Re: SPSS warning from regression

Posted by Richard Ristow on Mar 06, 2007; 2:38am
URL: http://spssx-discussion.165.s1.nabble.com/SPSS-warning-from-regression-tp1074315p1074324.html

At 05:50 PM 3/5/2007, Yifan Lu wrote:

>I am running multiple regression, but SPSS keeps telling me:
>
>>Warnings
>>There are no valid cases for models with dependent variable alldays.
>>Statistics cannot be computed.
>>No valid cases found. Equation-building skipped.
>
>Anybody also had such problem before? Any suggestions to fix the
>problem? Thank you!

The default (and recommended) practice for regression is 'listwise
deletion': a case is excluded if any of the dependent or independent
variables is missing. The problem you describe can arise when you lose
all data because each case has some variable missing.

>I checked the data and it seems all right. I also ran descriptives and
>the results come out right.

Descriptives won't tell you enough; they give the statistics on single
variables, not on all of them together.

Look at your variables, especially for
. Any that are missing for a large proportion of the cases
. Any that may be mutually exclusive - if one is present, the other
will be missing

In principle, you could get your problem from plain bad luck: you don't
have much missing data, but by chance every case has some missing
variable. (In this case, missing value imputation may be warranted.)

More commonly, though, there's something in the structure of your data
that ensures you don't have complete data for any case. Look for that
first.

-Good luck,
  Richard Ristow