inscription
PORTAIL D'INFORMATION GÉOGRAPHIQUE

Estimation of autoregressive models with two types of weak spatial dependence by means of the W-based and the latent variables approach : evidence from Monte Carlo simulations

Auteur(s) et Affiliation(s)

LIU, A.
Dept. of Spatial Sciences, Univ., Groningen, Pays-Bas
FOLMER, H.
Faculty of Spatial Sciences, Univ., Groningen, Pays-Bas
OUD, J.H.L.
Behavioural Science Institute, Univ., Nijmegen, Pays-Bas


Description :
This article examines the estimation of autoregressive models with two types of weak spatial dependence by means of the W-based and the latent variables approach through evidence from Monte Carlo simulations. It starts by comparing the two approaches by means of simulations in terms of bias and root mean squared error (RMSE) for different values of the spatial lag parameters, specifications of the weights matrices, and sample sizes. The simulation results show that compared with the single W-based models, SEM frequently has smaller bias and RMSE. Furthermore, SEM even outperforms the correctly specified W-based models (based on the same two weights matrices used for data generation) in many cases.


Type de document :
Article de périodique

Source :
Environment and planning A, issn : 0308-518X, 2014, vol. 46, n°. 1, p. 186-202, nombre de pages : 17, Références bibliographiques : 2 p.

Date :
2014

Editeur :
Pays édition : Royaume-Uni, London, Pion

Langue :
Anglais
Droits :
Tous droits réservés © Prodig - Bibliographie Géographique Internationale (BGI)