11
responded that they only eat goat meat on special occasions. We examine seasonal and
occasion related consumption (Christmas, 4th of July, Family re-union, Marriage,
Ramadan, Cinco De Mayo) against the set of explanatory variables used in previous
models. The response variables with multiple nominal levels in this model lead to the
adoption of multinomial logit model with the generalized logit function. In contrast to the
previous models, multinomial logit models have a fixed reference category, and all other
categories are coded to contrast with the reference category.
Model Selection and Checking
The survey data in this study encompass an extensive domain. The initial set of variables
for each model is selected from a large pool of variables on the basis of the demand
theory and previous publications. As such, the initial set of variables comes to including
the consumption of other meats, real income, age, education, household structure, gender,
and geographic variables, etc. The initial variables are tested against a set of statistical
criteria and those pass tests will be retained in the model.
Model diagnostic and checking have been the topic of discussion for decades and
ended up with a large body of literature (Hocking; Judge et al.; Wallace). However, more
questions raised then solved. Model selection and checking is still a complicated issue
and no simple steps could lead to the finding of the true model. We did not use the
stepwise regression procedure for its flaws extensively discussed. Rather, we tested
variables and models based on a set of statistical criteria (R2, Cp, P-value, and the Wald
and the likelihood ration statistics). Cp was close to the potential number of explanatory
variables to avoid the over- or under-specification of models. For each variable, we use
the Wald statistic to test its significance. For variable set, we use the likelihood-ratio
More intriguing information
1. The name is absent2. The Composition of Government Spending and the Real Exchange Rate
3. The name is absent
4. The name is absent
5. The name is absent
6. Growth and Technological Leadership in US Industries: A Spatial Econometric Analysis at the State Level, 1963-1997
7. STIMULATING COOPERATION AMONG FARMERS IN A POST-SOCIALIST ECONOMY: LESSONS FROM A PUBLIC-PRIVATE MARKETING PARTNERSHIP IN POLAND
8. Education as a Moral Concept
9. Sex differences in the structure and stability of children’s playground social networks and their overlap with friendship relations
10. Telecommuting and environmental policy - lessons from the Ecommute program