Graduate Students Seminar
Wednesday, April 24, 2019 · 11 AM - Noon
|Session Chair:||Qing Ji|
Speaker 1: Jing Wang
- Inverse Probability Weighting via Propensity Score in Estimation of Average Causal Effect
- Observational studies and Randomized Controlled Trials (RCTs) are the main types of studies for making inference on the effect of treatment on a response. In RCTs, randomization leads to unbiased comparison between treatment and control. However, in observational studies, there may exist some “confounders”, which are associated with both exposure and outcomes. Thus, unbiased estimation of the treatment effect from observational data requires methods to control confounding.
- Propensity scores were introduced in 1983 as a tool to estimate the causal effect from non-randomized data. Inverse probability weighting using propensity score is one of most widely used methods to reduce confounding. Two different simple weighted estimators and one “augmented” estimator, which has unique “double-robustness” property, will be reviewed, and part of their theoretical properties and implications will also be briefly discussed.
Speaker 2: Ellie Gurvich
- Effects of Seasonal Pesticides on the Honeybee-Varroa Destructor System Model
- The past few decades have had considerable loss in Western honeybee (Apis mellifera) colonies. This disappearance of honeybees, dubbed colony collapse disorder (CCD), has been attributed to be a combination of different factors. Many mathematical models have been developed that investigate the various factors that could lead to CCD. Of extensive interest has been the infestation of Varroa destructor mites, which serve as a vector carrying the Acute Bee Paralysis Virus (ABPV). Ratti et al. expanded on the original work of Sumpter and Martin for this type of model, and formulated an analog to the Ross-Macdonald model of malaria. This paper aims to investigate a simplification and addition of pesticides to the model done by Ratti et al.