Handling missing data: Multiple imputation for the applied researcher

Home / Research Methods  / Handling missing data: Multiple imputation for the applied researcher

A 4-hour workshop taught by Ryan S. Wells, Ph.D.

You may also be interested in our increasing web survey response rates workshop.

See when this workshop is offered next »

Overview of our missing data workshop workshop

Item-missing data is a serious concern for any quantitative researcher. Survey participants regularly skip questions, and administrative data is often incomplete. But what do to about this issue is not always straightforward. Researchers often analyze only complete cases (listwise deletion) or they impute missing information using means or regression analysis. These approaches can lead to biased estimates and artificially small standard errors.

This workshop will help the applied researcher to analyze the extent and patterns of missing data, and to decide how to address this challenge. Most of the workshop will focus on one common but powerful solution to missing data: multiple imputation. This technique has advantages over many other approaches, which will be discussed. At the end of this workshop, participants will understand the decisions, strategies, and code needed to successfully use multiple imputation in their research projects.

Expected outcomes of our missing data workshop workshop

By the end of the workshop, participants should understand the problems associated with missing data for quantitative research, as well as options for how to handle such data. Participants will gain specific insight about multiple imputation as an approach to dealing with missing data, including the assumptions and data requirements underpinning the method. By the end of the workshop, participants will understand the decisions, strategies, and statistical methods needed to successfully apply multiple imputation to their projects.

Who should attend?

The target audience is researchers who have taken more than one statistics course. Basic knowledge of regression is useful. Those who will benefit the most are those who conduct quantitative research and may encounter missing data in their projects. Software demonstrations will use Stata, but SPSS code and output will also be included for participants who are using SPSS for their research projects.

Agenda

  1. What is “missing data?”
  2. Nature and structure of missing data
  3. Dangers of missing data
  4. Possible ways to address to missing data
  5. Multiple imputation – the basics
  6. Pooling – Rubin’s Rules
  7. Number of imputations
  8. Variable selection / Auxiliary variables
  9. Comparing observed and imputed results
  10. Comparing multiple approaches to missing data
  11. Imputing complex survey or weighted data
  12. Common mistakes and confusion
  13. Communicating MI to other audiences

About the instructor

Ryan Wells is associate professor and program coordinator in higher education. Wells primarily uses quantitative research methods to study questions of college access and equity for underrepresented groups, as well as resulting social inequality. Topics for recent research on college access and equity include socioeconomic status, gender, and disability status. His research has been published in top journals such as The Journal of Higher Education, Teachers College Record, Research in Higher Education, and Community College Review.

Learn more about Dr. Wells here.

See when this workshop is offered next »