The purpose of this paper is to discuss a technique which can be used to solve estimation problems for which the more commonly accepted analytical methods such as linear regression analysis are inappropriate. The estimation method proposed consists of using computer alogrithms for numerical minimization of functions. The basic method of functional minimization is discussed and examples are provided. Various measures of fit for non-unique solutions, and statistical tests on the parameter estimates are addressed. The paper concludes that functional minimization is not a panacea for all parameter estimation problems, but for many models can provide parameter estimates and statistical tests which otherwise would not be available.
All Science Journal Classification (ASJC) codes
- Political Science and International Relations