# Why don't I see fiducial confidence intervals in a probit regression using version 16 of SPSS?

Probit regression, akin to logistic regression, is used to find the value of a continuous covariate which corresponds to an estimated proportion (e.g. correct) of a response. Fiducial confidence intervals (Finney, 1971) can be obtained for the covariate value corresponding to a proportion. For example in acoustics you might be interested in the degree of sound (in decibels) where a person can hear roughly 90% of sounds after they have performed a hearing test at different decibel thresholds.

According to SPSS's bug database, there was a problem with the display of the covariate confidence interval table in probit, and this was fixed in the 16.0.2 patch which can be downloaded for free from here.

Quoting the bug notes:

Look at the Confidence Limits table. It appears to be empty. Only if you edit it and select a value do you see the value. Even resizing the columns doesn't help.

This can be fixed by the user by selecting all cells and formatting them in the cell properties tab.

(with thanks to Jon Peck of SPSS for pointing this out).

There is an SPSS macro for doing Probit regression as described above.

Reference

Finney, D. (1971) Probit analysis. Cambridge University Press. QA276.8.F55 in Betty and Gordon Moore library.