Introduction

This chapter will discuss some issues in statistical inference in the analysis of choice models. We are concerned with two kinds of computations, hypothesis tests and variance estimation. To illustrate the analyses, we will work through an example based on a revealed preference (RP) data set. In this chapter, we present syntax and output generated using Nlogit to demonstrate the concepts covered. The syntax and output is, for the more familiar reader, largely self-explanatory; however, for the less familiar reader, we refer you to Chapter 11, which you may wish to read before going further. The multinomial logit model for the study is shown in the following Nlogit set up which gives the utility functions for four travel modes: bus, train, busway and car, respectively:

1. ;Model:

2. u(bs) = bs + actpt*act + invcpt*invc + invtpt*invt2 + egtpt*egt + trpt*trnf /

3. u(tn) = tn + actpt*act + invcpt*invc + invtpt*invt2 + egtpt*egt + trpt*trnf /

4. u(bw) = bw + actpt*act + invcpt*invc + invtpt*invt2 + egtpt*egt + trpt*trnf /

5. u(cr) = TC*TC + PC*PC + invtcar*invt + egtcar*egt

The attributes are act = access time, invc = in vehicle cost, invt2 = in vehicle time, egt = egress time, trnf = transfer wait time, tc = toll cost, pc = parking cost, and invt = in vehicle time for car. Where a particular example uses a method given in more detail in later chapters, we will provide a cross-reference.