Independent Candidates in a Parliamentary Election in India: A Poisson Regression Model

The paper attempts to explain the number of independent candidates in Indian parliamentary election in the year 2004. The statistical models developed are applications and generalizations of Poisson and Negative Binomial distributions. Our results suggest that the distribution of independent candidates can be explained well with a negative binomial probability model or its generalizations. Our results also help to identify three major factors behind the variations in the number of independent candidates. First, a major determinant of the number of independent candidates is political fractionalization. Results suggest that the number of non-independent candidates would typically lead to more independent candidates in the fray. Interestingly, our analysis points out that the major determinant appears to be political fractionalization at the State level rather than at the constituency itself. Second, we find some indirect evidence of presence of free riders. Free riders typically stand in urban constituencies and against the so called VIP candidates. Third, our results suggest that SC and ST constituencies would have typically lower number of independent candidates due to lack of potential candidates as compared to general constituencies