3 ms·

no it is not. in any case it's referring to household incomes and only between people who are 35 to 64. so again, no, 1/20 people are not making that much. agai

no it is not. in any case it's referring to household incomes and only between people who are 35 to 64. so again, no, 1/20 people are not making that much. again, percentiles do not work in a fashion where you can take one group (35 to 64) and apply it to the entire population. i repeat, that's not how percentiles work.

anpama 7y ago
That's exactly how they work: if the top 5% of people in some cohort earn $X or more, then that means 1 in 20 people in that cohort earn $X or more. OP may have confused one cohort (households) for another (individuals), but that's irrelevant to how percentiles work.
- percentiles 7y ago
  What you're saying is true but is irrelevant to my point. With the original post the claim was that 1/20 in Boston make more than 374k because the top 5% of people between 35 and 64 make more than 374k. Again this is not true and is not how percentiles work. What's so difficult to understand? If you have 98 people in Boston who are 18 who make $1 and 1 who is 40 who makes $374K and another who is 36 who makes $10K what does this mean? It means 50% of people in Boston between 35 and 64 make more than 10K, but not that half of the people in Boston in general make that much, as the original post suggested. The original poster was claiming he or she was not doing well because of the data NYT presented, but percentiles cannot be extrapolated from one group to another. As you've said it's only useful for a specific cohort in which it's measured. In my hypothetical example the person making 10k would be in the top 2% despite being in the lower 50% for income in their cohort.
  - sokoloff 7y ago
    Your reading of the original comment gives zero charity to their comment. While they did not specify for that cohort, I think even a plain reading would intuit that's what they had in mind. You're technically right.
DangitBobby 7y ago
Clearly the misconception isn't "that's not how percentiles work" it's "that's a percentile of only a subset of the population". I don't know why you keep repeating yourself. It is exactly how percentiles work in that subset.

anpama 7y ago

That's exactly how they work: if the top 5% of people in some cohort earn $X or more, then that means 1 in 20 people in that cohort earn $X or more. OP may have confused one cohort (households) for another (individuals), but that's irrelevant to how percentiles work.

percentiles 7y ago
What you're saying is true but is irrelevant to my point. With the original post the claim was that 1/20 in Boston make more than 374k because the top 5% of people between 35 and 64 make more than 374k. Again this is not true and is not how percentiles work. What's so difficult to understand? If you have 98 people in Boston who are 18 who make $1 and 1 who is 40 who makes $374K and another who is 36 who makes $10K what does this mean? It means 50% of people in Boston between 35 and 64 make more than 10K, but not that half of the people in Boston in general make that much, as the original post suggested. The original poster was claiming he or she was not doing well because of the data NYT presented, but percentiles cannot be extrapolated from one group to another. As you've said it's only useful for a specific cohort in which it's measured. In my hypothetical example the person making 10k would be in the top 2% despite being in the lower 50% for income in their cohort.
- sokoloff 7y ago
  Your reading of the original comment gives zero charity to their comment. While they did not specify for that cohort, I think even a plain reading would intuit that's what they had in mind. You're technically right.

DangitBobby 7y ago

Clearly the misconception isn't "that's not how percentiles work" it's "that's a percentile of only a subset of the population". I don't know why you keep repeating yourself. It is exactly how percentiles work in that subset.