Richard Dunn

Stochastic Panel-Data Models of Urban Shopping Behaviour: 1. Purchasing at Individual Stores in a Single City

In this paper, the first application of the NBD (negative binomial distribution) consumer purchasing model is presented in the context of purchasing at individual stores in a single city. Empirical results derived from a recently conducted spatially disaggregate consumer-panel survey in Cardiff, United Kingdom, are reported. The fit of the model, in general, is extremely good, although there are certain small but consistent discrepancies, the explanation of which has both statistical and geographical components.

Stochastic Panel-Data Models of Urban Shopping Behaviour: 2. Multistore Purchasing Patterns and the Dirichlet Model

In this paper, the first application of the Dirichlet model of heterogeneous buyer behaviour, in the context of multistore purchasing in a single city, is presented. Empirical results derived from a recently conducted spatially disaggregate consumer-panel survey in Cardiff, the United Kingdom, are reported. The fit of the model is, in general, extremely good, and its potential for studies of urban consumer behaviour is stressed.

A nonparametric approach to the incorporation of heterogeneity into repeated polytomous choice models of urban shopping behaviour

Abstract Longitudinal discrete choice models which use a mixing distribution to control for unobserved heterogeneity, and which are estimated by a nonparametric maximum likelihood mass point procedure, have so far been restricted to the analysis of dichotomous choices. This paper extends mass point methods to the case of polytomous choice problems and in so doing makes an important contribution to the analysis of longitudinal data and dynamic discrete choice models. The paper concludes with an empirical assessment of the performance of mass point methods in the analysis of urban shopping behaviour, using short-term panel data from the Cardiff Consumer Panel.

The Geography of Council House Sales in England — 1979-85

Analyses of the social, economic and spatial consequences of council house sales have tended to rely on local case study evidence or limited official statistics at the national level. This paper seeks to provide more systematic evidence for England as a whole. Sales which were completed between 1979 and 1985 are mapped by local authority to indicate the uneven geography of the privatisation of council housing in England. Statistical analyses of key variables are then used to draw out some of the factors at work contributing to high and low levels of sales. One of the principle conclusions is that sale of council houses is leading to a greater unevenness in the national pattern of housing tenure.

Stochastic Panel-Data Models of Urban Shopping Behaviour: 4. Incorporating Independent Variables into the NBD and Dirichlet Models

In this paper the authors describe the way in which the NBD and Dirichlet models of consumer behaviour may be extended to include independent or exogenous variables. Such methods allow the predictions obtained from these models to be disaggregated, that is, to be made conditional on the values of the exogenous variables. Empirical examples are presented of repeat buying of a branded good and at an individual store (for the NBD model), and of multistore purchasing (for the Dirichlet model).

The Cardiff Consumer Panel: methodological aspects of the conduct of a long-term panel survey

Panel surveys can provide high quality data of unmatchable detail. They are of central importance for the description and analysis of spatial behaviour which involves recurrent choice and movement, and for the assessment of change in behaviour over time. In this paper a discussion of some methodological aspects of panel surveys is provided via a description and appraisal of the design, conduct and characteristics of the Cardiff Consumer Panel Survey of 1982. The importance of the data base generated by the Cardiff survey, as a resource for geographical research, is assessed.

Stochastic Panel-Data Models of Urban Shopping Behaviour: 3. The Interaction of Store Choice and Brand Choice

In this paper, the first consideration of the interaction of the processes of store choice and brand choice at the level of individual stores is presented. Empirical results derived from a recently conducted spatially disaggregate consumer-panel survey in Cardiff, the United Kingdom, are reported. These results suggest that store loyalty and brand loyalty are both extremely low. On this basis it is argued that store and brand choice might profitably be regarded as a nested choice process, with store choice preceding brand choice. The application of the Dirichlet model to patterns of multibrand purchasing within individual stores is then discussed.

Graphical diagnostics for logistic oil exploration models

This paper demonstrates how some recently developed graphical diagnostic techniques can be used to explore and improve the specification of logistic oil exploration models. Techniques are applied to information on 124 hydrocarbon exploration wells drilled between 1948 and 1963 into the ‘B’ Division of the Mississippian (Osage Series) in a 13 by 13 square mile area of Stafford County, south-central Kansas.

A Dynamic Approach to Two-Variable Color Mapping

Abstract Procedures for high-interaction two-variable color mapping are described, whereby the user interacts with a graphic display to produce a color statistical map in a few seconds. The approach emphasizes the methodological benefits derived from the ability to examine the nature of the linkage between the statistical and spatial distributions of bivariate data. A series of examples illustrate this method.

Framed Rectangle Charts or Statistical Maps with Shading: An Experiment in Graphical Perception

Abstract This article presents and discusses the results of an experiment in human graphical perception designed to assess the relative accuracy with which quantitative information is extracted from a framed rectangle chart compared to an unclassed choropleth map. The basis for and results of this experiment are discussed in relation to recent developments in theories of graphical perception and statistical mapping.