Ancestors

Written by toybox [fr] on 2024-11-27 at 13:00

I just realised that mean values don't really make sense for numbers that you shouldn't add.

For example, the mean weight of a pumpkin is fine -- you want to know how much 1000 pumpkins weigh, you just take mean x1000. But the mean age of a population? You cannot add the ages, you can't have "the total age of 1000 persons".

=> More informations about this toot | More toots from toybox@framapiaf.org

Written by toybox [fr] on 2024-11-27 at 13:02

So I guess what I mean [haha] is: mean values are really a density, if you want to say something about data distribution you should use the median and that's it. The only time giving a mean value for descriptive purposes makes sense is when mean = median, otherwise it's just laziness as in "can't be bothered computing the median" or "my software just gave me the mean".

=> More informations about this toot | More toots from toybox@framapiaf.org

Written by toybox [fr] on 2024-11-27 at 13:04

(For density purposes like estimating the weight of 1000 pumpkins, of course, the mean value is more useful than the median.)

=> More informations about this toot | More toots from toybox@framapiaf.org

Toot

Written by toybox [fr] on 2024-11-27 at 17:46

I mean, even the unit of a mean value is not the same as the unit from the data! If age of people is in years or weight of pumpkins in kg, then the median age or weight is still years or kg, but the mean age or weight is years per person or kg per pumpkin. Don't report mean values with the wrong unit 😠

=> More informations about this toot | More toots from toybox@framapiaf.org

Descendants