Hello Denny,

Since you're using what is apparently existing data, the choice to use an amended version of measure was obviously beyond your control as a researcher.

This is not an unusual situation. I suspect that one reason for such changes is that the subsequent researchers believe that the amended wording may be more universally understood by respondents. You could:

1. run a pilot study, comparing the two versions. This could take a variety of forms, such as having a small group of respondents similar to the target population evaluate each pair of stimuli, and ask them to use the "think aloud" method to explain whether they would respond differently to the two versions. Alternatively, you could just each respondent version "A" and version "B", then compare the scores directly. Or, you could randomly assign respondents to version and then compare response distributions by item pair (and item intercorrelations). These last two methods would require more participants.

2. Go ahead and use the amended version data, since it has apparently been collected. Then, run estimates of score reliability and factor structure (using CFA), to see whether the apparent structure of the established instrument appears to hold with the amended version.

3. Have instrument development expert(s) review both versions, and indicate whether they consider the versions comparable in meaning. Any instances of incongruence should be noted in your write-up.

Good luck with your work.