SPEAKER：Paul Benjamin Lowry (Chair Professor, Pamplin College of Business, Virginia Tech)

TIME/DATE：2024年10月28日 14:00

CLASSROOM：同济大厦A楼308室

ABSTRACT

Statistical methods have been foundational in Information Systems (IS) research, with techniques like Structural Equation Modeling (SEM) and meta-analysis instrumental in testing research hypotheses against data. SEM allows for evaluating complex models, such as path, mediation, and confirmatory factor analysis (CFA) models. However, SEM has notable limitations in synthesizing results across multiple studies. In contrast, though underutilized in IS compared to SEM, meta-analysis is recognized for its ability to aggregate results across a broad body of literature, offering a more evidence-based approach suited to the IS field.

Despite its benefits, traditional meta-analysis falls short when dealing with models that involve mediation analysis, CFA, or SEM constructs. This limitation becomes particularly problematic when testing established IS theories, such as the Technology Acceptance Model (TAM) or Protection Motivation Theory (PMT), which rely on nomological networks of constructs that are difficult to synthesize across multiple studies.

To address these limitations, researchers have developed Meta Analytic Structural Equation Modeling (MASEM), which integrates the strengths of both SEM and meta-analysis. MASEM provides several advantages, including testing the fit of proposed models, estimating parameters such as regression coefficients, and extending analyses to models involving latent variables.

MASEM has become increasingly prominent in IS research, with several high-impact studies leveraging its capabilities. However, gaps remain in the application and interpretation of MASEM, particularly in utilizing the latest methodological advances. This presentation aims to introduce IS researchers to these advances and provide a comprehensive framework for conducting rigorous MASEM studies.

MASEM combines meta-analysis and SEM, allowing researchers to test structural models constructed from meta-analytic data. Unlike traditional meta-analysis, MASEM facilitates testing relationships among variables, including mediation effects, within theoretical frameworks. This makes it a powerful tool for theory testing in IS, particularly for comparing competing theoretical models and assessing the validity and applicability of these models across a broader range of empirical data.

There are several approaches to MASEM, including correlation-based and parameter-based methods. The most common approach, the univariate-r method (URM), has been widely used in IS research but has significant limitations. It treats SEM and meta-analysis as separate stages, which can lead to statistical inconsistencies. A more robust approach, Two-Stage Meta-Analytic Structural Equation Modeling (TSSEM), uses a multivariate approach to pool correlation matrices and addresses many of the limitations of URM, such as handling missing data and accounting for interdependencies among effect sizes.

This presentation outlines the latest methodological standards for conducting MASEM using TSSEM, focusing on addressing the statistical limitations of URM. It also lays out the theoretical implications for IS researchers who aim to test and build theory, and want to provide effective suggestions to practice on “what works.” By following these guidelines, researchers can enhance the accuracy and generalizability of their findings, ultimately contributing to a more robust body of IS research.

GUEST BIO

Prof. Paul Benjamin Lowry, Ph.D., is an Eminent Scholar and the Suzanne Parker Thornhill Chair Professor in Business Information Technology at the Pamplin College of Business at Virginia Tech, where he serves as the BIT Ph.D. and Graduate Programs Director. He also is a visiting professor at Nanyang Technological University in Singapore. He is a former tenured Full Professor at City University of Hong Kong and The University of Hong Kong. He received his Ph.D. in Management Information Systems from the University of Arizona and an MBA from the Marriott School of Business. He has published 290+ publications, including 170+ journal articles in the Journal of Management Information Systems (JMIS), Information Systems Research (ISR), MIS Quarterly (MISQ), Journal of the Association for Information Systems (JAIS), Information Systems Journal (ISJ), European Journal of Information Systems (EJIS), Journal of Strategic Information Systems (JSIS), Journal of Information Technology (JIT), Decision Sciences Journal (DSJ), various IEEE Transactions, and others. In the last 10+ years, he has been consistently ranked in the top 5 in the world in various top information systems journal rankings. He is on the senior board of editors at the JMIS. He also is an SE at ISJ and an AE at ISR. He has previously served as co-EIC of THCI; DE at DSJ, SE at JAIS, SE at THCI, guest SE at MISQ, guest SE at JMIS, guest SE at EJIS, guest SE at JIT, guest SE at SGR, AE at EJIS, AE at I&M, AE at ISJ, AE at ECRA, AE at SGR, AE at CAIS, and guest AE at MISQ. His research interests include (1) organizational and behavioral security and privacy; (2) online deviance, online harassment, and computer ethics; (3) HCI, social media, and gamification; and (4) business analytics, decision sciences, innovation, and supply chains.