This is survivorship bias in action. If you compared all university-educated developers to all self-taught developers, you might have a valid comparison. If yo

This is survivorship bias in action. If you compared all university-educated developers to all self-taught developers, you might have a valid comparison. If you only compare startup-employed university-educated developers to their self-taught brethren, you run into selection bias issues. For one thing, you are discarding all unemployed self-taught programmers (who may be very motivated, but not very good). Another issue is that university-educated programmers have various other characteristics (such as greater conformism) that reduce the likelihood they will want to work at a startup, or be successful in that environment. Self-taught developers who get hired to 'good' startups are also likely to have achievements that made their resume more appealing to the interviewer than a university degree; this means the self-taught developers who get hired are probably more accomplished.