>The two examples show exactly the problem of the median: it is not that different from the mean in many situation when we talk about salaries and incomes
???
Did you compute them? do you know how?
The Brazil data I just posted has median around 1383, mean of 2308.36, which is 65% larger.
An interesting test: compute with details the median and mean of the data I just posted. I already gave you the answers. I think you don't understand these terms.
That you don't like the outcome of the most two commonly used indicators is not the fault of the indicators - it's the result of you wanting to put things in the data that are not there.
Your claim about Brazil having "hardly anyone", which I showed was nonsense, shows exactly what you're trying to do - put things in data that simply are not there (and in the case of your claim, are mathematically invalid no matter the income distribution).
>Now, if you start using the percentiles as you're doing here, then you are looking deeper into the data
Percentiles have the exact same graph as dollars, or pesos, or anything you desire.
% = exact$ / total $.
So they have the exact same graph, merely a different label. Do you understand this? The % of total income versus income shows nothing different, there is no "deeper into the data" or "more accurate and reflective of reality" for this.
It's simply scaling ALL the numbers by a constant. No one changes place. No income compared to another changes in ratio. There is no subtle nuance pulled out of anywhere.
>which just proves my point.
Is your point that you don't know what any of these terms mean? Then yes, you proved your point. Mean, median, percent, ratio, decile, all the terms related to income - you have screwed them all up beyond reason.
Don't bother replying until you answer the questions I poste last time:
Do you admit 50% of people in Brazil make median income or more? What amount of people did you mean by "hardly anyone"?
Also answer: is the graph of income by % the same geometric shape as income by $, or yen, or peso, or any currency you care to pick, thereby giving no "looking deeper into the data"?