# Measure of central tendency for an ex-gaussian distribution

I know there won't be a clear answer to that question but I'm really curious to know your opinion on that matter. I deal with reaction times, and finding a good measure of central tendency is difficult due to the ex-gaussian shape and the outliers. Beyond the mean and median, I discovered recently the trimmed mean and winsorized data (that alter the shape of the distribution a lot since outliers are replaces by lowest and highest values).

I'm not expecting to have a clear answer here, I'm just curious to know the solution you guys have maybe found to deal with this kind of distribution.

• What do you mean by "measure of central tendency"? Do you mean estimates of the parameters? or how one should predict a future value, given a sample?.... Put another way, what would you use this measure of central tendency for? – probabilityislogic Oct 9 '11 at 4:07
• No i mean a measure to resume my datas in these particular kind of distribution, like the mean or median for a normal distribution. – user5084 Oct 9 '11 at 9:05

1. Quantify the distribution via quantiles (typically seq(.1,.9,.2)) and add quantile to the list of fixed effect variables in your analysis. This requires that you have some method for dealing with the continuous-yet-likely-nonlinear nature of quantile as a variable; I like generalized additive mixed effects modelling for this purpose.