Dynamic impact of social network on knowledge contribution loafing in mobile collaboration: a hidden Markov model. Issue 9 (15th October 2019)

• Record Type: Journal Article
• Authors: Jiang, Shan; Zhang, Xi; Cheng, Yihang; Xu, Dongming; Ordoñez De Pablos, Patricia; Wang, Xuyan
• Is Part Of: Journal of knowledge management. Volume 23:Issue 9(2019)
• Page Start: 1901
• Page End: 1920
• Publication Date: 2019-10-15
• Subjects: Hidden markov model -- Network centrality -- Knowledge contribution loafing -- Mobile collaboration; Knowledge management -- Periodicals; 658.4
• DOI: 10.1108/JKM-10-2018-0641
• Languages: English
• ISSNs: 1367-3270

Abstract

Abstract : Purpose: Social loafing in knowledge contribution (namely, knowledge contribution loafing [KCL]) usually happens in group context, especially in the mobile collaboration tasks. KCL shows dynamic features over time. However, most previous studies are based on static assumption, that is, KCL will not change over time. This paper aims to reveal the dynamics of KCL in mobile collaboration and analyze how network centrality influences KCL states considering the current loafing state. Design/methodology/approach: This study is based on empirical study design. Real mobile collaboration behavioral data related to knowledge contribution were collected to investigate the dynamic relationship between network centrality and KCL. In total, 4, 127 chat contents were collected through Slack (a mobile collaboration APP). The text data were first analyzed using the text analysis method and then analyzed by a machine learning method called hidden Markov model. Findings: First, the results reveal the inner structure of KCL, showing that it has three states (low, medium and high). Second, it is found that network centrality positively influences individuals involved in medium and high loafing state, while it has a negative influence on individuals with low loafing state. Research limitations/implications: The limitations are related to the single machine learning method and no subdivision of social network. First, this paper only uses one kind of text classification model (TF-IDF) to divide chat contents, which may not be superior to other classification models. This paper considers the eigenvector centrality, and not further divides the social network into advice network and expressive network. Practical implications: This study helps companies infer tendency of different KCL and dynamically re-organize a mobile collaborative team for better knowledge contribution. Originality/value: First, previous studies based on static assumptions regarding KCL as static and the relationship between loafing reducing mechanisms and team members KCL does not change over time. This study relaxes static assumptions and allows KCL to change during the process of collaboration. Second, this study allows the impact of network centrality to be different when members are in different KCL states.