Three Studies On Information Technology In Distributed Work Environment

Mr. Liu Runfeng
Dr Zhai Yingda, Assistant Professor, School of Computing

14 Apr 2023 Friday, 02:00 PM to 03:30 PM

MR20, COM3-02-59


This thesis seeks to explore the deviant behavior and digital communication patterns of information workers in both centralized office and distributed work environments. Effective information diffusion within an organization, facilitated by IT, is a critical factor for enhancing productivity. However, existing studies have limitations in establishing causal relationships and often fail to account for the dynamic nature of worker interactions in an organizational context. Therefore, this thesis aims to comprehensively examine the working and communication behaviors of workers in a social network, particularly in the context of transitioning from a centralized office to a distributed workforce.

The primary objective of the first study is to investigate the contagious nature of cyberloafing behavior in the workplace, taking into account group size and consistency. Using the social learning theory as the overarching theoretical framework, our conceptual model provides a theoretical explanation of the dynamics of workers' cyberloafing behavior under different organizational environments. Moreover, we propose three influence mechanisms - physical proximity, work nature closeness, and positional equivalence - under which social influence may have varying effects on workers' cyberloafing behavior under different conditions. We conducted our empirical analysis using observational network activity data and compared how new workers' behavior evolved after joining teams with different structures. Our findings reveal that social influence significantly impacts workers' cyberloafing behavior and is more pronounced in small groups with high consistency. Additionally, we found that close-work pairs holding a similar position exert a stronger influence on cyberloafing behavior.

In the second study, we propose a novel linear effect model to identify the average treatment effect in the presence of social interactions. The widely-used difference-in-differences (DID) model is commonly employed to draw causal inferences in quasi-experimental designs. However, it may produce a biased estimate when social interactions exist between the treatment and control groups, which is a common occurrence in empirical settings of information systems and management research. Our study extends the DID framework by decomposing the overall effect into the refined average treatment effect and the biases caused by social interactions with respect to different reference groups. We also explore the bias-variance trade-off in our estimation when subjects experience varying degrees of social interactions. Using an email network dataset from a natural experiment on working-from-home (WFH), we examine the policy effect of WFH on productivity and its implications for the hierarchical communication efficacy of firms. Our results reveal that WFH enhances workers' communication by increasing the number of new connections, fostering deeper conversations with more recipients, and promoting faster email response times with greater work-hour flexibility. Overall, our findings suggest that WFH promotes a flatter communication structure within an organization and makes workers more responsive.

As a complement to the second study, we utilized email traffic data to create both the managerial and email communication network, and investigated the evolving nature of the email communication network. By employing a social network analysis approach, we treated each individual as an autonomous node and generated a sequence of weekly communication networks based on email traffic logs. Specifically, we examined three key features of the network structure: the speed of information receipt, the extent of network reach, and the level of network homophily. The analysis revealed that the communication network underwent significant changes following the implementation of WFH, becoming more interconnected, exhibiting less homophily, and featuring shorter communication distances.

In summary, our research on the distributed work environment provides valuable insights into the dynamic nature of workers' behavior and communication patterns, especially in the context of the increasing prevalence of work-from-home arrangements. Our studies on organizational email communication demonstrate that WFH can enhance communication effectiveness and streamline communication structures by disseminating information to more relevant parties with fewer layers of communication. These results suggest that WFH could be a viable option for workforce managers to improve communication efficiency, especially during and after pandemics. Moreover, our study on cyberloafing sheds light on the existence of social influence in the workplace and highlights the importance of considering organizational environments when implementing formal controls, such as anti-cyberloafing policies. These findings provide valuable insights for managers and policymakers seeking to optimize work arrangements and enhance productivity in the modern workplace.