Due to the nature of survey data analysis, an analysis using the SURVEYREG procedure requires more memory than an analysis using the same regression model in the GLM procedure. For details about the amount of memory that is related to the modeling, see the "Computational Resources" section of the GLM documentation.

The additional memory that is needed in the SURVEYREG procedure to handle the survey design is described as follows:

Let

• H be the total number of strata
• n_c be the total clusters in your sample across all H strata if it's cluster sampling
• p be the total number of parameters in the model

The additional memory needed is (in bytes):

48H + 8pH + 4p(p + 1)H

For cluster sampling, the additional essential memory needed is (in bytes):

48H + 8pH + 4p(p + 1)H + 4p(p + 1)n_c + 16n_c

Other small amounts of additional memory are also used by the SURVEYREG procedure. However, when you have a large number of clusters or strata, or a large number of parameters in your model, the memory that is described above will dominate the total memory usage in the procedure.