When comparing the prevalence of cataracts among individuals with zoster to the prevalence among those without zoster, not accounting for diabetes status would tend to cause the crude cataract prevalence ratio to underestimate the true prevalence ratio. (Assume no additional sources of the following: bias, confounding, effect modification or measurement error).