of variables selected on the basis of the demand theory and fit logistic model with the
SAS logistic procedure.
Demand Model
(1) Model of current demand. This model examines households’ existing consumption
behavior and focuses on the explanation of consumer behavior with a set of demographic
and socioeconomic variables. The purpose of this model is to identify the current
consumers and to define the current niche markets. The dependent variable is valued with
the response to a survey question “have you or any member of your immediate family
member ever eaten goat meat?” The initial set of independent variables includes
consumption behavior of substitution meat such as beef, pork, and chicken; demographic
factors, such as household size, age, and race; and socioeconomic variables such as
education and household real income. A logistic regression with the general logit
function is fitted. To reduce the lost of information in original data, most original levels
of variables in the survey were kept. However, levels of some variables were combined to
avoid bias if few observations observed or to increase efficiency if similar impacts from
multiple levels. Models were first tested with the global hypothesis on whether the model
as a whole help to explain the variability of response variable, then each variable was
further examined on the basis of the type 3 analysis, and individual level was tested with
the Wald statistics. To avoid the lost of the degree of freedom, variables were eliminated
from the model if failed to pass statistical tests, even some variables may be the
influencing factors from the theoretical perspective. This treatment differs from other
studies in which variables were still retained even not significant statistically.