[Update: Sam Wilson weighs in.]

The noble Sam Wilson pointed out a problem with this morning’s abortion regressions.  My dependent variable, ABORTION, was constructed by another GSS user.  This user coded people who weren’t asked about abortion as “no’s.”  The true yes/no split on “abortion for any reason” is 41/59.  Here are the corrected regressions using ABANY as the dependent variable.

Regressing abortion views on sexual partners, age, age squared, year, and sex:

abortfixed1.jpg

Adding a control for ideology:

abortfixed2.jpg

Adding controls for church attendance and Biblical literalism:

abortfixed3.jpg

Main change: The absolute magnitude of all the main coefficients rises by roughly 50%, leaving the relative role of ideas and interests about the same.  Main difference: Number of male sexual partners now looks about 50% more important than number of female sexual partners. 

Still, I messed up, and I apologize.