October 11th, 2011
In mid-July I submitted the final version of my thesis to the library, and am now entitled to call myself Dr. Chawner. I don’t feel any different, though, and apart from the satisfaction of having completed (after far too long) a major piece of research, don’t expect my life to change. I’ve sent the abstract of my research to the people I interviewed, and to the discussion lists I sent invitations to complete the web-based survey to. In addition, I thought it might be a good idea to post a summary here, along with a few additional comments.
The purpose of this research was to identify factors that affect participants’ satisfaction with their experience of a free/libre open source software (FLOSS) project. The research built on existing models of user satisfaction from the information systems literature, and also incorporated two characteristics of FLOSS projects first identified by Ye, Nakakoji, Yamamoto, and Kishida (2005), product openness and process openness. The central research question it answered was, What factors influence participant satisfaction with a free/libre and open source application software project?
Richard Stallman’s reasons for setting up the GNU project and the Free Software Foundation arose from his frustration at being forced to be a passive user of software used for a Xerox printer. These suggest that being able to be an active participant in a FLOSS project is one factor that should be examined, and therefore the first sub-question this project answers is, What types of contributions do participants make to free/libre and open source software projects?
Several studies have shown that the extent of participation in a FLOSS project varies from individual to individual, and this variation leads to the second sub-question, Do the factors that influence satisfaction vary for different types of participation? If so, in what way?
A preliminary conceptual model of factors affecting participant satisfaction was developed, reflecting the key concepts identified in the literature. The main theoretical goal of this research was to test the model using empirical data.
The research used a sequential, mixed methods approach. The first, qualitative stage involved reviewing documents from selected projects and interviewing a purposive sample of FLOSS project participants. The second, quantitative stage involved an online survey of FLOSS project participants, and the data gathered were used to test the conceptual model.
The results of the first stage showed that participation in FLOSS projects was a more complex construct than previously reported in the literature. Seven categories of activities were identified:
- interaction with code;
- supporting the community;
- management; and
I illustrated the relationships between the categories in a diagram that places ‘use’ at the centre (where I think it belongs, since ’software is for use’, to paraphrase Ranganathan’s first law of library science), and groups the other six categories in three pairs. There is more detail about why I used this grouping in the full thesis. One of the reasons I was interested in identifying different types of participation, and how they relate to each other, is that I feel that the onion-style model many people use to discuss what happens in a FLOSS community (which usually places code at the centre) overemphasises the importance of coding, and diminishes the importance of the user. I don’t mean to imply that coding isn’t important, but I think that other aspects of particpation are also important.
Four attributes that modified these categories were also identified: organisational focus, role formality, remuneration, and time commitment.
Data from 154 responses to the online survey were used to test the model using stepwise multiple regression, which determined the effect of each of the variables on overall participant satisfaction. Moderated regression analysis was used to test the effects of three potential moderating variables. The results showed that that perceived system complexity had the largest effect, decreasing satisfaction if respondents perceived that the software was complex, while process openness and perceived developer communication quality accounted for the most variance in satisfaction. The final conceptual model, based on the data, is below.
Overall, the model explains 44% of the variance in satisfaction; the β values show the standardised effect of each predictor variable on satisfaction.
The main theoretical contribution of this research lies in its extension of satisfaction studies to FLOSS communities, showing that communication and openness are more important than in conventional software projects. Its practical contribution will help people involved in the management and governance of FLOSS projects to identify ways of increasing their participants’ satisfaction, which may in turn encourage them to contribute more.
The final version of the thesis is available in the VUW library’s research archive from:
Ye, Y., Nakakoji, K., Yamamoto, H., and Kishida, K. (2005). The co-evolution of systems and communities in free and open source software development. In Koch, S., editor, Free/open source software development, pages 59–82. Idea Group, Hershey, PA.
I’d welcome comments on the results here (while comments are open), or by email to me at firstname.lastname@example.org.