Estimating the Percentage of Workforce Health Programs in Ontario from Results of a Mail Survey at Different Follow-Up Stages

Abstract

Selection bias due to nonrespondents is a common concern in mail surveys. The authors developed a procedure to estimate characteristics of nonrespondents based on when subjects respond, using data from a survey of personnel managers from large Ontario companies. The study was designed to assess the nature and extent of employee assistance programs (EAPs) and health promotion initiatives. Results indicated that personnel managers who responded quickly were significantly more likely to have an EAP than those who responded later. Estimates of the proportion of nonrespondents with EAPs were made by extrapolation, based on when people responded. The estimated proportion of nonresponders with EAPs was nearly identical to the actual proportion as determined by follow-up telephone interviews. Analyses also showed that companies with EAPs were significantly more likely to have health promotion programs. Extrapolation procedures were therefore used to estimate the proportion of companies with these other health promotion initiatives.