von Davier's applied research has focused on utilizing psychometric methods in international large-scale assessment. In his roles at ETS and Boston College,he led the psychometric work on transitioning the PIAAC 2012,the PISA 2015,the TIMSS 2019 and the PIRLS 2021 from a paper-based to a computer-based trendline using mode effect models with data from studies that were designed to align results from paper and computer-based assessments. [34]
Another line of his research has concerned the more general issue of linking in large-scale educational assessments. [35] [36]
A third line of von Davier's research has discussed the response styles and correcting for survey response bias in self-reports. The applications range from mixture models for personality data [37] [38] to the pitfalls of attempts to correct response bias by anchoring vignettes. [39] More recently,his research's focal point was the use of process data in assessment to improve achievement estimation and contextualize assessment results. [29] [30] [40]
Publications
von Davier has authored and co-authored over 150 publications in peer-reviewed journals,edited books,monographs,and research report series. His h-index is 52. He co-edited several books on topics ranging from Latent Variable Models in Psychometrics to International Large-Scale Assessments and NLP in Assessment. [41]
von Davier's first book,Multivariate and Mixture Distribution Rasch Models:Extensions and Applications,explored the advanced applications and extensions of the Rasch model across various disciplines,including education,psychology,health sciences. Allan S. Cohen commented in the Journal of the American Statistical Association,"This book,published in honor of the retirement of Jürgen Rost,is an edited volume of 22 invited chapters written by eminent researchers in the field of item response theory (IRT)." [42] His next book,The Role of International Large-Scale Assessments:Perspectives from Technology,Economy,and Educational Research,published in 2012,discussed the significance of large-scale international assessments as catalysts for change in understanding the role of human capital distribution,impacting policy,education,and research. In 2013,he co-edited the Handbook of International Large-Scale Assessment:Background,Technical Issues,and Methods of Data Analysis,with Leslie Rutkowski and David Rutkowski,which explored the methodology,technical details,and policy implications of International Large-Scale Assessments (ILSA) in education. Terry Ackerman remarked,"This book is an excellent resource and guide to international large-scale assessments or ILSAs. The three editors have done an excellent job identifying a group of prominent scholars whose expertise ranges from international testing and behavioral statistics to educational policy." [43]
Alongside Randy E. Bennett,von Davier published Advancing Human Assessment:The Methodological,Psychological and Policy Contributions of ETS in 2017,detailing the advancements in human assessment made by ETS,covering measurement and statistics,education policy,psychology,and the development of widely used educational surveys and methodologies. Building upon this exploration of assessment methodologies,he co-edited Advancing Natural Language Processing in Educational Assessment in 2023 with Victoria Yaneva,which looked into the implementation,benefits,and challenges of using NLP in educational testing and assessment. In addition,his book,Handbook of Diagnostic Classification Models:Models and Model Extensions,Applications,Software Packages,provided an overview of diagnostic classification models (DCMs),discussing their development,application,and advantages in offering detailed evaluations of test taker performance across multiple skill domains compared to traditional assessment models. Yu Bao reviewed the book and stated,"The Handbook of Diagnostic Classification Models serves as a reference book that consists of a comprehensive collection of the majority of research topics and a summary of the influential publications within recent decades." [44]
In his highly cited studies,von Davier wrote the practices researchers can use for analyzing and reporting data from large-scale international assessments,addressing common issues and statistical complexities to ensure unbiased results. [45] He emphasized the importance of correctly using plausible values in large-scale survey data analysis to avoid biased estimates and underscored the need to follow established procedures and guidelines. [46] Additionally,he presented a diagnostic model for multidimensional skill profiles using maximum likelihood techniques,demonstrated its application with simulated and real data, [47] and introduced general diagnostic models (GDMs) for estimating skill profiles,suitable for polytomous data and missing responses,with a focus on TOEFL Internet-based testing (iBT) field test data. [48] In related research,he showed that the G-DINA and LCDM approaches to diagnostic modeling are special cases of the GDM. [49] Some of his later research focused on large language models,recurrent neural networks,and other so-called AI methods and how they can be used in automated item generation,automated scoring,and other applications in large-scale educational assessment. [31] [32] [33]
