RL2025-2
An Introduction to Missing Data Analyses for Educational Research
Wednesday, June 11, 2025
  
INSTRUCTORS

Brian T. Keller, University of Missouri

Craig K. Enders, University of California, Los Angeles 

This course provides foundational knowledge about missing data analyses. The course content includes missing data assumptions, Markov Chain Monte Carlo (MCMC) estimation, missing data imputation, incomplete categorical variables, and incomplete interaction effects. The course includes a mix of lectures and software demonstrations. Attendees will be provided with the following materials: lecture slides built around analysis examples from a real educational data set; free statistical analysis software, Blimp, developed by the instructors; a 100+ page white paper that provides details about a range of missing data topics; and a 250+ page annotated software tutorial guide that provides step-by-step instructions for 20 common statistical analyses. Because missing data are ubiquitous in educational research settings, the course content would appeal to virtually any educational discipline that relies on quantitative methods. The target audience includes graduate students, professors, and research professionals who use, but do not specialize in, quantitative methods. To maximize accessibility, the only prerequisite is a working knowledge of statistical concepts from a typical first-year graduate statistics sequence, in particular multiple regression. The course instructors co-develop the free software application Blimp, available at www.appliedmissingdata.com/blimp.

  
RL2025-3
Quantitative Tools for Qualitative Data Analysis: A Truly Equal Status Data Science Design for Transparency, Rigorosity, and StoryTelling
Tuesday, June 17, 2025
  
INSTRUCTOR

Manuel Gonzalez Canche, University of Pennsylvania

This highly applied hands-on virtual course is ideal for qualitative and mixed methods researchers. It will feature four methodological frameworks along with their respective free and no-code software tools designed to:

1. Classify long texts like interview or observation data (LACOID published paper access here https://cutt.ly/SwiFCSj4) based on machine learning,
2. Classify short texts like social media posts or open-ended responses (MDCOR published paper access here https://cutt.ly/Xwhk55sP) while also relying on machine learning,
3. Identify the distribution of sentiments and emotions in raw or classified texts (SENA paper in print access here https://cutt.ly/aeXMaBxS) using natural language processing and linguistic analysis, and
4. Capture the geospatial contexts where our participants' stories took place across time (GeoStoryTelling published paper access here https://cutt.ly/qrgauhna) via interactive visualization, geographical information systems, and multimedia integrations.

The common thread across all these published or accepted methodologies and software tools is their alignment with Truly Equal Status Design (TESD), wherein the balanced interweaving of qualitative and quantitative outputs is vital for researchers to be able to construct deep understandings of the structure and meaning of our qualitative sources of evidence. This interdependence helps researchers maintain clarity and transparency in the rationale driving the interpretation of the data and findings. In other words, following TESD, quantitative outputs are to be rendered useless without going back to our original qualitative inputs. It is only when we integrate and intermingle quantitative outputs with our unaltered textual data that deeply powerful and nuanced understandings may be gained and conveyed. In short, quantitative outputs are to be used as a map that may help us gain better, clearer, and more nuanced understandings of our participants’ experiences, as expressed in their unmodified narratives.

A value added to this course is that all processes may be achieved with software that runs locally, that is, without requiring us to upload our data to any server. Moreover, these software tools conduct a combination of machine learning text classification, natural language processing, and high level interactive visualizations without any statistical or computer programming or coding requirements. Our reliance on no-code software is an attempt to democratize access to data science among qualitative and mixed methods researchers.