Validity and reliability of dependent variables: Considerations for construction safety researchers

The issue of underreporting of construction related injuries and illnesses to government agencies and common interest groups has developed into a plague. The aim of this paper is to draw attention to this problem and offer recommendations for construction researchers. How do we know if interventions are effective if we use insensitive and unreliable measurements as dependent variables? How do we know what works, what does not, and what is most effective at reducing risk and enhancing safety? How can we compare across countries, companies, or trades? Descriptive and intervention research requires both valid and reliable measurements. Construction safety and health research, like all occupational safety and health research, relies heavily on lagging indicators such as injury and illness rates as dependent variables and descriptors. For quite some time, the validity and reliability of injury and illness statistics has been called into question. Poorly developed and managed incentive programs and the desire to look good have caused injury and illness rates to plummet, whereas fatality rates have not experienced the same reduction. However, the effect of this phenomenon on researchers has not been given sufficient attention. Through a review of literature and research methodology guidance we evaluate the use of lagging indicators as dependent variables in construction safety and health research. The practical implications will provide for a debate amongst conference participants and readers. The social implications are wide as we will reveal the professional and ethical implications for practitioners and researchers.