Quantile regression GU

Skewed data, ordinal or nominal data is common in medical research. For example when measurements are not following a Normal distribution or questionnaire based scales is used. One way to adjust for the skewness is to transform the variable (e.g. logarithm). This can lead to problems with the interpretation of the estimate. For example is a median of a log-transformed variable equal to the log of the median of the un-transformed variable (invariance). For a mean this is not true. In these situations quantile regression can be an option. Another area of use for quantile regression is calculation of reference intervals (an interval describing a normal population) or for analyzing quantiles of the response (e.g. analyzing how the relation between HbA1b and explanatory variables can differ for high and low HbA1c values).

Link to more information about the course

Organized by: Department of Medicine, University of Gothenburg
Nuber of HECS: 1.5
Number of course places: 7-30
Course administrator: Madeleine Retamales Toro, madeleine.glucksman@amm.gu.se
Course dates: week 17-19 (20 April – 8 May, 2015)
Dead-line for application: 21 October 2014,
but late applications can be sent to catrin.wessman@gu.se.