Hello, I am using EI to estimate crossover voting in the Top 2 Primary in California in 2012. The unit of analysis is at the precinct level and I match the turnout record of partisan voters with election return data. For simulation, I use the following formula. formula <- cbind(dem.return, rep.return, oth.return) ~ cbind(dem.show, rep.show, oth.show) dem.return : Vote Share of a Democratic Candidates at the precinct level rep.return : Vote Share of a Republican Candidates oth.return : Vote Share of a Non-Two Party Candidate dem.show : Ratio of Democratic Registered Voters among who casted ballot for the primary rep.show : Ratio of Republic Registered Voters among who casted ballot for the primary oth.show : Ratio of Non-two Party Voters among who casted ballot for the primary My quantity of interests are the ratio of democratic registered voters who voted for the republican candidates (Democratic Crossover to Republican Party) and the ratio of republican registered voters who voted for the democratic candidates (Republican Crossover to Democratic Party). Using precinct level data, I am trying to estimate population weighted average of crossover voting at the State Assembly District. In some districts, the estimate looks reasonable. However, in other districts, my estimates suggest that more than half of each partisan voters crossover to other parties. I wonder what would be the best way to examine how reasonable my EI estimates are. Thank you for your help. Best, Woo Chang KANG -- Woo Chang KANG Ph.D. Candidate Department of Politics New York University 19 West 4th St. 2nd Fl., New York, NY 10012. - --- EI mailing list served by HUIT Subscribe/Unsubscribe: http://lists.gking.harvard.edu/mailman/listinfo/ei More info on ei and eiR: http://GKing.Harvard.edu/eiR EI mailing list EI(a)lists.gking.harvard.edu To unsubscribe from this list or get other information: https://lists.gking.harvard.edu/mailman/listinfo/ei