Title: Using the Methods of Ridge Regression and Principle Components to Estimate the Parameters of Logistic Model Under Multicollinearity By Using Simulation
Abstract: The problem of the Multicollinearity , can appear in the model of propensity scores (PS) when estimating the average treatment effects (ATEs). In this research the logistic ridge regression and logistic principle components regression are used as an alternative methods to the maximum likelihood estimates in the propensity scores model. The average treatment effects (ATEs) estimators adopted the method of inverse probability weighted (IPW )and then estimate the average treatment effects (ATEs), where the use of simulations (Monte Carlo) to generate tracking data model of logistic regression and Multicollinearity problem depending on various factors, from simple correlation coefficient values and sample size and the number of independent variables and the constant value plus the adoption of different designs of propensity score in simulation study, this is due to the fact that estimates the average treatment effects (ATEs) has strengths and different accounts. And we use Bias and the mean squares error (MSE) as criteria for comparing methods of estimation.The results that have been obtained by using a simulation study indicates that the Bias and the mean square error (MSE) depend on the sample size and the degree of the correlation as well as the design propensity score model. I have observed that the estimation method of logistic ridge regression (LRR) and principle component logistic regression (PCLR) was the best of the maximum likelihood estimates (MLE)
Publication Year: 2017
Publication Date: 2017-09-01
Language: en
Type: article
Access and Citation
AI Researcher Chatbot
Get quick answers to your questions about the article from our AI researcher chatbot