Google Correlations: New Approaches to Collecting Data for Statistical Network Analysis

摘要: Social processes such as partisan alliances, clientelism, and elite cooptation could be fruitfully studied using recent advances in the analysis of network data (see Fowler et al. 2011). Yet collecting information on political elites conventional methods surveys archival records can costly and, authoritarian states, prohibitive. This paper introduces a new method for collection networks non-obtrusive web-based techniques. One possible indicator connectivity is frequency with which individuals appear at same events, fundraisers, groundbreakings, receptions. Using Google search scraping algorithm (Lee 2010) to capture how often pairs news articles reporting these I construct matrices given list that identify variety criteria. To assess cross-validity conceptual accuracy, compare from this previously collected connectedness three populations: 108th U.S. Senate, 110th House, Mexican business 2013. then supply an application collect Nigerian oil 2012. Conducting analysis, show appointments National Petroleum Corporation board directors are made basis (i.e., centrality) not necessarily technical experience or merit. These findings lend support hypothesis leaders use patronage lucrative bureaucratic positions order satisfy elites. Given many theories behavior aim understand individual- group-level interactions, potential applicability proposed technique very large, especially situations where intrusively