This blog entry notes how easy it is to fool yourself into thinking that a screening program is doing well because it finds problems “early while you still have a chance.” It outlines problems with non-randomized studies that are prone to selection bias, length bias, and lead-time bias. It also explains what these biases mean in non-technical language.
Bastian H. “What’s so good about “early,” anyway?” Statistically funny blog. 2013-12-29. Available in html format
You can find an earlier version of this page on my blog.