late look bias versus selection bias

This forum made possible through the generous support of SDN members, donors, and sponsors. Thank you.

aashkab

Full Member
10+ Year Member
15+ Year Member
Joined
Dec 16, 2007
Messages
277
Reaction score
0
I'm confused between the two.

For selection bias, a case arises where if you send out surveys to those who survived an MI, its considered a selection bias.

But in late look bias, if you sent out surveys to people with fatal diseases, only those who survived will return it back.

Okay, I think I just answered my own question but is the best way to look at this is that there is essentially a pre made selection in selection bias before you send anything out while in late look you don't know what will come back to you in terms of results because you don't know the implication of the disease?
 
Look at it more globally because if you only look at them in terms of this scenario you may miss the difference in a different context.

A selection bias is one in which the investigators OR SUBJECTS are permitted to chose whether theyre in a control group or study group (placebo vs test whatever). So in your example, if the physician is only sending surveys out to those who survived MIs, he is selecting to only send surveys to the subject group with no control survey to compare against.

A late-look bias is different and it seems like you already understand this. There is the example you described, but it also applies in a situation in say a situation where you poll all mothers who had a folate deficiency during pregnancy. Those who had complications are far more likely to remember details than those who had nothing happen to them, and their memory is likely to impact the information you get back.
 
Top