Both of these are great examples Angeline. The Paterno authors define their construct as: "a woman’s positive and negative emotional outlook toward potential pregnancy including the extent to which she desires to achieve or prevent pregnancy". It seems like it would be useful to see the whole distribution of this variable, as it might be quite bimodal (some women basically don't want to get pregnant and some women do), in which case it is a slightly surprising validation approach to treat it as a continuous variable. Still overall the face validity for the measure seemed convincing and the correlation with effective birth control strategies seemed like a reasonable construct validation.
It seems like the Kim et al paper is using one of the measures of pregnancy intentions that are criticized in the Paterno article? So the measurement challenges Paterno notes would be relevant for interpreting the Kim paper. How could lack of reliability or validity with the NSFG measure influence their results?
I agree that for comparing probability of unintended pregnancies it would be very helpful to have both absolute and relative #s.