The recent paper by Rodríguez-Romo et al. studied the relationship between different variables of the self perceived urban environment and physical activity in Madrid.1 To this end, the study participants were asked in a questionnaire about their perception of the urban environment related to physical activity (density, access to public transportation, access to food stores, paths, bike lines, leisure-time physical activity areas, traffic safety, area security, social environment and esthetics), and in another questionnaire they were asked about levels of physical activity. The authors then estimated the probability of having moderate to high levels of physical activity given certain environmental characteristics as reported by each participant.Research efforts trying to understand how environmental features may promote active transportation, not only leisure physical activity, is currently a relevant topic offering insight for possible political action.2

The use of self-reported measures is one of the approaches that can be used in the study of urban environment and health. It represents an unexpensive and feasible approach that can provide useful information about the environment. One of the problems associated with the use of self-reported for both result and exposure variables, is the possibility of same source bias.3–5 Same source bias may result in spurious associations between exposures and outcomes because the measurement error between the two variables is correlated. To be more precise, exposure self-reporting may be influenced by the self-reported outcome. For example, persons who are not physically active may be more likely to perceive their environment as less favorable for physical activity, regardless of the actual condition of the urban environment.

Given that the article by Rodríguez-Romo et al. models the probability of having the self-reported outcome on the self-reported environmental exposure same-source bias is likely to happen. Several tools are available to minimize same source bias in studies that rely on self-reported information by study participants, including the aggregation of environment perceptions by study participants living in the same neighborhood using techniques such as Conditional Empirical Bayes (CEB).6 The best way to get rid of same-source bias is avoiding self-reports of environmental measures by study participants and the use of other environmental measurements (different reporters, systematic social observation, georeferenced databases and GIS)4 but we acknowledge that measuring the environment is difficult and may not be feasible in all studies. Using self-reported measures may be desirable if the research question relates to how people perceive their environment and the effect of this perception on physical activity, although the interpretation of these results may be more difficult. Nonetheless same-source bias should be avoided in these situations too.