Logit Discrete Choice Model Research Articles

THE APPLICATION OF DISCRETE CHOICE MODELS TO WILDLIFE RESOURCE SELECTION STUDIES

This paper introduces the theory and application of discrete choice models to resource selection studies. Discrete choice models calculate the probability of an individual selecting a resource as a function of the attributes of that resource and all other available resources. The data for these attributes may be continuous or categorical. When availability is the same for all individuals and only two resources are available, the multinomial logit discrete choice model reduces to the logistic model. Discrete choice models and advances in GIS technology give the researcher flexibility in defining resource availability separately for each individual over time and space. The output of the discrete choice approach also provides managers with a tool to explore the effects of potential management actions and provides researchers with new hypotheses deserving of further investigation. To illustrate the application of discrete choice models to resource selection studies, we present a case study of summer diurnal bed site selection by elk (Cervus elaphus) in Custer State Park, South Dakota, United States. The results demonstrate the importance of factors relating to thermal regulation, hiding cover, and potentially forage, in elk bed site selection in this region.

Boat Location Choice: The Role of Boating Quality and Excise Taxes

A multinomial logit discrete choice model is estimated using data collected from a survey of documented boat owners to examine the factors that determine the boater's choice of state of principal use. Travel time and perceived boating quality significantly influence boater choice, but excise taxes that must be paid depending on the state in which the boat is located only negatively influence location choice at extremely high tax levels and only for high income boaters. It is unlikely that a state such as Maryland, which has a relatively high excise tax rate on boaters, could entice more boaters to keep their boats there, simply by lowering the tax rate from 5% to 4%. A more effective means of attracting boaters would be to maintain or improve coastal quality and make potential boaters aware of the improvement.

Painting by Numbers—Mathematical Models of Urban Systems

The objective of this paper is to contribute to the understanding of the phenomenon of mathematical modelling in urban studies and to stimulate a deeper debate about the use of models in planning. This is done by means of an identification of key assumptions made in the process of the interpretation of reality as a set of algebraic relations; assumptions which, it is argued, underlie virtually the whole of mathematical modelling but which appear never to be openly discussed. An example of the use of these assumptions is then given in the theoretical development of the logit model of discrete choice, and the paper is concluded with preliminary recommendations about the future development and use of the mathematical model.

Testing for Independence of Irrelevant Alternatives

In a recent article, Zhang and Hoffman discuss the use of discrete choice logit models in sociological research. In the present article, the authors estimate a multinomial logit model of U.K. Magistrates Courts sentencing using a data set collected by the National Association for the Care and Resettlement of Offenders (NACRO) and test the independence of irrelevant alternatives (IIA) property using six tests. Conducting the tests with the appropriate large sample critical values, the authors find that the acceptance or rejection of IIA depends both on which test and which variant of a given test is used. The authors then use simulation techniques to assess the size and power performance of the tests. The empirical example is revisited with the inferences performed using empirical critical values obtained by simulation, and the resultant inferences are compared. The results show that empirical workers should exercise caution when testing for IIA.

The additive structure of utility in discrete choice models

The logit model of discrete choice is generalized with respect to not only the shape of the error distribution, but also the additive cost structure. If utilities of alternatives are distributed according to different laws and many observations are accommodated, then a utility maximizer is shown to be governed by the logit model with an additive cost structure and an error distribution defined by the laws. The representation may break down in the case where the alternatives attracting positive choice probabilities have laws featuring either slow or fast decay, but a new result covers this situation.

Pedestrian Access to Transit: Model of Walk Trips and Their Design and Urban Form Determinants Around Bay Area Rapid Transit Stations

Walking links practically every trip taken, yet it has not been maximized as a primary mode of transportation, and the characteristics of walk trips are little understood. The Bay Area Rapid Transit (BART) system has been used for primary analysis of characteristics of pedestrian access to transit. A travel survey of individual riders was used to develop a disaggregate discrete choice logit model of walk access to BART stations. Characteristics with the highest probability of choosing walking over another mode are distance, availability of a car, and gender. Linear regression analysis of aggregate station area characteristics was developed; density, the number of parking spaces at the station, and income and education of the residents were identified as key factors in the decision to walk. When the logit and regression models were combined into a single model, there was a clear shift toward individual characteristics over aggregate station area characteristics that determines access-mode choice. Among physical characteristics, station areas with high levels of retail activity outside downtown areas contain the highest proportion of walk trips, followed by station areas with little or no parking.

