Why in case-control studies, cases and non-case are taken from "Two-different Population?"
Why don't they come from a single population?
$\begingroup$
$\endgroup$
2
-
$\begingroup$ Perhaps you could add some explanation & a reference. The trivial answer that cases are sampled from the population of cases & controls from the population of controls, the two populations together forming the single "at-risk" population you're interested in, may not be appropriate. $\endgroup$– Scortchi ♦Commented Dec 12, 2014 at 12:55
-
$\begingroup$ I edited your title so it to be more precise. $\endgroup$– TimCommented Dec 12, 2014 at 13:02
Add a comment
|
1 Answer
$\begingroup$
$\endgroup$
4
Let's take the definition of population in statistics:
population is a complete set of items that share at least one property in common that is the subject of a statistical analysis
this is a quote from simple definition on Wikipedia. So, you define your population of interest precisely and then sample from this population. Comparing cases to controls means that you want to compare two samples taken from two populations, e.g. population of marijuana smokers, to population of non-marijuana smokers. This is a simplified answer, however as it is a self-study question, I hope it'll lead you a little bit further.
-
$\begingroup$ Why isn't it like that : from a single population , i separated the marijuana smokers and non-marijuana smokers. And then investigate whether my interested risk factor are more prevailing in the marijuana smokers group to non-marijuana smokers ? $\endgroup$– timeCommented Dec 12, 2014 at 18:47
-
$\begingroup$ Hint: Refer to the definition of population. You could also go through some textbooks on introductory statistics to find out more on how population is defined and what is the reasoning behind it. $\endgroup$– TimCommented Dec 12, 2014 at 18:51
-
$\begingroup$ that is, there is no common characteristics between marijuana smokers and non-marijuana smokers. so they can't come from same population. $\endgroup$– timeCommented Dec 12, 2014 at 19:41
-
$\begingroup$ No, rather: you define the two populations you want to compare (e.g. people who have a name starting with "Z" vs. others), than during the analysis you check if there are any differences. It is your arbitrary choice. $\endgroup$– TimCommented Dec 12, 2014 at 19:44