Correcting bias in self-rated quality of life: an application of anchoring vignettes and ordinal regression models to better understand QoL differences across commuting modes

Melanie Crane, Chris Rissel, Stephen Greaves, Klaus Gebel

Likert scales are frequently used in public health research, but are subject to scale perception bias. This study sought to explore scale perception bias in quality-of-life (QoL) self-assessment and assess its relationships with commuting mode in the Sydney Travel and Health Study. Multilevel ordinal logistic regression analysis was used to analyse the association between two global QoL items about overall QoL and health satisfaction, with usual travel mode to work or study. Anchoring vignettes were applied using parametric and simpler nonparametric methods to detect and adjust for differences in reporting behaviour across age, sex, education, and income groups. The anchoring vignettes exposed differences in scale responses across demographic groups. After adjusting for these biases, public transport users (OR = 0.37, 95 % CI 0.21-0.65), walkers (OR = 0.44, 95 % CI 0.24-0.82), and motor vehicle users (OR = 0.47, 95 % CI 0.25-0.86) were all found to have lower odds of reporting high QoL compared with bicycle commuters. Similarly, the odds of reporting high health satisfaction were found to be proportionally lower amongst all competing travel modes: motor vehicle users (OR = 0.31, 95 % CI 0.18-0.56), public transport users (OR = 0.34, 95 % CI 0.20-0.57), and walkers (OR = 0.35, 95 % CI 0.20-0.64) when compared with cyclists. Fewer differences were observed in the unadjusted models. Application of the vignettes by the two approaches removed scaling biases, thereby improving the accuracy of the analyses of the associations between travel mode and quality of life. The adjusted results revealed higher quality of life in bicycle commuters compared with all other travel mode users.