Transportation and automobile use is a major concern today in the United-

States. The use of automobile has impacts on congestion, urban dynamics, environ-

ment and on the economy in general. Good indicators of transportation demand

are the number of vehicles owned by a household and the total number of miles

traveled.

This thesis aims at building a model that can predict the total vehicle miles

traveled and number of cars owned by households, simultaneously. The discrete-

continuous model that we present correlates the error terms of a utility-based probit

with the error term of an ordinary regression. The objective is to capture the

relationship between preferred ownership alternatives and miles traveled.

We successfully show that households with high utility for owning a lot of

cars also drive more and that households with high utility for owning few cars drive

less. The correlation is between utilities and miles traveled. It also correlates the

two transportation demand indicators without assuming that one precedes the other

and, thus, does not suffer from circular variable inclusions.

The thesis ends by incorporating sampling weights into the model before pa-

rameters are estimated. We find slight changes in parameters' values calculated

with weights. The difference however, is more quantitative than qualitative since

the general analysis we make with the weighted coefficients remains the same, only

the magnitude of the effects change