Anais
Resumo do trabalho
Métodos e Técnicas de Pesquisa em Administração · Métodos e Técnicas de Pesquisa em Administração
Título
STRUCTURAL EQUATION MODELING: IS IT STILL WORTH LEARNING?
Palavras-chave
Open Science
Quantitative Methods
Theory building
Agradecimento:
Ambos os autores são bolsistas PQ do CNPq – Brasil.
Autores
-
Diógenes de Souza BidoUNIVERSIDADE PRESBITERIANA MACKENZIE (MACKENZIE)
-
Cesar Alexandre de SouzaFaculdade de Economia, Administração e Contabilidade da Universidade de São Paulo - FEA
Resumo
Introdução
Although structural equation modeling (SEM) has a long history and is a widely used method in various areas of Applied Social Sciences, it remains a challenging topic for students and new researchers, because learning requires a great effort for many years, and even more so when one has little experience with quantitative methods in general.
The lack of knowledge about this method causes researchers to avoid research problems that require this method, or, even worse, prevents them from formulating such research problems.
The lack of knowledge about this method causes researchers to avoid research problems that require this method, or, even worse, prevents them from formulating such research problems.
Problema de Pesquisa e Objetivo
Even if the researcher does not intend to use this method in their research, they are likely to have to read, study, or express an opinion about an article that employs this method at some point. Still, with superficial reading, one tends to believe that the authors did a good job rather than finding research gaps.
Thus, this article aims to exemplify the critical reading, rereading, and estimation of an alternative model based on the data published in the article under analysis.
Thus, this article aims to exemplify the critical reading, rereading, and estimation of an alternative model based on the data published in the article under analysis.
Fundamentação Teórica
The model is composed of three multidimensional constructs: (i) Facilitating Team Learning Behaviors (FTLB): Task learning, team learning, Boundary crossing, Error communication; (ii) Team learning behavior (TLB): Sharing, Co-construction, Constructive conflict; (iii) Team performance (TP): Team effectiveness, Employee satisfaction, Client satisfaction.
Metodologia
First, we will make a critical reading, comparing the hypotheses with the methods used, results obtained and we will highlight the weaknesses; secondly, we will reread considering the results that are available in the article itself; and thirdly, we will redo the tests of the hypotheses using as input the matrix of correlations available in the article itself, estimating the measurement and structural model through the modeling of structural equations based on covariances using the SmartPLS 4 software with the recently made available option (maximum likelihood) and the lavaan package.
Análise dos Resultados
Convergent validity and reliability were considered adequate because all constructs had an average variance extracted (AVE) greater than 0.5, as well as Cronbach's alpha and composite reliability values greater than 0.7. Apesar da alta correlação entre dois construtos (0.9) a validade discriminante foi confirmada. All structural coefficients (direct effects) were significant (p < 0.001), with very high R² values.
Conclusão
Even when one does not intend to apply SEM, the potential gain from the critical reading and the rereading grounded in the method can lead readers to a depth of understanding that is not possible with floating reading or with the abstracts generated by artificial intelligence (at least for now). We hope that this article raises awareness of these points and serves as a tutorial for beginners.
Contribuição / Impacto
Regarding the article by Professor Tan et al., we can highlight the following: Firstly, they should not have discarded any dimension in the qualitative stage. Secondly, all facilitators promote Team Learning Behavior, which, in turn, enhances team performance. For future studies that use these scales, we recommend taking some a priori measures to minimize common method bias.
Referências Bibliográficas
Cudeck, R. (1989). Analysis of Correlation Matrices Using Covariance Structure Models. Psychological Bulletin, 105(2), 317–327.
Kline, R. B. (2023). Principles and Practice of Structural Equation Modeling (5th ed.). The Guilford Press.
Little, T. D. (2024). Longitudinal Structural Equation Modeling (2nd ed.). The Guilford Press.
Tan, L., Kocsis, A., Burry, J., & Kyndt, E. (2023). Performance of architectural teams: The role of team learning, reflexivity, boundary crossing and error communication. Design Studies, 87, 1–31.
Kline, R. B. (2023). Principles and Practice of Structural Equation Modeling (5th ed.). The Guilford Press.
Little, T. D. (2024). Longitudinal Structural Equation Modeling (2nd ed.). The Guilford Press.
Tan, L., Kocsis, A., Burry, J., & Kyndt, E. (2023). Performance of architectural teams: The role of team learning, reflexivity, boundary crossing and error communication. Design Studies, 87, 1–31.