You're only looking at one of the variables (the number and/or proportion of highly educated people). If the other variable (the number and/or proportion of wel

You're only looking at one of the variables (the number and/or proportion of highly educated people). If the other variable (the number and/or proportion of well-paying jobs) is abnormally low, the first variable can be quite low and still produce the same effect. It's the ratio that matters, not absolute numbers.