Testing household preferences for the importance of the frequency and severity of water quality impairment

Abstract Water quality indices are employed by governments largely as a means of communicating the multifaceted nature of water quality and aquatic ecosystem health to the general public. Given the complexity of responsibility for oversight of freshwater quality in Canada, the Canadian Council of Ministers of the Environment (CCME) has developed an index based on the severity, frequency, and scope of water impairment. An important feature, and potential shortcoming, of this approach is that the three attributes of water quality are weighted equally. If households, however, weight these attributes differently, then the index’s ability to convey information to the public may be weakened. This issue is examined by eliciting household preferences for a hypothetical water quality protection program that reduces the severity and frequency of impairment using a discrete choice experiment (issues of scope are not included in the analysis). Latent class and mixed logit models are estimated. The latent class models, which outperform the mixed logit, indicate the presence of two preference classes that hold dramatically different preferences for benefits of the protection program. While one group of respondents is unresponsive to the severity and frequency of impairment, there is evidence that the other group may assign different weights to the attributes. These findings suggest the CCME index could convey different information to the two groups.