- Note: Please see the referenced paper, The Sharing Economy in Computing: A Systematic Literature Review. In Proceedings of the 21st ACM conference on Computer supported cooperative work & social computing. ACM. by authors: Tawanna R Dillahunt, Xinyi Wang, Earnest Wheeler, Hao Fei Cheng, Brent Hecht, and Haiyi Zhu. 2018. This paper contains the full bibliography of the works cited in the description of the method below as well as a more complete description of the method in Appendix A.
We conducted a systematic literature review (SLR) of sharing economy research in the ACM Digital Library (DL).
Identifying Search Terms / Corpus Collection
The first step of any SLR is to identify relevant work and to achieve this, we identified our search terms. The sharing economy has been referred to by many terms. We sought to build a comprehensive set of keywords to include terms for the sharing economy that may have been used in older literature. To accomplish this, we conducted an iterative search of repeated, relevant keywords in our corpus. First, we identified search terms based on common terms that have been used in HCI and in well-known and cited monographs of the sharing economy literature [Botsman and Rogers 2010; Sundararajan 2016]. The initial search terms were: “sharing economy,” “collaborative consumption,” “peer-to-peer exchange,” “physical crowdsourcing,” and “gig economy.”
Next, we extracted the author-defined keywords from the corpus resulting from our first search of the digital ACM library. We compiled a list of all keywords that appeared in two or more papers, and then filtered those keywords that were overly vague (e.g., multi-agent systems, transportation) or conceptually unrelated to the sharing economy (e.g., dynamic pricing, experimentation). We considered the remaining repeated keywords to be synonymous or closely related to the sharing economy, and performed a search with that list. We performed this process three times, until the search yielded no new papers, for a total of four searches. Ultimately, we identified the following 21 keywords, which served as seeds for our final ACM search: “sharing economy,” “collaborative consumption,” “peer-to-peer exchange,” “physical
crowdsourcing,” “gig economy,” “algorithmic management,” “collaborative economy,” “local online exchange,” “mobile crowdsourcing,” “network hospitality,” “on-the-go crowdsourcing,” “platform economy,” “ridesharing,” “social exchange,” “surge pricing,” “timebanking,” “micro tasking,” “microtasking,” “situated crowdsourcing,” “workplace studies,” and “spatial crowdsourcing.” This search procedure yielded 354 papers. Next, we filtered articles using specific selection criteria to assure their
appropriateness, or fit to the SLR, as described next.
Our first selection criterion was that articles had to be written in English owing to the constraints of the research team and the general tendency for high-quality research in computing to be published in English (for better or worse). Second, we filtered articles to exclude research that was not substantively connected to the sharing economy (despite its use of one of the keywords) or research that fell below a quality threshold. To implement this second stage, we utilized a version of the approach by Busalim and Che Hussin [Busalim and Hussin 2016]. Specifically, researchers read each paper and ranked each article as high, medium or low on the following four criteria: (1) the paper’s topic was related to the sharing economy, (2) the paper had a clear research methodology, (3) the paper described the data collection process, and (4) the
paper had key findings that aligned with the posed research questions. Those criteria that ranked as high received a score (loading factor) of 2, medium 1, and low 0 [Nidhra et al. 2013]. Research papers that received a 0 for the first criterion (papers unrelated to the sharing economy) were immediately excluded. Any paper that received a 1 or a 2 on the first criterion and had a total score of 5 passed through the filter; if the total score was 4 or below, the paper was excluded. It is important to point that out we did not exclude short papers, extended workshop papers, panels, or alt.chi paper purely due to their format; they were processed like full papers. After the second stage of the filtering process, we were left with a corpus containing 119 sharing economy articles. Finally, seven papers were marked as “Accidentally made it through filters.” These papers were irrelevant to the sharing economy and included magazine articles that we identified as think pieces (see Appendix
A.2.2 Selection Criteria of the cited paper for details). This left us with a final corpus of 112 papers.
We extracted excerpts from each work to provide: (1) the stated or implied problem or research question that was being addressed, (2) the central purpose or focus of the study, (3) a brief statement about the sample population or subjects, and (4) key results that related to the proposed study. In addition, we extracted the list of authors,
publication year, title, publication venue, geographic region, the research questions, and how each paper defined the sharing economy. We also extracted keywords and coded the methodology employed as quantitative, qualitative, mixed methods, design, math modeling, or literature review. In some cases, two methodology codes were applied. For example, a paper that presented the design and field deployment of a new sharing economy application, and reported results from a qualitative user study, it would be coded as both “design” and “qualitative.”
To identify topical themes in the literature, three members of our research team coded each paper in the final corpus. This was beneficial for addressing both research questions. We used the selection criteria to extract and summarize article contents as attribute codes [Onwuegbuzie 2016; Saldaña 2012] into a shared spreadsheet. This allowed us to develop a synopsis for each work and to sort the literature according to publication year, methodology used, populations sampled, instances or applications of the sharing economy studied (e.g., Airbnb, Lyft), and by publication venue. More specifically, we used hybrid coding to assign topical codes to each paper [Trochim 2006]. We leveraged existing frameworks [Owyang 2016] to create a priori codes. We also coded openly in our initial coding phase and followed this with focused coding. Focused coding allowed us to categorize significant or frequent codes that emerged from our data [Saldaña 2012].