Paul Bailey and Blue Webb, members of the EdSurvey team, recently published an article, “Expanding NAEP and TIMSS Analysis to Include Additional Variables or a New Scoring Model Using the R Package Dire.” Featured in the journal Psych in its special issue on computational psychometrics, this article showcases the new Dire software and its utility for ushering in a new era of data use at the National Center for Education Statistics (NCES).  

The article explains that NCES’s assessment designs often incorporate short assessments that are designed to provide accurate population-level estimates but tend to be imprecise at the student level. To address this issue, a sophisticated conditioning model is essential for obtaining unbiased estimates―and this is where the Dire software comes in. When used in conjunction with EdSurvey, Dire (short for “direct estimation”) provides a way for users to create a new conditioning model that incorporates variables, including those they merged for their analysis. The user needs to bring together all of the item parameter information as well as test scaling data. EdSurvey does most of the work, and the process becomes just a few lines of code. 

To learn more about Dire, explore its utility in analyzing NAEP data, and view example work with conditioning models, read the full article here. To stay informed on EdSurvey developments and further publications related to the NAEP research community, consider signing up for our mailing list.