages are more effective in comparing the percentages in countries like Croatia and the US where the italian native speakers are the 0.43% and the 0.44%. So at least 2 decimal points are required when we compare the number of speakers. That's fine rounding the figures to the thousand, even tough it seems less precise, dunno, probably we can meet at the hundred. Regarding the alphabetical order, it's ok. Please don't replace statistical sources with the UN source. It took several months of research to gather all the references I've used for all the countries I've sourced. FrankCesco26 (talk) 14:25, 9 November 2018 (UTC)

When the difference is between 0.44 and 0.43, does it really make a difference to show an extra decimal? Again, we're providing the impression of a level of accuracy that is just not there, especially when we're using survey data which certainly has a much larger margin of error (Eurobarometer and whatnot). To clarify, using the exact figures is fine if they come from an official source (say, the Swiss census recording precisely 267,617 Italian speakers in Ticino), but when they're simply calculated by multiplying the population for a percentage (as is the case for Italy) it can be misleading, especially when it is not explained by a source.

Also, I'm puzzled by your opposition to using the UN figures, the version you reverted to still has plenty of them. They have the advantage of having consistent methodology and notes in English, which the primary sources don't always have.--eh bien mon prince (talk) 14:34, 9 November 2018 (UTC)

@Underlying lk:Well, considering that we are comparing census data, I think that it is reliable enough to have more detailed percentages. If there is the original source and the UN data, you should prefer the original source, as it shows important details which the UN statistical division simply doesn't show. If we use the Eurostat population figures, which is a coordinated statistical bureau which directly gathers the datasets from official sources like the national statistical institutes (every country has one), I wouldn't say that the calculated population figures are that misleading, I'd say it really depends on the statistical weight of the survey. Even most censuses are surveys and rely on a sample, even if it is greater than "normal" surveys, for example Switzerland's figures come from the yearly census, which interviewes about 200.000 individuals yearly, and many figures are obtained by adding up multiple years, 3 or 5. On balance, in my opinion, we should give an approximation to the hundred to every survey we use, since the thousands are statistical significant when we use the surveys with the right weights. FrankCesco26 (talk) 13:22, 10 November 2018 (UTC)