Introduction to Quantile Regression for Education Research

Session Type: Professional Development Course

Abstract

One of the most commonly used quantitative statistical tools in education research is the regression model. These models are limited by strict assumptions on the types of data that can be used and the types of relationships they can describe. Quantile regression (QR; Koenker & Basset, 1978) was designed to overcome these shortcomings, and is beginning to be used in education research. It is an excellent fit to education research as it can analyze and accurately describe heteroschedastic relationships, which are common when examining development of new skills (floor effects) or skill mastery (ceiling effects). QR also allows for more complex research questions to be asked of the data, such as comparing whether the strength of a relation is significantly greater at the high- or low-end of the outcome.
In the proposed methodological mini-course, provided in combination lecture, discussion, and hands-on activity format, attendees will get a conceptual introduction to quantile regression, guidance in crafting research questions suitable for quantile regression, hands-on practice using provided data and your own data, and learn best practices in model building and presentation of results for publication. Come to this workshop if you are at any stage of your career, have a working knowledge of R or SAS, have experience analyzing and interpreting data using regression or regression-based models, and want to learn how to fit quantile regression within your research area. Please bring a laptop with SAS or R and Rstudio installed, and your own de-identified dataset in a .csv format.

Sub Unit

• Professional Development and Training Committee

Instructor

• Jessica Logan, The Ohio State University - Columbus