For qualitative discrete variables, such as the explanatory variables used in this study,
dPi/dXij does not exist. Probability changes are then determined by:
(∂Pl/ ∂Xij) = Pi(Yi :Xij= 1) - Pi(Yi :Xij= 0)
The following model was developed to predict the likelihood of making frequent use of
nutritional labeling (i.e. those who usually or always consult nutritional labeling on the
food products they purchase). The model was tested under the specification:
Prob = β0 + β1 Male + β2Age1 + β3Age2 + β4Age3 + β5 Suburban
+ β6 Rural + β7 Household_Size + β82Children + β9 Prime + β10 Organic
+ β11 Media + β12 Education + β13 HiJncome
Where:
|
Prob |
= 1 if the participant usually or always checked nutritional |
|
Male |
= 1 if the individual is male and 0 otherwise |
|
Age1 |
= 1 if the individual is older than 65 years of age and 0 |
|
Age2 |
= 1 if the individual is 51 to 65 years of age and 0 otherwise |
|
Age3 |
= 1 if the individual is 36 to 50 years of age and 0 otherwise |
|
Suburban |
= 1 if the individual resides in a suburban neighborhood and 0 |
|
Rural |
= 1 if the individual resides in a rural neighborhood and 0 |
Household_Size = 1 if the number of individuals living in the household were 4
or more and 0 otherwise
2Children = 1 if 2 or more individuals under the age of 17 resided in the
household and 0 otherwise