Agreement between self-reports and on-site inspections of compliance with a workplace smoking ban.

Introduction: This study compares self-reports on compliance with a workplace smoking ban with on-site inspections of the same workplace, in the Netherlands, to assess the validity of self-reported compliance by employees.

Methods: A total of 360 companies had participated in the telephone survey (in October and November 2006) and were also visited by inspectors directly after the survey to establish compliance. The sampling frame included companies with 5 or more employees, stratified according to the number of employees and type of economic activity. We calculated the agreement, the under- or overestimation and the predictive values, and explored nonresponse research.

Results: The percent agreement on compliance between the two measures was 77.5%, the McNemar test was not significant, and the agreement coefficient with first order correction was .68, indicating moderately strong agreement. Furthermore, the results indicate a slight overestimation of compliance. Concerning the predictive values, we found most variance among the self-reported noncompliance: 55.2% of those reporting noncompliance did in fact comply.

Conclusions: This study allows to conclude that self-reports on compliance with a workplace smoking ban are largely valid and that social desirability is negligible. For agencies enforcing the workplace smoking ban, these results indicate that a strategy to identify noncompliance among responding companies might be useful. Moreover, such a strategy reduces the burden of inspecting among complying companies.