Type | Thesis or Dissertation - Bachelor or Statisitics |
Title | Nonparametric inequality measure based on ranks |
Author(s) | |
Publication (Day/Month/Year) | 2006 |
URL | http://www.stat.duke.edu/~dp55/rank.pdf |
Abstract | In this paper we study the presence of inequality among the states of India with respect to several attributes. In this context we propose a non-parametric measure of inequality based on ranks. We use nonparametric measure as it requires minimal assumptions on the model and is unaffected by small measurement errors. The description of our proposed statistic incorporates several factors that are likely to affect inequality. Due to its generalized nature, it is difficult to get a closed-form expression of the value of the pro-posed statistic for large data-sets, except for a few highly restricted cases. Thus we have taken resort to combinatorial optimization techniques involving Markov chain Monte Carlo simulations. We have used the simulated annealing algorithm because of its manifold advantages. Finally we use the statistic to conclude if there is any trend in inequality amongst the states over the past decade. |