Knowledge production and sharing in crowd-based online communities is an increasingly significant phenomenon (e.g., Huang et al. 2014, Ren et al. 2012, 2016). Finding out what motivates contributions to such crowd-based knowledge production and dissemination is key for enhancing innovation and knowledge flows and hence social welfare. We study career concern as a motivational driver for solving problems of others and developing public stocks of knowledge in online collaboration communities (OCCs). Our empirical context is Stack Overflow, an online platform founded in 2008, in which community members post questions and answers on computer programming-related issues for various programming languages. We build on the related phenomenon of open source software (OSS) development (von Hippel and von Krogh 2003, von Krogh et al. 2012) and presume that community members try signaling their competencies by answering questions of others. By linking this behavior to changes in demand for software developers in the external job market (cf. Boudreau 2012), we examine whether career concern influences contribution behavior. If career concern motivates community members, we expect them to actively respond to the job market demand for specific skills by adapting their contribution level.
Our empirical strategy is to utilize individual-level contribution data from Stack Overflow along with programming language-level data from the job market for software developers. Stack Overflow implements sophisticated reputation mechanisms to encourage and quantify member contributions. We exploit the unexpected launch of an internal job market—Stack Overflow Careers—as a natural experiment to draw causal inference on the relationship between career concern and member contribution. Stack Overflow Careers, launched in November 2009 and upgraded in February 2011, is a for-fee-service within the community that allows members to post their CVs, which potential employers can browse to identify job candidates using customized search criteria. Because members’ contributions in the community become part of their CVs, members can signal their programming skills and overall competencies directly to the recruiters. To capture the demand-side conditions of the job market, we obtain data from IT Jobs Watch, an Internet platform that tracks IT-related job advertisements in UK. The data contain the quarterly number of vacancies and wage offers for each of the programming languages. The combination of both data sources represents a unique dataset that longitudinally traces the contribution behavior in an OCC contingent on job market conditions, allowing us to study community members’ career motivation as a driver of their contribution behavior. Furthermore, the fine-grained language-level data on member contribution and job market conditions allow us to examine the heterogeneity in the effect of career concern by the degree of language specialization.
We find a causal relationship between the demand for software development jobs and the magnitude of member contribution in the corresponding programming language. Specifically, our analysis shows (1) that the number of external job vacancies in a programming language positively influences the number of contributing members and the total number of contributions; (2) the increase of total contribution comes primarily from new members contributing in that language; (3) enhanced information flows following the introduction of Stack Overflow Careers amplify the effect of job vacancies on contribution intensity; and (4) members with greater language specialization react more sensitively to demand conditions. We also find weak evidence of a positive interaction effect between information flows and language specialization on the job vacancy-contribution intensity association.
We claim three contributions. First, we contribute to the emerging literature on OCCs (Faraj et al. 2011, Hew and Hara 2007, Majchrzak et al. 2013, Ren et al. 2016), such as question and answer communities (e.g., Answers, Yahoo! Answers, and Baidu Knows), and online communities of practice, such as healthcare (Hara and Hew 2007) and law (Wasko and Faraj 2005). We show that career concern could be a significant motivator of contributions in such communities.
Second, we extend the motivational literature on OSS development (Lerner and Tirole 2002, von Hippel and von Krogh 2003, von Krogh and von Hippel 2006). Although our setting differs from traditional OSS projects, which are of much larger size with interdependent collaborative activities (e.g., Linux (Bagozzi and Dholakia 2006, Lee and Cole 2003), Apache (Roberts et al. 2006), and Mozilla (MacCormack et al. 2006)), we believe our findings are readily extendable to OSS communities. Compared to OSS development, our setting is less hospitable for pursuing career motives because contributors can signal only a very limited spectrum of their skills by answering isolated questions. Thus, our finding of support for career concern as an effective contribution motive in this conservative setting is striking. Moreover, our utilization of large-scale archival data provides important improvement over the OSS literature, which has mainly relied on self-reported data (Hertel et al. 2003, Lakhani and Wolf 2005, Orman 2008, Roberts et al. 2006, Spaeth et al. 2015) and anecdotal evidence (Raymond 1999, Himanen 2001). This helps reinforce our understanding of the motivations for contributing to OSS development (e.g., von Krogh et al. 2012) and supports the theory of private-collective innovation (von Hippel and von Krogh 2003).
Third, we conjecture a paradigm change regarding the role of institutions of higher education. These institutions have traditionally functioned to improve career prospects by providing quality signals for their clients. However, OCCs also seem to perform this function, potentially replacing these institutions’ role as signal providers, at least in the area our study examines. We argue that such a paradigm change has positive consequences for social welfare. In the institutions of higher education, the immediate outcomes of the signaling production process are wasted (Arrow 1973, Holmström 1999, Spence 1973). In OCCs, however, outcomes themselves serve an immediate purpose. The exchanged knowledge has at least one (often many more) recipient who profits from the outcomes. Since the recipients will approve (via voting) only if they see value in the contribution, the generated signal is directly tied to recipients’ benefit. Because of this coupling, social waste from signal production may be much lower in OCCs than that in the traditional institutions of higher education.
REFERENCES AVAILABLE FROM AUTHORS
HBS_OUI_Submission_Short_Final.pdf 