Multiple Linear Regression Techniques in Assessing the Effects of Truncation on the Gamma Distribution

Abstract

This study examines, through Monte Carlo simulation, the effects of truncation on the gamma distribution (with one through four parameters). Specifically, three types of truncation (right, left, and double truncation) are considered. Computer facilities were used to generate 400 random samples from the gamma distribution with different parameter values for different sample sizes, shape, scale, location, power, and degree of truncation. Correlation and regression analysis demonstrated that the degree of left truncation has a significant correlation with the measures of central tendency for all distributions. The degree of right truncation had a significant relationship with the measures of deviation. On the other hand, the kind of truncation had a significant unique contribution for all models. However, the type of truncation had significant unique contributions for all models except two (scale and location models for gamma with three parameters).