Dmitry Abbakumov

Welcome to my webpage! My name is Dmitry Abbakumov. I am a psychometrician (PhD KU Leuven) and data scientist with experience building scaled psychometric and learning analytics products. My professional mission is to create and deliver added value for EdTech projects through data-driven insights and psychometric inference. Since 2021, I am leading the evidence based learning and technology-enhanced assessment streams at Practicum by Yandex as its Chief Psychometrician. On this website, you can find my CV, publications, and information on my projects. You can reach me at mailbox@abbakumov.com.

Current Project

At Yandex, I am working on a psychometric machine for evidence-based learning. The engine consists of theoretically-grounded, data-driven, explainable solutions to show when and why learning happens and to explain how learning products work. I have presented the first results at the Yet another Conference on Education 2021. Please find the video (in Russian only, sorry) below.

Publications

Kalinichenko, N., Velichkovsky, B., & Abbakumov D. (2021). Empirical evaluation of Russian version of the Technology Acceptance Questionnaire. Psikhologicheskie Issledovaniya, 14(78), 7. PDF (in russian)

Abbakumov, D., Kravchenko, D., Kuskin, W., & Urban, A. (2020). How rewatching video lectures impacts students' performance in assessments in MOOCs. https://doi.org/10.35542/osf.io/vcd5g

Kravchenko D., Bleskina I., Kalyaeva E., Zemlyakova E., & Abbakumov D. (2020). Personalization in education: from programmed to adaptive learning. Journal of Modern Foreign Psychology, 9(3), 34-46. PDF (in russian)

Abbakumov, D., Desmet, P., & Van den Noortgate, W. (2020). Psychometrics of MOOCs: Measuring learners' proficiency. Psychologica Belgica, 60(1), 115-131. PDF

Abbakumov, D., Desmet, P., & Van den Noortgate, W. (2020). Rasch model extensions for enhanced formative assessments in MOOCs. Applied Measurement in Education, 33(2), 113-123. PDF

Abbakumov, D., Desmet, P., & Van den Noortgate, W. (2019). Measuring growth in students' proficiency in MOOCs: Two component dynamic extensions for the Rasch model. Behavior Research Methods, 51(1), 332-341. PDF

Abbakumov, D., Desmet, P., & Van den Noortgate, W. (2018). Measuring student's proficiency in MOOCs: Multiple attempts extensions for the Rasch model. Heliyon, 4(12), 1-15. PDF

Abbakumov, D., & Lebedeva, M. (2016). Russian as a foreign language grammar and vocabulary placement test: Design, pilot test, and psychometric analysis. Russian Language Abroad, 5, 70-75. PDF (in russian)

Abbakumov, D. (2011). Comparing the effectiveness of verbal and numerical tests for predicting results of employees' productivity. Organizational Psychology, 1(2), 92-99. PDF (in russian)

Selected Talks

Building the engine for evidence based learning. Yet another Conference on Education by Yandex, Moscow, Russia, 2021

Educational measurement for EdTech: What we need for evidence based pedagogy? eLearning Stakeholders and Researchers Summit by HSE University and Coursera, Virtual, 2020

Ensuring the validity of data and the validity of predictions in digital learning. Times Higher Education Virtual Digital Transformation Forum, Virtual, 2020

Combining explanatory IRT and psychological networks for understanding and modeling online learners' difficulties. International Meeting of the Psychometric Society, Virtual, 2020

Measuring student's activity in MOOCs using extensions of the Rasch model. International Meeting of the Psychometric Society, Santiago, Chile, 2019

Learners' activity in MOOCs from a psychometric perspective. Coursera Partners Conference, London, the UK, 2019

Measuring student's proficiency in MOOCs: Multiple attempts extensions for the Rasch model. International Meeting of the Psychometric Society, New York, USA, 2018

Measuring student's proficiency in MOOCs: Multiple attempts IRT extensions. Coursera Partners Conference, Tempe, USA, 2018

Measuring the impact of video lectures on learner's productivity. Coursera Partners Conference, Boulder, USA, 2017

Interest and interestingness: The new perspective on students and content. Coursera Partners Conference, the Hague, the Netherlands, 2016

Psychometrics of MOOCs: The basics. EduLab, Rome, Italy, 2016

Computerized adaptive testing algorithm for summative assessment. Computerized Adaptive Testing Summit, Princeton, USA, 2014

Doctoral Dissertation

Psychometrics of MOOCs: How to Measure Proficiency?

Supervisor: Prof. Dr. Wim Van den Noortgate, Co-supervisor: Prof. Dr. Piet Desmet

Date of Public Defense: 13 September 2019

The dissertation consists of five stand-alone chapters following the introductory chapter – four empirical and one conceptual. In the four empirical chapters, we have proposed extensions for the most common IRT model, the Rasch model, aimed at solving the following problems. In Chapter 2, we have improved the proficiency measures by modeling the effect of attempts and by involving non-assessment data such as learners’ interaction with video lectures and practical tasks. In Chapter 3, we have modeled individual growth in proficiency through the MOOC as an effect of the cumulative sum of video lectures a learner has watched before responding on a summative assessment item. In Chapter 4, we have established a more nuanced insight on the role of proficiency on the learners’ performance by involving one extra latent effect, the effect of learners’ interest. In Chapter 5, we have proposed a way to measure learners’ activity (e.g., watching videos, reading texts) as influenced by a latent learner characteristic and a latent content characteristic. In the final, reflective and conceptual, chapter (Chapter 6), we have summarized a connection between psychometrics, as a scientific discipline, and MOOCs, as an industry, and have sketched the future development of psychometrics of MOOCs.

Full Text PDF

The Open Lab (Archived)

After defending the PhD, I have initiated a non-profit project, The Open Lab for Psychometrics of Digital Learning. The lab worked in 2019/2020 academic year and brought together scholars and experts in psychometrics, machine learning, and educational sciences from HSE University, University of Colorado Boulder, University of Illinois Urbana-Champaign, Coursera, and Pearson to study how digital learning works with psychometric methods. During that time, we have created an explanatory psychometric network of difficulties and problems of digital learners, explored how rewatching video lectures impacts learners' performance in MOOCs, and organized five international online seminars.