4 CONSUMER WELFARE CHANGES AND THEIR VALUATION
4.4 V ALIDITY AND RELIABILITY OF THE CHOSEN METHOD
4.4.1 Validity
Validity can be examined from various perspectives, the most important of which in willingness to pay studies are content and construct validity (Mitchell and Carson 1989). In the context of a willingness to pay study, the theoretical construct to be obtained is the maximal consumer willingness to pay, or the maximum amount of money the consumer would actually pay for the existence of the good to be valuated on the markets (Mitchell and Carson 1989).
Content validity consists of measuring conceptually relevant issues. In willingness to pay studies, particular emphasis falls on the careful construc-tion of the survey form, especially its actual hypothetical scenario and will-ingness to pay question.
The validity of CV studies has often been challenged, because the differ-ent willingness to pay questions (open question versus dichotomic, dichoto-mic question versus payment card) have elicited varying willingness to pay estimates, although in theory they should all describe the same theoretical construct: Hicksian consumer surplus. All question formats have their weak-nesses and strengths, of which the most studied are the effect of the selec-tion of the bids or end points, the so-called anchoring effect, incentive com-patibility, and the restrictions imposed by the statistical characteristics of each question format.
Construct validity means the ability of a given measure to generate the results expected from the study, in which case the measure as a whole
works as it is supposed to. In willingness to pay studies, construct validity can be estimated where parallel markets exist, or for example where ex-perimental markets can be created for the goods (Mitchell and Carson 1989).
Hanemann (1994) outlines three ways to improve the validity of a study: to repeat the study, to compare the results of the study with other results, and to compare the results of the study to real behaviour whenever possible.
To improve the validity of this study, researchers and experts representing various disciplines and research institutes were asked to review the survey form and provide comments and feedback. The survey form was also pre-tested prior to the actual survey.
4.4.2 Reliability
In a CV study, the attributes of the good to be valuated are described to the respondents using a scenario defined to the highest possible degree of clarity and concreteness, and a sufficient level of credibility. By creating the attrib-utes and credible markets for the good, a situation is constructed where will-ingness to pay depends on or is, in other words, contingent on the scenario.
This is also how the name for the method was derived.
Hanemann (1994) states that reliability can be affected in the following stages: during the formation of the sample and formulation of the survey, in the valuation scenario, the questioning structure, and during the analysis of the research data. The most crucial element in a CV study and its reliability is the formulation of the valuation question. To improve reliability, the valua-tion scenario must be as realistic and concrete as possible. In this study, the hypothetical bias cannot be considered a major problem, because the good to be valuated (increased quality information) becomes concrete in the con-text of buying beef, which is a familiar decision-making situation for most respondents.
The actual willingness to pay question should be formulated so that it restricts valuations to a situation that behaviourally resembles voting, where the responses are of the Yes/No type, or so-called closed questions (Arrow et al. 1993, Hanemann 1994). The dichotomic questioning format was chosen because it can be implemented as a computer-aided survey. For example, the NOOA Panel (Arrow et al. 1993) specifically recommended the dichoto-mic questioning format.
The willingness to pay question is constructed around a specific good, to ensure its maximal relevance and credibility to the respondents. The actual scenario should seem as credible as possible, to encourage the respondents
to express their willingness to pay, even in the absence of actual transac-tions. The method of payment, e.g., taxes or price increases, must have a credible linkage to the supply of the good in question.
Since the mid-1980s, the most prevalent question format has been the closed question. The deciding argument has been the fact that the closed question format is easier and more familiar to the respondents. In the course of normal commerce, we have to think whether we are willing to buy a given product at a given price. According to Hanemann (1994), people are more willing to tell what a given good costs than what it would be worth to them.
The respondents have strategic reasons for quoting prices lower than the full value of the good. The closed question format does not offer strategic rea-sons for lower price quotes.
CV studies have been criticised for the so-called hypothetical bias, or the fact that in reality the respondents will not pay the quoted prices. The hypothetical bias can be reduced by making the scenario as realistic as pos-sible. In this study, willingness to pay is linked to the buying of a private good, whereby the situation is familiar to the respondents from previous ex-perience.
In this study, the construction of the scenario presented problems with the issue of scope, or changes in the amount of the good to be valuated. In the scenario, the changes could not be explicitly defined, because detailed description and cataloguing of all available quality information would have complicated the actual willingness to pay scenario. Thus, the willingness to pay scenario of this study pointed out to the respondents that it is in fact possible for consumers to receive highly detailed quality information.
In terms of critiquing the questions framed in this study, one might speculate whether the outcome would have been different if the respondents were queried about each quality-related piece of information separately. It was the purpose of this study to determine, through focused questions, what specific quality information the respondents consider relevant to produce, but the willingness to pay for each information requirement was not surveyed. Detailed findings on the requirements for new information in addition to the current package labelling are presented in Chapter 5.6.