Endogeneity or reverse causality can lead to biased estimation of effects and to inaccurate interpretations of cause and effect water-use relationships. However, biased estimations of the effects of designed interventions on water–related behaviour are rarely discussed. This study investigates the endogeneity of psychological factors in water-related behaviour using an instrument variable (IV) approach. Data from eight household water treatment (HWT) studies in Asia, Africa, and South America are utilized. A combination of several socio-economic characteristics, such as education and accessibility, is used as a control variable and three psychological factors, i.e., perception of risk, attitude, and social norms, are used as predictors of the adoption of HWT. Variables related to country level institutional quality are used as the IVs to remove the effects of reverse causality of households having adopted HWT on the psychological factors. Results show that psychological factors and HWT adoption indeed co-evolve, resulting in the emergence of reverse causalities, indicating that endogeneity exists in the water-related behaviour. Institutional quality indicators are found to be valid instruments for psychological factors such as attitude and norms towards HWT, but not households’ perception of risk in context of HWT. This suggests that institutional performance “heavily” influences households’ attitude and norms regarding water-related behaviour. The effects of households’ attitude and norms on HWT behaviour are underestimated by 59 and 40%, respectively, if the feedback effects of adoption behaviour on psychological factors are not considered. Further, households’ experience of using water-related technologies and external nudges, such as from policy makers, are more important drivers of the behaviour compared to risk perception. Despite the challenge of finding valid instruments for psychological factors, the study recommends that the effects of reverse causality addressed here need to be considered in future water-related behavioural studies. The conventional regression analysis approaches that ignore such effects are no longer suggested for analysing water-related behaviour.