Journal of Statistics Applications & Probability
Estimated Path Analysis Parameters Using Weighted Least Square to Overcome Heteroskedasticity at Various Sample Sizes
This study aims to determine a better parameter estimation method between the OLS and WLS methods to overcome the problem of heteroscedasticity in path analysis and to find out the comparison of standard and adjusted errors between the two methods at various sample sizes 𝑅. Path analysis is a complex regression analysis with a direct or indirect causal relationship between several variables. The data used in this research is simulation data. The path analysis model formed consists of two correlated exogenous variables, one endogenous variable and one intermediate variable with the relationship between variables used limited to linear form. The results of this study indicate that the WLS parameter estimation method is better than the OLS method in estimating the path analysis parameters that have heteroscedasticity problems. The parameter estimator between the two methods has no significant difference, but the standard error of the WLS method is smaller than that of the OLS. In line with this, the p-value of the significance of the WLS method parameters was almost entirely significant for the five relationships at various levels of heteroscedasticity. it can also be concluded that the larger the sample size, the smaller the standard error for both the OLS and WLS methods. The models goodness from the adjusted value of the WLS method is higher than the adjusted value of the OLS method.
Digital Object Identifier (DOI)
D. Adyatama, A.; Solimun; A. Efendi, S.; Nurjannah; and B. T. Mitakda, M.
"Estimated Path Analysis Parameters Using Weighted Least Square to Overcome Heteroskedasticity at Various Sample Sizes,"
Journal of Statistics Applications & Probability: Vol. 12:
2, Article 24.
Available at: https://digitalcommons.aaru.edu.jo/jsap/vol12/iss2/24