mix of characteristics becomes available to consumers. Many CV stud-
ies focus on how much consumers would be willing to pay above the
going market price for certain attributes. Some argue that empirical stud-
ies should distinguish between acceptability and willingness-to-pay. Or-
ganic produce or fresh produce grown under Integrated Pest Management
(IPM) regime are examples of goods still little known to consumers,
while various opinions circulate about crops with the attribute mix al-
tered through genetic engineering. An acceptability study may be a pre-
requisite for willingness-to-pay investigation.
Consumer purchase or consumption decisions can be described by a
binary variable. The variable assumes the value of one if the decision
to purchase was made and equals zero otherwise. The categorical nature
of the dependent variable presented a problem in the application of sta-
tistical methods and the interpretation of estimation results. The theo-
retical framework and the appropriate estimation technique were
reconciled by McFadden (1981). McFadden’s approach transforms the
algebraic expression of utility into a statistical equation that can be es-
timated by probit or logit technique. Probit technique is considered su-
perior to the logit approach because its application requires rigorous
distribution assumptions. The logit approach relaxes the normality as-
sumption and, by rule of thumb, empirical logit models provide similar
results if estimated using large data sets.
Factors that influence the decision to purchase may be different from
those responsible for a premium a consumer is willing to add to the com-
parable price. Consumer surveys probing for willingness-to-pay ask re-
spondents to choose from several price premium alternatives. The
selection of a premium begins at the value of zero and is constrained by
investigators at a level perceived as the reasonably high. The selection
of each alternative is coded as categorical variables assuming values of
zero to record the choice of the first alternative, two to record the choice
of the second alternative, and so on. This coding scheme requires a dif-
ferent estimation technique if a categorical variable becomes the de-
pendent variable. Ordered probit or ordered logit techniques are applied
and estimated coefficients interpreted as indicators of a probability of
the respondent’s choice falling into a particular category describing the
range of premiums.
In the case of all four techniques (i.e., probit, logit, ordered probit,
and ordered logit), the practical interest lies in the ability to identify the
effect of each explanatory variable on the dependent variable. The mar-
ginal effects of quality attributes measure by how much the probability
Market Valuation of Quality 237