Sustained Participation in Open Source Software Project Communities

ABSTRACT

Sustained participation is critical for the viability of open source software (OSS) project communities (OSSPCs), and this paper explores how sustained participation is maintained in viable OSSPCs. With the lens of the integrative model of trust (IMoT), hypotheses regarding interactions between trust and community citizenship behaviors (CCBs – as OSSPC participating activities) are developed. Both a qualitative study and a quantitative study are conducted, and data analysis confirms both the Trust→CCBs and the CCBs→Trust hypotheses along the time dimension, revealing CCBs-Trust interactions as a mechanism for maintaining sustained community participation. Further, while it is found that CCBs have an accumulative overall positive impact on trust, alternating positive and negative impacts of CCBs on trust over time are identified. In addition, a delayed impact is also identified in the Trust→CCBs relationship. These findings are explained from several theoretical perspectives, which provide directions for future research and help community management to maintain sustained participation.

KEYWORDS:

• OSS
• sustained participation
• integrative model of trust
• social learning system
• diffusion of innovation
• intellectual capital creation
• stage model
• vector autoregressive model

Acknowledgments

The first author appreciates the sabbatical leave during the Fall of 2018 awarded by the University of Massachusetts at Dartmouth to concentrate on the research. The second author’s work on this project is partially supported by the Social Sciences and Humanities Youth Foundation of the Ministry of Education in China (18YJC630150). All opinions, findings, and conclusions discussed in this paper are those of the authors and do not necessarily reflect the views of the Sciences and Humanities Youth Foundation of the Ministry of Education in China.

Notes

¹ http://www.chuckboecking.com/idempiere-review, January 23, 2021.

² Cognitive trust is more easily to be assessed by researchers. While the non-inclusion of affective trust is a limitation, it will not impact hypothesis testing.

³ https://groups.google.com/g/idempiere, Nov. 2018.

⁴ Besides two coauthors, a colleague helped with thread analysis. Since 6 threads are assessed by only one or two raters, 163 threads are used when calculating the Kappa scores.

⁵ Sentiments are generated using Amazon Comprehend which will be discussed soon.

⁶ www.bruceclay.com/blog/sentiment-trust-signal/, July 17, 2020.

⁷ aws.amazon.com/comprehend/features/. May 26, 2021.

⁸ github.com/icy/google-group-crawler, January 11, 2020.

⁹ Messages in those threads with only one participating member and threads written with a language other than English are not included for the study.

¹⁰ There are not enough data for more than 8 weeks per period.

¹¹ If the duration is not set as 6 weeks, either VAR model does not fit the data or hypotheses are not supported, inconsistent with the viable iDempiere community.

¹² www.r-bloggers.com/is-my-time-series-additive-or-multiplicative/, December 30, 2019.

¹³ For lnTrust, the DF statistic is −4.0707 and the P-value is 0.01; for lnCCB, the DF statistic is −3.4397 and the P-value: 0.01527.

¹⁴ Chi-squared = 45.614, df = 48, p-value = 0.5711.

¹⁵ Chi-squared = 58.836, df = 45, and p-value = 0.08079.

¹⁶ 0.8096, 0.8096, 0.7606, 0.7606, 0.7285, 0.7285, 0.7009, and 0.7009.

¹⁷ Chi-squared = 20.398, df = 1, p-value = 6.289e-06.

¹⁸ F-Test = 5.5196, df1 = 4, df2 = 98, p-value = 0.000475.

¹⁹ F-Test = 2.4302, df1 = 4, df2 = 98, p-value = 0.05267.

Additional information

Funding

The second author’s work on this project was partially supported by the Social Sciences and Humanities Youth Foundation of Ministry of Education in China [18YJC630150].