The impact of weight matrices on parameter estimation and inference: A case study of binary response using land-use data

Abstract

This paper develops two new models and evaluates the impact of using different weight matrices on parameter estimates and inference in three distinct spatial specifications for discrete response. These specifications rely on a conventional, sparse, inverse-distance weight matrix for a spatial autoregressive probit (SARP) model, a spatial autoregressive approach where the weight matrix includes an endogenous distance-decay parameter (SARPα), and a matrix exponential spatial specification for probit (MESSP). These are applied in a binary choice setting using both simulated data and parcel-level land-use data. Parameters of all models are estimated using Bayesian methods.In simulated tests, adding a distance-decay parameter term to the spatial weight matrix improved the quality of estimation and inference, as reflected by a lower deviance information criteriaon (DIC) value, but the added sampling loop required to estimate the distance-decay parameter substantially increased computing times. In contrast, the MESSP model's obvious advantage is its fast computing time, thanks to elimination of a log-determinant calculation for the weight matrix. In the model tests using actual land-use data, the MESSP approach emerged as the clear winner, in terms of fit and computing times. Results from all three models offer consistent interpretation of parameter estimates, with locations farther away from the regional central business district (CBD) and closer to roadways being more prone to (mostly residential) development (as expected). Again, the MESSP model offered the greatest computing-time savings benefits, but all three specifications yielded similar marginal effects estimates, showing how a focus on the spatial interactions and net (direct plus indirect) effects across observational units is more important than a focus on slope-parameter estimates when properly analyzing spatial data.a yiyi

Volunteered geographic information quality assessment using trust and reputation modelling in land administration systems in developing countries

“…Six different land use types were identified in the study area to classify each land parcel. A six point rating scale related to the classes was designed to help experts provide subjective ratings based on a land use weighted matrix table (Wang et al 2013): rating values of four and above represented values denoting positive accuracy measures of a land use classification, while those of three and below denoted negative accuracy measures.…”

‘No country for old men’? The multiplier effects of pensions in Portuguese municipalities

“…Direct and indirect effects are different from the estimates of the response parameters and from the estimates of neighbors' covariates. As LeSage and Pace (2009) and Wang et al (2013) discussed, these differences can be explained by the feedback effects (Indirect effects), which return to the municipality after passing through the surrounding municipalities. In contrast, direct effects account for the (spatially immediate) effects that the explicative variables produce in the place in which they were observed.…”

“…Authors such as LeSage and Pace (2009) and Wang et al . (2013) have properly detailed the choice between the models’ specifications and the potential specifications of the spatial weight matrix W . The methodological strategy focuses on a Bayesian comparison approach.…”

“…Regarding the analysis of the spatial weight matrices, we follow Delgado and Robinson (2015) and we refer to a diversity of row-normalized geographical-distance matrices (inverse distance matrices with cut-offs and n-order contiguity matrices, and k-nearest neighboring matrices). Authors such as LeSage and Pace (2009) and Wang et al (2013) have properly detailed the choice between the models' specifications and the potential specifications of the spatial weight matrix W. The methodological strategy focuses on a Bayesian comparison approach. This approach enables the construction of a Table of Posterior Model Probabilities, which signals the best alternative among the models.…”

Towards an East German Wage Curve: Nuts Boundaries, Labour Market Regions and Unemployment Spillovers

“…This seems in contrast to LeSage and Pace (2014) who find little theoretical support for marked differences in inference. However, LeSage and Pace (2014) do not rule out the possibility that different matrix choices have an impact on the results (see also Wang, Kockelman and Wang 2013).…”