Sensitivity of MRQAP Tests to Collinearity and Autocorrelation Conditions

By David Dekker, David Krackhardt, Tom A. B. Snijders. Psychometrika, 2007

805 citations recorded by the source corpus.

Abstract Multiple regression quadratic assignment procedures (MRQAP) tests are permutation tests for multiple linear regression model coefficients for data organized in square matrices of relatedness among n objects. Such a data structure is typical in social network studies, where variables indicate some type of relation between a given set of actors. We present a new permutation method (called “double semi-partialing”, or DSP) that complements the family of extant approaches to MRQAP tests. We assess the statistical bias (type I error rate) and statistical power of the set of five methods, including DSP, across a variety of conditions of network autocorrelation, of spuriousness (size of confounder effect), and of skewness in the data. These conditions are explored across three assumed data distributions: normal, gamma, and negative binomial. We find that the Freedman–Lane method and the DSP method are the most robust against a wide array of these conditions. We also find that all five methods perform better if the test statistic is pivotal. Finally, we find limitations of usefulness for MRQAP tests: All tests degrade under simultaneous conditions of extreme skewness and high spuriousness for gamma and negative binomial distributions.

Subfields: Sociology and Political Science, Statistical and Nonlinear Physics

View Paper: https://www.semanticscholar.org/paper/adb415e4187062096c869cb83bc91b3a149210cc

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Sociology and Political Science

In Statistical and Nonlinear Physics

Citations in the corpus

Geography, Networks, and Knowledge Flow (2007, methodology): “However, the double Dekker semi-partialing method minimizes the effect of, and is robust to the effect of, collinearity (Dekker et al. 2005).”
The dynamics of animal social networks: analytical, conceptual, and theoretical advances (2014, positive): “Multiple Regression Quadratic Assignment Procedure (MRQAP) is another method that can be used to investigate the synergistic effects of multiple factors on network structure (Dekker et al. 2003; Dekker et al. 2007; Croft et al. 2011).”
Statistical Models for Social Networks (2011, methodology): “Statistical Models for Social Networks 7 permutation of regression residuals (Multiple Regression QAP, or MRQAP) by Krackhardt (1988) and Dekker et al. (2007). Another method was recently developed by Lindgren (2010).”
Network structure is related to social conflict in a cooperatively breeding fish (2013, methodology): “We performed separate analyses for each dominance matrix for each social group, using the double Dekker semipartialling method (Dekker et al. 2007) with 2000 permutations per analysis.”
Beyond and Below Racial Homophily: ERG Models of a Friendship Network Documented on Facebook1 (2010, methodology): “Furthermore, for the OLS regressions above, we rely on the “double semipartialing” permutation technique (Dekker et al. 2007) to calculate significance levels for parameter estimates.”
Multilevel social organization and space use in reticulated giraffe (Giraffa camelopardalis) (2014, methodology): “assignment procedure (Krackhardt 1988; Dekker et al. 2007) to determine the e"ect of shared space use on AS (arcsine transformed).”
Spatializing the Social Networks of Gangs to Explore Patterns of Violence (2011, background): “Though these are highlighted elsewhere (see Tita and Greenbaum 2009), the current research focuses specifically on the role of urban street gangs in explaining the spatial distribution of violence (see Decker 1996; Wilkinson and Fagan 1996; Cohen et al. 1998; Cohen and Tita 1999; Rosenfeld et al. ”
How Do Cultural Producers Make Creative Decisions?: Lessons from the Catwalk (2009, methodology): “2002) was used to run the QAP regressions because it presents the most reliable procedure under a number of multicollinearity and autocorrelation conditions (Dekker et al. 2007).”
Exponential Random Graph Models for Social Networks (2014, neutral): “L’inte´reˆt de cette classe de mode`les est alors de traquer la de´pendance des observations a` l’e´chelle locale plutoˆt que l’e´vacuer comme c’est le cas par exemple des proce´dures QAP.”
Where Do Creative Interactions Come From? The Role of Tie Content and Social Networks (2011, methodology): “Finally, to test further the robustness of our regressions to potential network autocorrelation, additional models (not reported here) using multiple regression quadratic assignment procedures were estimated (MRQAP; Krackhardt 1988, Borgatti et al. 2002, Dekker et al. 2007).”