I live in Manhattan. In a related story, I hang out with a lot of lawyers. Enough that it’s pretty difficult for me to make generalizations about them.
However, I have noticed some commonalities in interests among my legally-inclined friends:
Stuff Lawyers I Know Like:
- Complaining about the job market
- Complaining about the absurd amount of debt they have
- Watching (& over-analyzing) Law & Order
Now, a caveat. Anyone who knows me knows I dislike very few things. I’m not judgmental and judicious, which somehow makes sense. But I hate Law & Order.
I can’t really put a finger on why I hate it. Maybe because it makes my friends suspicious of others to the point of paranoia. Maybe because it’s so locked into its formula that I can set my watch by when they find the victim and when they end up in court. Maybe I just don’t get it.
But I do know that, almost without exception, my lawyer-friends love the show. They say that they watched it while they were in college. And as they complain about the glut of lawyers who probably also watched the show, I wondered: are there too many lawyers because of Law & Order?
So being a data geek, I looked at the data.
The Law School Admission Council (LSAC) tracks info on how many LSATs are administered each year and how many applicants there are each year. Unfortunately, applicant data only goes back a few years, so I looked at the historical relationship between LSATs and applicants (for every 100 LSATs, there have been 60-65 applicants). I applied this relationship backwards as far as I had data, to 1987* to estimate applicants. So I get a history of Law School Applications:
There’s some literature that says that there is a relationship between the demand for higher ed and wages. It’s a little dicey but let’s say that it might hold for law schools. We’ll graph the annual change in GDP (scale on the right side of the graph, in %):
Looks like it’s not a totally crazy idea. We get an r-squared of .314**. A weak relationship, but ‘meaningful’. Still, notice that there are more applications than we might expect in the early 2000s.
Law & Order
Law & Order ran for 20 years: 1990-2010. In the beginning, it wasn’t really that popular. For the first 8 years, it couldn’t crack the top 20 shows on TV. That’s about as popular as Modern Family was in 2010. And then, in 1999, with no warning it took off. Which leads us to…
1999 was also the launch of Law & Order: Special Victims Unit. Combined, the two shows had almost 30 million viewers. In less than 5 years it would be more popular than the original. The 1999 season (which runs into 2000) matches up with the time people would be preparing for law school entry in 2001. Here’s a graph of law applicants vs. SVU viewers
Hmm. So does this improve our model fit? Let’s look at the model:
We now have an R-squared of .60, with significance for both variables (and a slightly lower AIC). We still have serious auto-correlation, though.
Translation for you lawyers: STOP WATCHING SVU! For every million people who watch the show per year, this estimates that
850 500 more people apply to law school than we would expect otherwise***.
EDIT: It was annoying me that I hadn’t accounted for auto-correlation, so I reran the analysis with a lag on Law.Applicants. This knocks the DW-stat and the impact of GDP just out of ‘significance’, though both, I think, are likely making an impact. (Note: I didn’t muck with the data to push these out of significance. Scout’s honor). Here are the results- now estimating 500 more applicants per million views.
Also, I meant to mention this in the original post, but 2001 was also the year that TNT relaunched as ‘We Know Drama’ TNT with L&O reruns on *all* the time: I don’t have any data on their viewers, otherwise I’d test that effect.
*This is the first of many questionable things I will do with the data here. I should say this is more for fun than anything. This involves an olympic long-jump of faith so don’t take it too seriously. At the very least, causality could go the other way as future lawyers go out of their way to watch their future selves kick ass on TV.
**For stat geeks: the relationship is highly auto-correlated, with a D-W stat of .66.
***I also checked a model that included the number of bachelor degrees awarded each year. To my surprise this performed worse by AIC and was less significant than SVU watchers (and just as auto-correlated).