The Summary platform in JMP offers an easy way to report many statistics, over multiple “By” groups, in data table format. The Distribution platform offers up an even greater variety of statistics, and we have had some customers ask for an easy way to get a Summary-style report of these. We’ve also received requests to add trimming and Winsorizing capabilities to all statistics.
If you’re not familiar with trimming or Winsorizing, don’t worry — the concepts are pretty simple: Trimming removes data from the “ends” of the sorted data set, while Winsorizing replaces these data with less extreme values.
For example, consider the data {1, 2, 3, 4, 5, 6, 7, 8, 9, 10}. If we perform a 40% trim (perhaps a high percentage in practice but convenient for illustration), we keep only the middle 60% of the data, leaving us with the new data set { 3, 4, 5, 6, 7, 8 }.
If we perform a 40% Winsorizing, instead of removing the 1, 2, 9 and 10, we replace them with the minimum and maximum of the trimmed 40% data set. The result is the new data set { 3, 3, 3, 4, 5, 6, 7, 8, 8, 8 }. Winsorizing allows the extreme data in the original data set to contribute meaningful information to the analysis, while preventing it from having undue influence.
With the Extended Summary Statistics Add-in, you can compute any of the statistics on raw, trimmed or Winsorized data. The add-in has been designed with the look and feel of the Summary platform. So if you’ve used Summary, you’ll be up and running instantly. In addition to offering almost all of the statistics that the Summary and Distribution platforms offer, the add-in also will compute ranks, and presents the results in either a stacked or an unstacked (original) format.
I’ve placed the add-in in the “Stat” section of the add-ins on the JMP File Exchange. You can download it after logging in to your free SAS profile. Enjoy!
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.