3
$\begingroup$

Could you please explain, what does "data dependent quantiles" mean, as mentioned in the following paragraph:

Quantiles appear intuitively appealing to epidemiologists as they can be thought of in terms of low, medium and high risk groups. Moreover, the association between exposure and outcome can be described in terms of a relative risk between these groups. However, these perceived benefits are outweighed by several important problems that arise when a continuous variable is categorized, particularly if data dependent quantiles are used to form categories.

Many Thanks...

$\endgroup$
3
$\begingroup$

I believe the intent is simply to describe the situation where specific quantiles of the data are used to form groups, such as dividing the data at its median (a "median split") to get two groups, or putting the highest third of the data into one group, the lowest third into another group and all the middle third into a group.

This is distinct from a split where the dividing line is determined not by a quantile of the data at hand, but on some other criterion (such as some externally determined standard of what divides high from low).

$\endgroup$
  • 2
    $\begingroup$ +1 -- as a side note it is also possible to have something akin to "population quantiles" for a reference population, such as birthweight distribution across infants. $\endgroup$ – James Stanley Apr 1 '15 at 20:42

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.