An empirical study of ordinal condom use measures.

Accurate condom use assessment is critical in sexually transmitted infection-prevention research. Ordinal condom use frequency measures may be problematic due to subjective interpretation by respondents. To assess this potential bias, we examined the ordinal condom use labels (e.g., "never," "rarely," etc.) assigned by college students to 17 scenarios that described how frequently a hypothetical couple used condoms. Scenarios varied by condom use frequency (0% to 100%) and by whether frequency was described by stating the number of protected acts out of 20 total acts of intercourse, out of 100 total acts, or as a proportion of acts that were protected. There was substantial interpersonal variability in the labels assigned by participants in all 17 scenarios and inconsistent use of the "never" and "always" labels to characterize 0% and 100% condom use, respectively. The assigned labels varied as a function of the number of total acts (20 vs. 100) and whether condom use was expressed as a count or a percentage, which suggests that participants did not convert the number of protected acts to an equivalent proportion. These results call into question both the reliability and validity of ordinal condom use measures.