The Population Division requested that the Center for Survey Methods Research staff conduct cognitive testing on newly-proposed question wording for the veterans questions for the 2003 American Community Survey. Staff conducted cognitive interviews on a shortened form of the self-administer version of the ACS. This report documents the findings of this research. The results showed that the new changes did not present a problem for respondents in the self-administered format. However, these changes are not directly translatable to a CATI format for the follow-up instrument, and additional testing of the CATI version is recommended.
The one problem that surfaced was the supplemental instruction in the period of service question. Some respondents in the military did not include all relevant periods of service; rather, they omitted response categories during which they only served a short period of time. As a result of this finding, we have recommended a change to the supplemental instruction to encourage reporting of all periods of service.