Abstract:
To compare the estimation in linear regression model using orinary least square (OLS) method, prarmetric bootstrap (PB) and nonparametric bootstrap (NBP) method when the distribution of errors are normal, uniform, logistic, double exponential, smallest extreme value (SEV) and greatest extreme value (GEV), by comparing mena square error, variance, biasedness, mean of mean square error and relative efficiency are criteria for comparing the point estimations. The confidence coefficient from confidence interval of the estimation methods coefficient is criteria for comparing the interval estimation. For point of estimation, it is found that when the distribution of error are from normal, logistic, and double exponential, OLS method is the most efficient, followed by PB method and NBP method is the least efficient. For interval estimation, it is found that when the dustribution of errors are from normal, logistic, SEV and GEV, OLS method is the most efficient, followed by PB method and NPB method is the least efficient. For the case when the distribution of errors are uniform and double exponential, PB method is the most efficient, followed by OLS method and NPB method is the least effecient.