Speaker: Jack Davis
Title: "Don't just take the middle!" Estimation methods of underlying values of binned data.
Abstract: Binned data is numerical data for which only a pre-defined range or 'bin' for the data is known. It is commonly used in social science surveys to simplify a numerical response (e.g. How many times have you...) into a multiple choice format. Many statistical methods require data to be recorded as specific numbers. This talk showcases a maximum likelihood method of estimating the underlying values of binned data. The talk expands upon this method with two ways to incorporate covariate data into those estimates using proportional odds logistic regression.