AN APPLICATION OF DIAGNOSTIC TESTS FOR THE INDEPENDENCE FROM IRRELEVANT ALTERNATIVES PROPERTY OF THE MULTINOMIAL LOGIT MODEL

Abstract

Statistical tests are proposed to diagnose the validity of the independence from (of) irrelevant alternatives property of the multinomial logit model. Application of the tests is illustrated by the use of actual travel data representing urban modal choice in the San Francisco area. The property as it applies to travel demand forecasting is discussed, and the common misconception that the property holds for market shares in heterogenous populations is shown by examples to be incorrect. The relation of the property to the basic assumptions of the model is described, and it is shown that the validity of the property in disaggregate modeling is an empirical issue that depends on the model specification and data in a particular application. A series of diagnostic tests for the property are developed and applied to actual travel data. /Author/

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 220 citations exist upstream.