Concussion Tests and a Voting Mess [538]

Here’s the roundup of what I’ve been working on at FiveThirtyEight over the last week:

The ‘One Person, One Vote’ Case Relies On Statistics That Nobody Has

The plaintiffs are challenging the usual method (counting total number of people living in a district) and are asking that states use the total number of eligible voters instead. The trouble is, we don’t have robust statistics on the number of eligible voters. If the Supreme Court were to set new standards for districting, we would need to overhaul the nation’s statistics and surveys.

I was kind of surprised, once I started digging in, that we don’t have these numbers — just a series of fiddly approximations that probably don’t match our standards for election law.  Even the more exhaustive attempts to figure out how many eligible voters there are throw up their hands at a certain point:

In a 2001 paper in the American Political Science Review, Michael McDonald and Samuel Popkin suggested two other possible factors to consider when estimating the total number of eligible voters. First, some states have residency requirements that might make people who have recently moved ineligible. Second, citizens can lose their voting rights if they are ruled mentally incompetent. McDonald and Popkin thought both of these factors were more trouble than they were worth to model: The researchers estimated that only 1 percent of the voting-age population failed to meet residency requirements and that only 0.1 percent were mentally incompetent to vote.

If the Supreme Court requires eligible voter estimates for districting, they may need to rule on whether McDonald and Popkin were right to exclude those groups — and on a host of other methodological questions — as the states scramble to commission constitutionally compatible surveys.

You can read the whole thing at FiveThirtyEight, and, if you like, you can head there by way of SCOTUSblog, which linked to my piece!

Then I wound up covering sports, through the lens of epidemiology:

We Don’t Know How Well The NBA’s Concussion Tests Work

So, if the tests are pretty accurate, why wasn’t Thompson identified as possibly concussed until after the game?

It’s hard to say, especially without information on the NBA-specific protocols, but sometimes tests are less accurate in real life than they are in a trial. Or a concussed player could just be unlucky; 5 percent or so pass a screening test in error.

If Thompson hadn’t mentioned that he felt ill, questions might not have been raised about whether the test failed. Players are a check on the test, but they may not feel free to speak up if they’re passed in error.

Continue reading at FiveThirtyEight…

And here’s a little bonus paragraph that didn’t really belong in the article (since the NBA does use baseline tests in their concussion protocols), but I was just interested personally in how well the tests did when you didn’t have a pre-concussion result to compare to:

Both the SCAT and the CCAT are best when administered as a before-and-after test, letting doctors compare the player’s post-injury score to a baseline from the beginning of the season. If a baseline isn’t available, both tests are a lot weaker. The SCAT sensitivity drops to 83% and the CCAT to 69%, making both much more likely to miss concussions.

And, if you’re wondering where you get concussable people to run tests on (the answer is not that the scientists thwak them on the head)… most of the studies I read for this piece were done on rugby players.  Now there’s a grim warning about a sport — “Number one choice of scientists who need concussion victims!”