# Economists Do It With Models

## Today’s Quiz: Can You Spot The Flaw In This Reasoning?

#### August 11th, 2010 · 23 CommentsEducation · Fun With Data

Ezra Klein posts the following graph:

The caption on the graph and the post is “The benefits of a college degree in one graph.” You mission, if you choose to accept it: Explain, in 200 words or less, what is wrong with this line of reasoning. (I’ve always been curious as to which readers have been paying attention. 🙂 )

I’ll give my own answer tomorrow.

Tags: Education · Fun With Data

### 23 responses so far ↓

• 1 Christopher // Aug 11, 2010 at 3:57 pm

Correlation is not causation. What if people with greater drive and skill are more likely to both pursue college degrees and jobs?

• 2 Merid // Aug 11, 2010 at 4:08 pm

Well, I think the reasoning is wrong and misleading because from the data that is given, we cannot be sure that those without diploma are without jobs in greater proportion for lack of education. It could simply represent cyclical unemployment affecting the lower end of the services and manufacturing industry. As Christopher eloquently demonstrated, it is what logicians call the cum hoc ergo propter hoc fallacy.

• 3 Daniel // Aug 11, 2010 at 4:14 pm

The comparison is flawed from the outset, since the number of subjects becomes smaller and smaller as you rise in level of education. There will necessarily be a smaller pool of those with BAs vs those with No Diploma. EX: Population = 1,000, Total BAs = 100, Unemployed BAs = 50 (5%), Total No Diploma = 500, Unemployed ND = 130 (13%). However, the proportion of Unemployed ND to Total ND is actuall LESS than that of Unemployed BAs vs Total BAs (130/500 vs 50/100).

• 4 Bo // Aug 11, 2010 at 4:29 pm

#1) As stated before correlation does not mean causation (see Christopher’s post where it explains)

#2) The graph assumes that employment and whether you are or aren’t is everything. However many would be highly concerned with the wage they would make in their field. Say 5% of teachers are unemployed and the median salary in the field is \$30,000 but 8% of hair stylists are unemployed and the median salary is \$50,000 (making these numbers up just for a hypothetical situation). Hair Styling is the more lucratice career even though it has a slighty higher unemployment rate but makes 66% more money.

#3) The graph also doesn’t account for what is lost in order to get college degree, 4 years in the work force, lots of money, etc. and if the slight gain in employment makes up for that.

• 5 Andrew // Aug 11, 2010 at 4:32 pm

What I am curious about would be a similar graph in boom years. While I feel like it follows the same trend, it may be drastically changed. Can we get some of THAT graphic content?

• 6 Tim McC // Aug 11, 2010 at 4:38 pm

Correlation does not imply causation, Ezra!

Richer/smarter/ambitious/lawful people are more likely to go to college. These same folks would be more likely to get jobs regardless of education.

• 7 Warren J // Aug 11, 2010 at 6:47 pm

Ha! I’ve been talking about this for years. Yes, more ambitious tend to get a better college education and tend to get better jobs. Whoopee!

Even more pronounced is the correlation between employment and the education of your highest-educated parent. So, the conclusion is that you increase employment by sending parents to school, right? Uhhh….

The correlation is even better between income and your parents education…

• 8 Naked Negro // Aug 11, 2010 at 7:18 pm

Data is derived from CPS, which does not disaggregate students over 16 years old. It’s also summer, so some of those students are presently HS graduates with no college and no summer jobs.

• 9 misterxroboto // Aug 11, 2010 at 8:26 pm

trick question. there is no flaw in the reasoning. attempting to find one is a genetic fallacy.

• 10 Dave M. // Aug 11, 2010 at 9:12 pm

cum hoc, ergo propter hoc…?

I thought it was called “post hoc, ergo propter hoc.”

• 11 Dave M. // Aug 11, 2010 at 9:15 pm

There’s also the fact that unemployment doesn’t count all those potentially underemployed grads and undergrads.

• 12 Naked Negro // Aug 11, 2010 at 10:52 pm

Also, the graph wasn’t put together by Demetri Martin.

