Statistical Disclosure Control

[Sme]

Does SAS offer anything in the way of Statistical Disclosure Control?

[Doc_Duke]

There is no "PROC SDC" if that is what you are asking.  There are multiple ways to address the issues of Statistical Dissclosure Analysis using SAS.  You can use UNIVARIATE and FREQ to look at distributions and joint distributions to understand where you might need to trim data or combine categories.  You can fuzz the data with a jittering macro.  If you want to go further, you can create synthetic datasets using PROC MI with instructions for the recipient on how to use MIANALYZE to put the data back together again.

IMHO, the hard work in Statistical Disclosure Control is in the analyses in order to understand the controls that need to be put into place.  Statistical Policy Working Paper 22 ( http://www.fcsm.gov/working-papers/spwp22.html ) is a good introduction on that.

Doc Muhlbaier

Duke

[daynebatten]

I realize this is an old thread, but for the benefit of those finding it on Google... There are obviously a lot of ways to accomplish statistical disclosure control (fuzzing, small cell suppression, etc.). All of these can be accomplished in SAS, even if they aren't implemented in actual SAS features.

If you're interested in small cell suppression specifically, I did a significant amount of work implementing various algorithms for small cell suppression in SAS while working as a government analyst. I've (rewritten and) published macro code for these algorithms for the benefit of others that might be interested in using them. There's two methods I've specifically implemented in macro code:

1. An optimal method that uses PROC OPTMODEL to determine the least-costly pattern of secondarily suppressed cells necessary to protect your data. Details here: Optimal Cell Suppression in SAS - Final Macro
2. A heuristic method that uses a simple algorithm to find an acceptable (but not necessarily optimal) solution. This is useful if you don't have a license for SAS/OR, or if your data set is large enough that PROC OPTMODEL is choking on the size of the optimization problem. This macro will work on pretty much any set of data you can sort via PROC SORT. Details here: A Heuristic Cell Suppression Approach in SAS Macro Code

I hope somebody ends up finding this useful!

[byCRC]

"Hello, yes, those are old and discontinuous responses even though the topic is very current. I need to start working on that subject. I usually process data for users and then perform Statistical Disclosure Control using a method simpler than 'local suppression', which works well for tables. Now I need to apply Statistical Disclosure Control to databases, and I'm not sure which software or methods to use. Any ideas?"