I'm running the GENLIN procedure in SPSS Statistics and comparing results from Version 19 or earlier with those from Version 20 or later. In the earlier results I have a variable with a parameter estimate shown, but with a missing Wald chi-square statistic. In later releases the statistic is there. Why is this happening?
The most likely scenario here is that the predictor variable is on a large enough scale that its standard error is effectively 0 and the Wald test, which requires forming a ratio of the parameter estimate to its standard error, would involve dividing a number by 0. Beginning with Version 20, GENLIN performs internal linear transformations on predictor variables in order to avoid numerical precision issues like this, doing estimation on "nicer" transformed values, then back-transforming the results to the original scale. You may be able to obtain the same result for the test statistic shown in later releases by rescaling the predictor variable. Dividing a predictor by 10 moves the decimal place one unit to the right in terms of the scale of the parameter estimate (and accompanying standard error). Moving the decimal a number of units to the right may allow you to get the test. If interest truly centers on the original scale, you can then divide the coefficient and its standard error by the same power of 10 used to rescale the variable and obtain the same equation you get in later releases.
Rate this page:
Copyright and trademark information
IBM, the IBM logo and ibm.com are trademarks of International Business Machines Corp., registered in many jurisdictions worldwide. Other product and service names might be trademarks of IBM or other companies. A current list of IBM trademarks is available on the Web at "Copyright and trademark information" at www.ibm.com/legal/copytrade.shtml.