Accounting for protest zeros in contingent valuation studies : a review of literature

The Contingent valuation (CV) approach is commonly used in environmental and agricultural economics and is becoming increasingly popular in the valuation of health and health care. Whatever the context, CV surveys risk eliciting "protest" responses where respondents state a zero valuation for a good even though their true valuation is greater than zero. Protests may result in study sample, information and hypothetical biases in analysis if censored, or cause inaccurate reporting of the true economic value of the good if uncensored. We review the prevalence of zero valuations, the classification of "protests" and the sociodemographic variables associated with protesting in CV studies in across a range of contexts including health. The results of the search found there is consensus that zero valuations due to concerns about taxation and/or trust in government should be classified as "protests". Those motivated by inability to pay should be classified as "true" zeros and retained in the data. A "don’t know" option and follow-up questions should be used to detect protesters. Our results show it is unlikely those protesting are representative of the population, hence, removing protests will lead to unrepresentative samples. Therefore, econometric techniques allowing valuation estimates to be "debiased" should be utilised. Whilst much of the evidence on the issues of zero and protest valuations comes from the fields of environmental and agriculture economics, this is the first paper to review the classification, modelling and the sociodemographic variables of zero and protest bids within health economics. We have identified a number of lessons of best practice for the future designs of CV studies conducted in health.