Generation of data with specific marginal risk difference

• Main Authors: Kazem Mohammad, Mohammad Ali Mansournia, Safoora Gharibzadeh
• Format: Article
• Language: English
• Published: Tehran University of Medical Sciences 2018-07-01
• Series: Journal of Biostatistics and Epidemiology
• Subjects: Data systems; Risk ratio; Causality; Computer simulation; Monte Carlo method
• Online Access: https://jbe.tums.ac.ir/index.php/jbe/article/view/154
• ISSN: 2383-4196; 2383-420X

Background & Aim: Simulation studies are important statistical tools to investigate the performance of statistical models in specific situations. For a binary outcome and exposure, one of the most important statistical measures will be the risk difference (RD). To assess the quality of estimators in estimating the effect of the exposure, a data set with a specific effect measure is require. Methods & Materials: Monte Carlo simulation can be helpful in situations when there is a proper data  generating  process.  In  this  paper,  another  technique  will  be  presented  to  generate  data  with specific marginal risk difference (MRD).  Results: Convergence of simulation methods in the same scenario reached in a few iterations using the proposed method.  Conclusion:  The  proposed  method  is  recommended  over  the  current  method  due  to  less  time consumption; this issue is important in studies with different scenarios.