What do I do when the program says Insufficient Workspace Memory?

Check one of the following:

1. You can increase the amount of memory that Gauss uses by increasing the number on the following line in the file gauss.cfg (file name is gsrun.cfg for the Windows version of Amelia):

     workspace =  50           # workspace size in megabytes
   

   Increase it as much as you like, but if you make it larger than your physical RAM, ${\mathfrak{A}melia}$ will use virtual memory by swapping to disk, and hence run much slower. (FYI: RAM is inexpensive these days.)

2. You may be using more memory than you think. The number of variables in the model is increased if you use _Amnoms, which creates dummy variables for all but one of the categories in the named variable. If you inadvertently name a continuous variable, you will be asking ${\mathfrak{A}melia}$ to create a very large number of variables.

3. Examine the descriptive statistics to verify that you have successfully gotten the variables you wanted, and in the right format, into ${\mathfrak{A}melia}$.

4. If any of your variables have an especially large fraction of missing observations, try leaving that variable out and running it again. Similarly, if you have a combination of variables for which listwise deletion would result in very few observations, try cutting out one of these variables and running it again. These steps will at least enable you to localize the problem.

5. To get a feel for how much memory you need, start by imputing just a few variables and then add variables gradually until you see a problem.

6. Verify whether you really need all the variables in there at the same time. If you plan to run separate analysis models with subsets of variables, you could impute the subsets separately.