There are far more young than olds. Age is a pyramid. Bigger base/ smaller top. What we are looking for is something with a small SD that is also not a pyramid in structure. I.e.- lots of variation, and a fairly equal distribution of the variation

Edit: Look at the graph you posted turned vertically. It's essentially a pyramid; you just chose a visual that plots on a different axis. Google " age pyramid" or population pyramid ... http://www.indexmundi.com/united_states/age_structure.html

(http://www.autoadmit.com/thread.php?thread_id=3592014&forum_id=2#33133526)

Why don't you just make your argument without the pyramid claim?

(http://www.autoadmit.com/thread.php?thread_id=3592014&forum_id=2#33134916)

it also depends on how you're quantifying the difference. for example, how do you compare a difference in a numerical trait like age to a binary trait like gender.

(http://www.autoadmit.com/thread.php?thread_id=3592014&forum_id=2#33133427)

of course the scores are roughly normally distributed anyway, that's why they don't use some other distribution instead.

(http://www.autoadmit.com/thread.php?thread_id=3592014&forum_id=2#33134987)

(http://www.autoadmit.com/thread.php?thread_id=3592014&forum_id=2#33137357)

(http://www.autoadmit.com/thread.php?thread_id=3592014&forum_id=2#33133449)

if you're asking what single trait, all other things being equal, would cause the biggest difference in two different americans, obviously 70 iq/130 iq would probably be the biggest reasonable difference yea. otherwise, it's probably some geographic difference like coastal urbanite v. breadbasket farmer.

race would be a big one except there's very little difference between asians and white strivers, also hispanics and poor whites. obviously upper class asian and urban black would be the biggest.

(http://www.autoadmit.com/thread.php?thread_id=3592014&forum_id=2#33133530)

The answer would probably be sex then, given that half the time you get a pretty meaningless same-sex pairing but half the time you get an opposite-sex pairing. Men and women are very different - you might get bigger differences between black poor and rich asian, or 140-iq versus 60-iq, but because the probability of that pairing coming up is so low you're better off just picking sex.

(http://www.autoadmit.com/thread.php?thread_id=3592014&forum_id=2#33133599)

it'd be incredibly unlikely to draw two people at random where one has an IQ even 4x the other.

income even if you draw the top 10% against bottom 10% that's a massive relative difference. not even sure it'd make sense since the bottom of the income distribution without government assistance is 0. is a middle class person infinitely better off than a poor?

(http://www.autoadmit.com/thread.php?thread_id=3592014&forum_id=2#33134974)