Hok Chio Lai
University of Southern California, CA·Psychology·Early-career
Over the past two decades, data sharing has created many large public datasets in the social and behavioral sciences. Combining several datasets can give researchers more statistical power and let them ask broader questions.
- Fields
Psychology
Mathematics & Statistics
Social Science
- Topics
integrative data analysis
data harmonization
psychometrics
- Grants
- 1 known grant, through Jul 2027
What they work on
- Develop a robust optimization algorithm for data harmonization that puts different instruments measuring the same concept on a common scale while also correcting for measurement bias across groups.
- Develop and validate two-stage random-effect path analysis, a method for analyzing harmonized scores from pooled datasets that correctly accounts for measurement error.
- Release open-source software for these latent variable methods and create guidelines describing when they perform well and when caution is needed.
More about the research
This approach is called integrative data analysis (IDA). However, different studies often measure the same concept, such as depression or reading skill, with different questionnaires or tests. Scores also contain measurement error. In addition, a test may work differently across groups or over time, a problem known as violating measurement invariance. Existing IDA methods have no single framework that handles all of these problems, so pooled results can be biased or inaccurate.
Why it matters
Researchers in psychology, education and other social sciences increasingly pool data from many studies but often face mismatched measures. Free, validated tools to harmonize these measures and correct for measurement error could make pooled findings more trustworthy. They could also let researchers answer questions that no single dataset can address.
Known grants
From NIH and NSF research grants. Labs often have other funding we can't see.
Closest labs
Peter Halpin
University of North Carolina at Chapel Hill, NC·Education Research
Develop estimators of treatment effects that are robust to item-level heterogeneity, by combining robust statistics with item response theory, the framework linking student ability to the probability of answering each test question correctly.
Amy Willis
University of Washington, WA·Mathematics & Statistics
Develop robust statistical methods for differential abundance, which test whether microbes are more or less common between groups while accounting for bias and variability introduced by sequencing.
Tracy Sweet
University of Maryland, MD·Education Research
Train three cohorts of 17 early career STEM education scholars through year-long intensive institutes on research design, measurement and data analysis for large-scale studies in K-12 and postsecondary settings.
The school
University of Southern California: 47 labs on PI Pages, $1.3B a year on research, 882 research doctorates a year.
Reach out
A start for your email. Fill in the brackets.