Background/how this came up:

A student has numeric ratings on a variety of items in questionnaires and sports performance data and would like to model the performance on questionnaire scores. He wanted to do an analysis (don't ask me why) where some of the data are split into quantiles so he can compare high, medium, and low performance groups. I suggested he do a linear regression instead so that he wouldn't lose power by throwing away the continuous data - but I realized that the only papers I've seen on this being bad were about doing median splits.

The question:

Intuitively, it seems that quantizing your data will sacrifice power (based on arguments against median splits)- but how is this affected by the number of quantiles you choose to use? Are there circumstances where using, say, quartiles isn't much worse than doing a regression? For a given sample size, can a minimum number of quantiles be calculated to stay above a given power level?

  • $\begingroup$ The answer certainly depends on the distribution of the continuous variable you're discretizing (by quantile splits, apparently). What kind of distribution does the continuous measurement have? $\endgroup$ – Macro Mar 2 '12 at 3:06
  • 2
    $\begingroup$ What is the conceivable motivation for using quantile groups? I can't imagine how this would either improve the analysis or shed light. You are right to believe this will sacrifice a good deal of power in general. By the way, his definition of "low performance group" will change arbitrarily if new subjects' data are added, if using quantiles to stratify the subjects. $\endgroup$ – Frank Harrell Mar 2 '12 at 3:54
  • 5
    $\begingroup$ Not an answer to your question, but my currently favorite answer to the underlying problem: When one has a continuous predictor variable, traditionally we are taught to either assume its effect as linear (losing power to detect non-linear effects) or quantize it (losing power to detect linear effects). An elegant alternative is generalized additive modelling (gam), which permits the data itself to determine the degree of non-linearity that the model attempts to fit. The mgcv package in R implements gam with cross-validation to avoid over-fitting, and has revolutionized my work. $\endgroup$ – Mike Lawrence Mar 2 '12 at 4:14
  • 3
    $\begingroup$ Even simpler is to use regression splines. $\endgroup$ – Frank Harrell Mar 2 '12 at 18:02
  • 1
    $\begingroup$ One major “reason” to do it is that it seems simpler for someone who has had years of exposure to ANOVA (learning, teaching or simply reading articles using it), does not know much about statistical modeling or the GLM and needs a p-value ASAP (e.g. most psychology students or researchers). Psychologists work very hard to make everything a 2x2x… ANOVA design. $\endgroup$ – Gala Mar 6 '12 at 8:56

Quantile regression is a type of regression that aims at estimating either the conditional median or other quantiles of the response variable. This kind of regression is desired if conditional quantile functions are of interest. To see the advantages and applications of quantile regression, I refer to this wiki page. You may want to use this type of regression in your work.

| cite | improve this answer | |
  • 1
    $\begingroup$ I think you confused quantile regression (with regard to estimating quantiles of Y; can be a good idea) with quantile grouping on X. $\endgroup$ – Frank Harrell Mar 2 '12 at 13:53
  • $\begingroup$ @Frank Harrel: RobTeszka wants to split questionnaire scores, for each person, into quantiles such that can compare their performances according to a few groups. So, I think it is quantiles of Y. If I am wrong, let me know please. $\endgroup$ – hbaghishani Mar 2 '12 at 16:35
  • $\begingroup$ Estimation of quantiles is different from using quantiles as boundaries to create artificial, arbitrary groupings of inherently continuous variables. Quantile grouping is analogous to classifying a person's wealth based on their net worth and the net worth of their neighbors around them. Looking at it another way, suppose the net worths of people in a sample is $145k, $145.1k, $145.2k, $145.3k. Cutting at the lower quartile would cut at $145.1k making the analysis only relative but on a scale that is likely to be nothing but noise. $\endgroup$ – Frank Harrell Mar 2 '12 at 18:01
  • 1
    $\begingroup$ Thank you to the community member who flagged this reply for attention. Please note that a good way to react to replies you believe are wrong or misdirected is through appropriate voting: this helps future readers get a sense of the reaction of the community as a whole. Therefore, please upvote questions and answers whenever you can and do not be afraid to downvote answers (and the occasional question) when it seems called for: this helps make your opinion known to all and enhances the value of every thread here. $\endgroup$ – whuber Apr 1 '12 at 18:53
  • 2
    $\begingroup$ @Macro, a stock reason for declining a flag (as supplied by the SE team) is "flags should not be used to indicate technical inaccuracies or an altogether wrong answer." For more info please read the SE meta posts at meta.stackexchange.com/questions/80269/… ("If an answer is simply low quality, please just downvote it instead of flagging") and meta.stackexchange.com/questions/10848/…. If you have additional questions, let's start a meta post. $\endgroup$ – whuber Apr 3 '12 at 5:02

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.