Estimating the Probability of Return Visits Using a Survey of Tourist Expenditure in the Balearic Islands

Using data supplied by the annual Survey on Tourist Expenditure in the Balearics, the returning intention (‘yes’ or ‘no’) has been modelled using a LOGIT model of discrete choice, which enables research of the factors (most of them qualitative) influencing this decision. The special importance of return visits to the Balearics is highlighted. The results presented correspond to the 1994 survey, although earlier studies, for 1989 and 1992, give us a temporal comparison of results. In addition, a comparison of the results for the two predominant nationalities of visitors to the Islands, British and German, is discussed.

The nested logit model and representative consumer theory

Two versions of the nested logit discrete choice model are linked to representative consumer theory.

Discrete-Choice Logit Models

Multinomial logit discrete-choice models are now being used more frequently in sociological research, but often without appropriate attention to the central role played by the independence of irrelevant alternatives (IIA) property. In this article, three tests of the IIA property are discussed and then how to apply them is illustrated, using a data set previously analyzed by Hoffman and Duncan. The IIA tests decisively reject the IIA property in the single/welfare/remarriage choice model.

The Credit-Constrained Consumer: An Empirical Study of Demand and Supply in the Loan Market

This article presents an empirical analysis of credit-constrained households in which models of both demand and supply sides of the market for loans are simultaneously estimated. We develop a discrete-choice logit model of the consumers' decision on whether or not to apply for credit. Combining this with a reduced-form logistic model for the banks' credit-granting decision, we then estimate the model with cross-sectional data, as a nonlinear, ordered, sequential logit. Our findings shed light on the criteria used by banks in assessing credit applications and reveal the strong dependence of preferences on demographic characteristics. The results also suggest that agents face significant fixed costs in applying for loans. The fact that we jointly estimate models of both supply and demand sides of the credit market enables us to examine whether particular groups in the population are more frequently credit rationed because they tend to demand more loans or because they are viewed as bad credit risks by lenders.

THE LOGIT AS A MODEL OF PRODUCT DIFFERENTIATION

The logit discrete choice model is argued to be flexible, tractable, and intuitively sound as a demand model of product differentiation under oligopoly. The free entry equilibrium product range is greater than or less than the social optimum, depending on cost and demand conditions and the degree of heterogeneity of consumer tastes. If the market provides too many products, then the difference between equilibrium and first-best optimal product diversity is negligible. However, when the market solution yields too few products, the market may seriously underprovide diversity. Copyright 1992 by Royal Economic Society.

A Heterogeneous Conditional Logit Model of Choice

Aggregate forecasts using McFadden's conditional logit model of discrete choice harbor the unrealistic implicit assumption of a random-utility distribution that is homogeneous across both alternatives and individuals. This article relaxes that assumption. A choice model is developed that describes the random-utility component as a probability-mixture model. Some numerical results illustrate that the derived model is not constrained by the independence-of-irrelevant-alternatives property. An experimental test of visual perceptions demonstrates the potential superiority of the model.

A logit analysis of telecommunication network bypass decisions

This paper is concerned with the growing trend for business and institutional users of telecommunication services to bypass public telephone networks by installing private facilities made available by recent technological advances. The factors leading to bypass decisions are determined, using discrete choice logit models estimated with data gathered in a specifically designed large survey of manufacturers and financial institutions. The results clearly show that the larger the firm in terms of numbers of employees and locations and in terms of telecommunication activity, the higher is the likelihood of that firm to engage in bypass. The forecasting applicability of the models is also outlined.