Affective risk perception index as a screening tool for construction workers.

• Affective Risk Perception Index (ARPI) was developed. • ARPI scores were analysed to create four categories of ARPI. • Cut-off score of the ARPI is 42 out of 60. • ARPI can be used to predict the risk-taking behaviour of construction workers. Affective risk perception has been receiving attention from occupational safety researchers. This study proposes the Affective Risk Perception Index (ARPI) to serve as a screening tool and provides a guideline to researchers and safety practitioners for the use of the ARPI. 536 construction workers participated in a questionnaire survey. ARPI scores were analysed using the 15th percentile, median and 85th percentile to create four categories of ARPI, namely 'poor' (0–20 points), 'moderate' (21–41 points), 'good' (42–55 points) and 'excellent' (56–60 points). Hierarchical multiple regression was used to examine the predictive value of the ARPI for the self-reported risk-taking behaviour of construction workers and compare the four categories of the ARPI in predicting such behaviour. Compared to personal factors (including perceived behavioural control, attitude towards risk-taking behaviour and outcome expectancy), the ARPI made the relatively smaller contribution to explaining the variance of the self-reported risk-taking behaviour of construction workers. Results showed that construction workers who were classified as good or excellent in the ARPI engaged significantly less in risk-taking behaviour than those who had a poor or moderate level of affective risk perception. Furthermore, no statistically significant differences were observed between poor and moderate levels of affective risk perception and between good and excellent levels of affective risk perception in predicting the self-reported risk-taking behaviour of construction workers. Therefore, the cut-off score of the ARPI is 42 and the ARPI serves as a practical screening tool for predicting the risk-taking behaviour of construction workers. [ABSTRACT FROM AUTHOR]