Reply #15: Other than the ethical considerations, [View All]

it doesn't work. Far too many false positives. It's like the idea of requiring AIDS testing for marriage licenses. The disease is so uncommon, that even if you have a very good test, say 99.9% sensitivity and specificity (AIDS tests are pretty close to that good; terrorism tests aren't even in that universe) the false positives completely overwhelm you. I've seen this argument presented occassionally (ie the point I'm making) especially against all the airport nonsense, but most people really don't get it. If you like I can show you my trusty teaching slides on the topic, but try it yourself. Make up a 2x2 table, say with a million people in it. SPlit them up in columns of terrorists and not terrorists. Assign that split on whatever portion of the population you think are terrorists. Then for the rows, decide how good your screening test is. Split the same million people up on terrorist test positive and terrorist test negative. Now you've got the outside margins of your 2x2 table. Then fill in the 4 cells of table - true positives upper left, false positive upper right, false negatives lower left, true negatives lower right. It'll be very revealing. You've got to have a fairly common condition before a good test works. And a crappy test never works.