Network Autocorrelation Models
- 1 November 1984
- journal article
- research article
- Published by SAGE Publications in Sociological Methods & Research
- Vol. 13 (2) , 155-200
- https://doi.org/10.1177/0049124184013002001
Abstract
For network effects models of network autocorrelation we use Monte Carlo methods to study the relative properties of three estimation methods. The methods are the iterative maximum likelihood estimation (Ord, 1975; Doreian, 1981), ordinary least squares, and a regression-based “quick and dirty” substitution for iterative MLE. Of the three, OLS is clearly the inferior estimation method and MLE the superior method. We recommend the use of the maximum likelihood method when network autocorrelation models are to be estimated.Keywords
This publication has 8 references indexed in Scilit:
- A Spatial Autocorrelation Model of the Effects of Population Density on FertilityAmerican Sociological Review, 1983
- Network autocorrelation: A simulation study of a foundational problem in regression and survey researchSocial Networks, 1982
- Testing a structural model of perception: Conformity and deviance with respect to Journal norms in elite sociological methodologyQuality & Quantity, 1982
- Maximum Likelihood Methods for Linear ModelsSociological Methods & Research, 1982
- Sexual Division of Labor in African Agriculture: A Network Autocorrelation AnalysisAmerican Anthropologist, 1981
- Linear Models with Spatially Distributed DataSociological Methods & Research, 1980
- Estimation Methods for Models of Spatial InteractionJournal of the American Statistical Association, 1975