DOI: 10.2307/2110459
关键词:
摘要: Reexamining Miller and Stokes's 1958 representation data, this article demonstrates that Stokes may have underestimated the extent of congressional in their classic study. It is argued correlations between sampled constituency opinion behavior are more seriously attenuated by sampling error than has been recognized. Unlike opinion, measures based on simulation shown to correlate relatively highly with behavior. Evidence presented which indicates elections a major source representation.