Hacker News Reader

Comment by diego | Hacker News Reader

Anecdotal evidence + overgeneralization + bad usage of statistics. For example:

"If we use wealth as a measure of success, Forbes’ January 2012 edition reveals that about 85% of America’s 400 Richest People (shown left) do, in fact, have a college degree."

So there is a correlation between extreme wealth and college education. What does that show? Do you account for the fact that rich families have a much easier time sending their kids to college? If you flip it and say "15% of America's 400 richest people don't even have a college degree" that sounds impressive as well. What did you expect? What does that have to do with startups?

I abhor these empty advice posts based on a single data point. I wouldn't give out generic advice about whether to stay in school or not, that's a personal decision that completely depends on the circumstances.

View on HN

My thoughts exactly.

Instead of using misleading correlations, the article could have just said:

"If you come from a rich family, you can do pretty much whatever you want."

That'd be more fun to read.

Fair point about a single number not painting the whole picture. Any thoughts on what stats might show a better correlation?

Another datapoint was an informal one-year follow-up to the first wave of Thiel's 20 under 20 (http://qr.ae/14o5S). Of the select few that were chosen based on their skills/ideas/potential and given access to Thiel's network, quite a few ended up actually going back to school.

Your argument is something like, "If you want to have a successful startup, you should complete college, because most of these hyper-successful people have completed college."

You then say the probability of having a college degree, given that you are on the Forbes 400 list, is 85%. That doesn't necessarily tell you the answer the question "What is the probability of making it to the Forbes 400 list, given that you have (or don't have) a college degree?"

Let's think about what you are trying to do. You want to take the population of college attending startup founders (or future founders), and make a prediction about which members of that group will be hugely successful based on whether or not they complete college. Let's say you run a survey of your readers now and then follow up with them 10 years from now. This would allow you to calculate several things:

* The probability of great success with college completion.

* The probability of great success without college completion.

Do you think those numbers are going to be appreciable? Do you think they'll be similar in magnitude? I suspect they'd be much smaller than 1%. In other words, I don't think that college completion is a strong predictor of great (Forbes 400 level) success. You haven't really established a causal relationship between finishing college and becoming wildly successful.

I suspect (and I think this is the point the GP is making) that your last name (is it Walton? or Pritzker?) or your parents' income would be much stronger predictors of this type of success.

I'd love to see socioeconomic status pulled out of the equation.

i.e. Take 1000 offspring of middle/upper class folks. Measure the difference between those who chose to go to college and those who did not.

Measuring the economic success of people who don't go to college is (most of the time) measuring the success of people who can't afford to go to college.

Ok, so lets say you did that. Do you have a hypothesis about what might be different?

College is actually reasonably affordable, there are accredited four year university degrees that cost about $25K over four years. Generally those are 'state' schools but still $25K it what it costs to buy a car, so if one can choose between buying a car and going to school we could do some interesting comparisons there.

In every survey I've ever seen, whether it was done by the Census bureau or the chamber of commerce, people with college degrees were more likely to have better economic success than someone who had not completed college. The numbers are quite skewed toward college graduates with STEM [1] degrees.

By and large, today, for the general case, if you want to increase your chances for economic success, getting a STEM degree from an accredited college is your most highly leveraged investment.

[1] STEM - Science Technology Engineering Mathematics

Coming at this 7 days late, but...

The median household income in the US is $49k, I believe. 40% of households spend more than they bring in and have considerable consumer debt. Of course, 50% of households make LESS than $49k income and 2+ kids.

Here's a list of state schools with instate and out of state tuition: http://www.kiplinger.com/tools/colleges/

This is their "best values" list. Nowhere is a $25k over 4 year school that I can see.

You'll need to account for living expense, textbooks, etc., for those 4 years as well (or maybe you're assuming that our hypothetical student can find a supporting job while going to school-- note that youth unemployment is approx. 2X the national average, and that the cheaper schools here are in states with generally higher unemployment).

Saying that college is reasonably affordable for most americans is just crazy.

All that aside, my hypothesis is that college impacts success (STEM-degrees having a higher impact), but that socioeconomic starting point has more of an impact. When you have three things that are highly correlated (success, starting point, and education) it's really hard to argue for causation with controlling for the third variable. When I look at all of my high-school peers who didn't finish college, they're all employed and making great money (I went to a mediocre private school in Maryland), but half of contemporary college grads (presumably most near the median background) can't find jobs (http://www.cbsnews.com/8301-18563_162-57434159/half-of-colle... ).

That's not really that surprising though... If someone drops of out school for the 20 under 20, the chance that they'd go back to school if they fail is incredibly high. So then unless all 20 of them are guaranteed not to fail, you'd expect to see "quite a few" going back to school.

I guess was looking at it more like of those that dropped out, how many ended up not succeeding, with the correlation implying that dropping out wasn't the right choice.

Agreed that there are quite a few confounding variables though. Any stats majors out there that know how to adjust for socioeconomic status + other biases?

Reply on news.ycombinator.com