|Title:||Methods and Software for Handling Network Data and Text Data in Structural Equation Modeling|
|Principal Investigator:||Zhang, Zhiyong||Awardee:||University of Notre Dame|
|Program:||Statistical and Research Methodology in Education [Program Details]|
|Award Period:||3 years (07/01/2021 - 06/30/2024)||Award Amount:||$861,354|
|Type:||Methodological Innovation||Award Number:||R305D210023|
Co-Principal Investigators: Yuan, Ke-Hai; Wang, Lijuan
The purpose of this grant is to combine structural equation modeling (SEM) techniques and data science methods to model network and text data. Network and text data are increasingly collected in many fields of research, business, and government. For example, to study student behaviors, it is important to understand the context of behaviors because students are not independent entities but are typically connected with one another, which naturally leads to the collection and analysis of network data. For teacher evaluations, narrative comments on different aspects of teaching can provide teachers rich information and valuable feedback over and beyond numerical ratings. Such data, however, present analytical challenges which have not yet been fully met by existing statistical techniques and software.
The research team will create easy-to-use software package called BigSEM to implement the proposed methods for analyzing network and text data. The research team will develop BigSEM as both an R package to allow future growth in capability and a web application (https://bigsem.org) so that researchers can conduct complex data analysis through drawing a path diagram. As part of the software development process, the team will conduct simulation studies to test the functionality of BigSEM, prepare real-data examples of the software's capabilities, and conduct multiple rounds of user-testing. The research team will also publish in journals and at conferences, while making available on the website: the software code, the real data used for illustration, and an extensive user's guide for BigSEM.