> Starting conditions are a factor, but a terrible predictor. If you look at income by the quintile you are born into, it is closer to random than 1:1 predictive.
Closer to random would mean a 20% chance of ending on any quintile from any quintile. From your report (which has data from 1991, quite few things changed since then), the probability of being at the top 20% is 3% if you are in the poorest quintile, 40% if in the richest. That's not "closer to random" and definitely not a terrible predictor, given that's just a single factor in social studies.
Not to mention that the quintile method is just an approximation. You don't know if the 24.5% of people in the bottom quintile who get to the next one moved from being 0% to 39% or from 19% to 21%.
> This means most people born into the bottom 20% improve their economic status, and most people currently in the bottom 20% started above it
Well, that's a minimum. If starting conditions weren't a factor you'd have 80% improving their status, not 55%. You can rephrase the statistic in another way, such as saying that 13% of the people at the lowest quintile should be in the highest income bracket, but they're not there because they started poor. Or that 22% of the people in the highest quintile are there not because they worked hard enough, but because they just started there.
Random would mean 20% of ending in the same quintile, perfect predictor would mean 100%, so 40% is much closer to random.
>you can rephrase the statistic in another way, such as saying that 13% of the people at the lowest quintile should be in the highest income bracket, but they're not there because they started poor.
You say "should" but it seems like this is only true if you expect (and want) completely random results.
> Random would mean 20% of ending in the same quintile, perfect predictor would mean 100%, so 40% is much closer to random.
Something that doubles or reduces by 10x the random probability is not "closer to random". I doubt that if you tell something that a factor is "closer to random than to a perfect predictor" they'd expect something like that.
> You say "should" but it seems like this is only true if you expect (and want) completely random results.
There is quite a lot of randomness involved in where one ends up. There is a random distribution of ability, of work industry, location, country, luck, connections, timings... quite a lot of factors so that you'd expect that randomness to dominate over the whole population and things be close to ~20%. In fact, countries with higher social mobility tend to be fairly close to that 20%. In Spain (which is by no means the best at social mobility, but it's the only one I have data at hand from) we are pretty far from that 43% chance of staying at the lower quintile, and also far from the 3% of moving from bottom to top: https://widgets.elpais.com/mapbox/2020/07/atlas-renta/pc_nac...
>Something that doubles or reduces by 10x the random probability is not "closer to random". I doubt that if you tell something that a factor is "closer to random than to a perfect predictor" they'd expect something like that.
This shouldn't be a matter of what the layman thinks, it is a question of statistics. If someone said or thought that the class you were born into has a bigger impact on outcomes than other factors, they would be wrong.
Closer to random would mean a 20% chance of ending on any quintile from any quintile. From your report (which has data from 1991, quite few things changed since then), the probability of being at the top 20% is 3% if you are in the poorest quintile, 40% if in the richest. That's not "closer to random" and definitely not a terrible predictor, given that's just a single factor in social studies.
Not to mention that the quintile method is just an approximation. You don't know if the 24.5% of people in the bottom quintile who get to the next one moved from being 0% to 39% or from 19% to 21%.
> This means most people born into the bottom 20% improve their economic status, and most people currently in the bottom 20% started above it
Well, that's a minimum. If starting conditions weren't a factor you'd have 80% improving their status, not 55%. You can rephrase the statistic in another way, such as saying that 13% of the people at the lowest quintile should be in the highest income bracket, but they're not there because they started poor. Or that 22% of the people in the highest quintile are there not because they worked hard enough, but because they just started there.