The Election Polls Are All Wrong!?!
What can the science of preventative screening teach us about election polling?
The Frontier Psychiatrists is not a politically themed newsletter. That being said, I love a good math-nerd opportunity. Today is one of those days! Today’s column is about “sampling” populations.
We use statistics in biomedicine because we usually do not plan on treating or testing the entire eligible population. If we did, we wouldn’t have to wonder what the chance was that the test result was incorrect. In medicine, we have the concepts of sensitivity and specificity.
Our imagined test has a margin of error, which we call sensitivity and specificity:
Sensitivity = TP / (TP + FN)
Measures how often the test correctly identifies a condition when it is present.
Specificity = TN / (TN + FP)
Measures how often the test correctly identifies the absence of a condition.
At a large enough population size, even a very sensitive and specific test, we will have enough false positives and negatives that it won’t correctly catch every case of illness, nor will it screen out every healthy person …