Marcus Berzofsky is a senior research statistician who has focused on survey sampling and methodology while at RTI International. He has designed complex survey samples and developed post-survey weight adjustments for both establishment and household surveys using both single and multi-stage designs. He has determined the appropriate sample size and allocation for samples based on frames created from administrative sources, commercial sources, found data, and random digit dial telephone lists. He has worked with multi-frame studies and developed composite estimators to combine respondents which may appear on more than one frame. Additionally, Dr. Berzofsky has developed new methods for selecting samples from social media and creating hybrid estimates by combining nonprobability samples with probability samples. On the post-survey side of the design, he has developed imputation, weighting, and estimation methods to minimize bias and precision. Furthermore, Dr. Berzofsky conducts research in all aspects of total survey error including measurement error, classification error, nonresponse error, coverage error, and processing error.
Currently, Dr. Berzofsky works on studies for the U.S. Bureau of Justice Statistics (BJS) as well as several state governments and universities. He is the Principal Investigator on the NCS-X NIBRS Estimation project whose aim is to develop imputation and estimation methods to provide national- and state-level estimates data from the National Incident-Based Report System – the nation’s administrative data source of crimes reported to the police. Additionally, he is Co-PI on the National Victimization Statistical Support Program which evaluates and develops the methodology used in producing victimization statistics including those from the National Crime Victimization Survey. He additionally designs campus climate surveys for several universities to measure sexual assault rates among students at their schools. Furthermore, he works on population-based studies assessing access the health care and the health care status of the general population and the Medicaid eligible population.
