Improving meta-analyses on hypothetical bias by using separate models for private and public goods

ABSTRACT

Previous meta-analyses examined hypothetical bias studies with the aim of identifying characteristics of stated preference (SP) design that potentially reduce hypothetical bias. However, previous meta-analyses combined observations for both public and private goods in their models. In this paper, we argue that eliciting values for public and private goods should be considered in separate models. Individuals’ behaviours, the choice of mechanism and the efficiency of mitigation techniques are specific to each type of good. Separating the models should allow more precise model specification and better identification of design effects. With two meta-regressions hierarchical mixed-effect models we provide the first meta-analysis for public and private goods separately. This approach provides specific information regarding SP design and better insight into the efficiency of mitigation techniques to reduce hypothetical bias.

KEYWORDS:

• Stated preference
• hypothetical bias
• meta-analysis
• willingness-to-pay
• private goods
• public goods

JEL Classifications:

• C10
• H41
• Q26
• Q28

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 An individual can always choose not to buy a private good, if or once offered.

2 The rational of perceived consequence for price or provision of private goods applies to products with innovative or novel attributes. Some experiments using common items (i.e. chocolate bar) lack any reasonable consequence other than academic curiosity (Carson and Groves 2007).

3 A more detailed explanation of the model specifications can be found at Atozou et al. 2020.