Mismatch between measures of “greenness”: Evidence from a field experiment on self-reports and observed behavior

We run a field experiment with n = 599 participants selected from the waiting room of a town hall of a medium-sized German town that demonstrates a clear mismatch between three measures of “greenness” frequently used by researchers: self-reports, donations to environmental charities, and product choice experiments. We find that an aggregate index of self-reports on 20 different past pro-environmental behaviors only weakly correlates ( r = . 09 * ) with an observed pro-environmental product choice. Our preferred own money donation task also exhibits only weak correlation with both a self-report index ( r = . 15 * ) and the product choice task ( r = . 17 * ), but as our experimental variation shows, this is only the case if the donation paradigm is modified to allow sorting between prosocial and environmental charities. Conversely, we show that correlations disappear for the donation task if it is structured without a competing non-environmental prosocial option. Our results are unaffected by whether the respondent is a student or working professional and robust to different ways of aggregating individual self-report items. Our results support the concern that pro-environmental behaviors are difficult to measure consistently with the instruments that are to-date most commonly applied by researchers, but that there is some room to modify experimental designs to produce measures that yield more correlated outcomes across instruments.