• 13 Lisa // Aug 11, 2010 at 11:07 pm

I think Daniel is right on the money. You should be more concerned with the raw number of people unemployed than with the percentage of some subclasses here.

Also, I don’t think he’s assuming causation–he’s simply showing you the correlation between degree and population unemployment rate. I think it’s your readers that are assuming that by showing the correlation, he’s arguing for proof of something.

• 14 Seth // Aug 11, 2010 at 11:17 pm

Who are we trying to explain this to? If it is other readers of this blog then fine correlation doesn’t equal causation, but I think most people don’t understand what that means.

If it was my intro to Econ students I would ask.
Do they think they would have a better chance at a higher paying job with a college degree? They answer yes.

Then I ask in the 30 plus classes they take in college how much of what they learn do they think they’ll use in their job after they graduate? A little from Principles of Economics, but most of what they will need to know they learn on the job.

So what do employers learn from finding out someone has a college degree? That a person will show up, can do work, and is capable of learning.

It is not clear that the college taught them that, but the diploma verifies for the employer that they can do it. So it’s not the diploma that makes the person get the job, but the person who will eventually get the job because they work hard enough to get a diploma.

• 15 Viktor // Aug 12, 2010 at 3:26 am

Of course the fault is that they assume that it is better to have a job than to not have a job!

(Okay, not totally serious here.)

• 16 Naked Negro // Aug 12, 2010 at 7:09 am

One sentence summary of a few points raised so far.

The graph does not control for age.

• 17 Thorsten // Aug 12, 2010 at 9:27 am

The graph seems to suggest that going to college reduces your risk of unemployment. That may not necessarily be true due to endogeneity: good students are more likely to perform well on the job market; and they are also more likely to (successfully) pursue a higher level of education. Therefore college graduates are BY NATURE more suited for the job market and therefore less likely to be unemployed. For that reason we can NOT conclude from the graph that a given person can increase his/her chances of finding a job by staying in school.

• 18 Naked Negro // Aug 12, 2010 at 9:29 am

I take it back. The source data is limited to persons over 25 years of age.

• 19 BradyDale // Aug 12, 2010 at 10:28 am

Well, not that I want to discourage going to college, but I always figured it was a relative competition thing. It’s not like Unemployment would drop to 5% if everyone had a bachelor’s degree.

In fact, once upon a time, they just stressed everyone graduating high school, but then everyone did that and so it became “you gotta go to college.”

If everyone did that, it would be getting a Masters or a PhD that would give you a competitive advantage in the workforce.

That’s not to say we wouldn’t be an overall more productive economy if EVERYONE went to college, but not everyone CAN get thru college so it’s a bit of a red herring. And we need folks out there getting paid fairly for doing non-Bachelors Degree work.

We need a manufacturing sector and that sector needs unions. Sorry, I’m going a little off topic here, but it’s silly on its face to think everyone can go to college and it’s also silly to think that an economy can’t function if non-knowlege workers get paid enough to support a family. The liberals are wrong in the former and the right-wingers are wrong in the latter.

• 20 nohype // Aug 12, 2010 at 12:59 pm

If Ezra Klein is the good liberal that he claims to be and if he believes that college leads to higher income, should he not conclude that we should get rid of college because it is an inequality-generating mechanism of society? Certainly the state should not be supporting these inequality-generating institutions by subsidizing them, right?

• 21 Omar Ting // Aug 12, 2010 at 2:10 pm

It doesn’t tell you how shitty the jobs are nor how much they get paid. Maybe the 88% of non-diploma people make 10 x’s as much money as those with bachelor’s degrees. Unlikely but can’t be known with the given information. What are the benefits of a degree then? Graph only shows more likely to have a job, not the quality/pay of the job.

• 22 Warren J // Aug 13, 2010 at 12:24 am

I have to make another comment, here. Commenters are being rather presumptive regarding the source of this data. A careful check of the source reveals … wait for it … no source!

People are assuming this is the result of Summer, but if this was produced from the CPS (as suggested by some), the CPS averages results over the calendar year.

• 23 Naked Negro // Aug 13, 2010 at 12:27 am

Here’s the source data.