Search the Lounge

Categories

« The Musical Side of the Sears Bankruptcy Story [UPDATED] | Main | Did Chief Justice Marshall Suborn Perjury in Marbury v Madison? »

October 23, 2018

Comments

Feed You can follow this conversation by subscribing to the comment feed for this post.

Matt

I wish that the dubious nature of randomly testing people for things where false positives are not implausible was better known in general. Here in Australia, there is _massive_ random testing of drivers for drunk driving and for using drugs while driving. (In the year and a half I've lived here, I've been tested at least 10 times as much as I have been in my whole life before that in the US.) These tests are often given in the morning (around 10am) or early afternoon (1pm or so.) These don't seem to be times well designed to really be picking out true positives. The tests are surely not 100% accurate. Even if they give follow-ups, it's a huge waste of people's time, causing lots of stress, and probably giving some unfair punishment. All either because people don't understand probabilities or because they don't care.

Ellen Wertheimer

What is the impact of false negatives? I cannot imagine that tests that have a high percentage of false positives would not also have a significant percentage of false negatives.

Steve L.

I am sure you are right, Ellen. I believe that tests are designed, to the extent possible, so that false positives (which can be checked) are more likely than false negatives.

And of course, they can only test for illnesses and conditions that have been identified. Perhaps the greatest false negative is when the physician says "there's nothing wrong with you," when often it should be "we have no test that can determine what is wrong with you."

The comments to this entry are closed.

StatCounter

  • StatCounter
Blog powered by Typepad