1
$\begingroup$

SPSS Warning The large number of columns in the design matrix will cause integer overflow. The large number of columns may be due to too many levels in one or more factors, or to higher-order interactions among factors with many levels. It may also be due to too many factors This command is not executed. How do I resolve this?

$\endgroup$
7
  • 3
    $\begingroup$ Probably you indicated a continuous variable as a factor? $\endgroup$ – ttnphns Mar 6 '13 at 7:43
  • 1
    $\begingroup$ I googled the error message. It is SPSS. I've flagged it because it is yet another common sense case that shouldn't be migrated but that the current meta consensus says should be. A failure to migrate will be a tacit acceptance among the voting members that the current policy doesn't quite make sense. $\endgroup$ – russellpierce Mar 6 '13 at 17:52
  • 1
    $\begingroup$ May be we should have a tag for numeric issues in statistical packages. We can throw people off to math SE, but in all likelihood they will be lost. $\endgroup$ – StasK Mar 7 '13 at 15:59
  • 1
    $\begingroup$ @Russell If you would like to bring up policy issues, please start a discussion on meta. I deny your "tacit acceptance" assertion, though, because it appears to me this question is on topic here. The test of that belief is that there are good answers that rely on statistical reasoning or analysis: the OP needs to understand how software in general constructs design matrices and factors; the solution is likely to apply in any statistical computing environment, not just SPSS; so it does not appear to be an SPSS-specific question. $\endgroup$ – whuber Mar 8 '13 at 15:01
  • 1
    $\begingroup$ @whuber: That discussion is already on meta (meta.stats.stackexchange.com/questions/1514/…). There are no new policy issues to raise. However, I was highlighting the one(s) that are relevant. In this form, "the question cannot (readily) be stated solely in English and mathematical notation". In addition, the question can "be formulated in an agnostic manner so that anyone used to that programming language or software might provide a definitive answer without resorting to their statistical knowledge". $\endgroup$ – russellpierce Mar 8 '13 at 15:36
2
$\begingroup$

As @ttnphns suggests, you likely have misspecified your model. The only place continuous variables should be placed is under covariates. If you've double and triple checked this, then look at how many IVs (factors) you have and how many levels each IV (factor) has.

When you have a lot of IVs (or IVs with many levels) SPSS will tend to produce this error. As the error message says, there is a problem due to "due to too many levels in one or more factors, or to higher-order interactions among factors with many levels" or "It may also be due to too many factors". A google search on this error message shows a comment from David Nichols who a signature line that identifies him as "Principal Support Statistician and Manager of Statistical Support" where he implies that the design matrix is just too large.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.