Allocator is an application of the Random procedure. The function of the Allocator is to randomly allocate "cases" (or objects or subjects) to "groups" (mathematicians sometimes prefer to say "blocks"). The allocator has been programmed to allow you to simulate basic statistical processes as well as to allocate larger sized samples to groups using pre-specified minimum group sizes. For example, you can simulate the throw of a dice by entering 0.16666.. in six boxes and by asking for a single case, i.e. throw. An optimum sample size and how to optimally allocate cases to an experimental and a control group can be calculated using the sample size or power procedures and then applied using the Allocator. For the technical details of the Allocator the user is referred to the Random procedure help page. A maximum of seven groups can be specified, from 'A' to 'G'. For a higher number of equally sized groups use the Order & Group procedure and if that doesn't satisfy you contact a SISA consultant.
A group is specified by entering the proportion by which you want the group to be represented in the final sample in the appropriate "Groups" input box. The total of the proportions should add to 'one'.
In the options box input the sample size you want in the "Numbers of Cases" box. If you want to allocate only one case this case is allocated to a group according to the probabilities specified in the Groups input boxes. With a high number of cases the procedure can be (very) slow. You might get a message box asking you if you want to abort the script because this might cause your computer to run slowly, answer "no" to this question, your computer will allocate the cases eventually.
For the function of the seed consult the Random procedure help page. If you want to redo an allocation you should note the seed you used.
For help on which method to use consult the Random procedure help page.
The results are presented with per row the case number, the group to which this case is allocated, and the random number on the basis of which this allocation has been done. If you check the "Present by Group" box the case numbers are presented for each group.
If you do not check any of the two "Group Size" boxes the proportions for each group in the resulting sample are approximate according to the probabilities specified in the Groups input boxes. The allocation of cases has been done on a fully random basis and the size of the groups has been randomly determined. If you check the "minimal" checkbox the program will continue sampling cases until each group has at least the number of cases in the sample which is equal to the proportion specified in the Group input boxes relative to the number specified in the Number of Cases box. Statistical power is guaranteed in this case and the procedure is fully random. The resulting sample size will mostly be higher than the number specified in the Number of Cases box. If you check the "exact" checkbox each group has exactly the proportion in the sample that is equal to the proportion specified in the Group input boxes. Statistical power is guaranteed but the procedure is not fully random.
For drawing a random number from any Uniform,Weibull, Normal, Gamma or Beta Distribution please consult the Random procedure help page.
For ordering cases randomly from 1 to N and to randomly group them into equally sized groups use the Order & Group procedure .
To pick n cases out of N with or without replacement use the n out of N procedure .
To order n cases into r groups use the Order & Group procedure .