I need to perform a conditional logistic regression, and unfortunately must use SPSS in this case. The data should be composed of groups, in which there is a subject and 3 matched control. Each combined group should be numbered serially, so it could be used as strata. Suppose I have column A:
This will do it:
compute group=trunc(($casenum-0.1)/4)+1.
Note that this will not work if the order changes or if there are groups with less or more than 4 lines.