Figure 3 in Chen et al. 2021, Distribution of scholars and collaboration between continents
Delighted to have a new open source scoping review paper in New Media and Society led by PhD student Yan Chen, Using social media images as data in social science research. This paper is the result of one of her comprehensive exams and allows us to see some of the strengths and biases of this emerging practice. There is a distinctly English-language and small-data approach to these datasets, compared to ‘culturomics’ approaches emerging in conservation contexts, for instance, that use lots of machine learning approaches. There is also a chaos of approaches to ethics and copyright in the handling and use, characteristic of a method that has not yet matured and is subject to a constantly shifting context in terms of platforms and usage norms.
Clusters of landscape, cultural ecosystem services and demographics associated with Instagram images in Chen et al 2020
As decision-makers tackle the challenge of adapting Bay of Fundy dykelands to climate change, they need to understand who uses and values dykelands and salt marshes, and for what. This new paper in Ocean and Coastal Management, Comparing cultural ecosystem service delivery in dykelands and marshes using Instagram: A case of the Cornwallis (Jijuktu’kwejk) River, Nova Scotia, Canada, used four months of geocoded Instagram data to understand the cultural ecosystem service (CES) tradeoffs that might result from removing/realigning dykes and restoring salt marshes where dykelands can’t be sustained. Dykelands provide a much wider set of CES for a wider demographic than do marshes for this set of social media users. However, a big surprise is that while salt marshes were present in many photos they were not named as such; users spoke only about the dykes and dykelands behind those marshes. As such, the marsh CES in the dataset came from visitors to an impounded freshwater wetland trail which is a local attraction walkable from the downtown centre of Kentville. Many of the messages triangulate well with the 2016 online Q survey I ran with Nova Scotians about the same topic and the paper provides another nice case study as to the utility of social media data for social impact assessment. One of the really great things about this paper is that it is a real ‘lab’ output. The work was initiated as a follow-up to that 2016 study and to inform the new ResNet work when I knew Camille was going to be joining as an intern from AgroCampus Ouest. PhD student Yan collected a few months of Instagram posts for Camille to analyze with her help, postdoc Tuihedur helped with statistics, and then Yan picked it up again to write up after Camille went back to France. I’m proud of this paper and this collaborative team.
The second paper from Yan Chen’s MES thesis is now out in Society and Natural Resources, Leveraging social media to understand younger people’s perceptions and use of hydroelectric energy landscapes. It is a research note demonstrating the utility of manual coding and conceptual mapping of a year of Instagram images around two hydroelectricity sites to predict how changes might affect young residents. Unlike her first thesis paper in Landscape and Urban Planning, which carried out spatial mapping of value ‘hotspots’–a method widespread in today’s growing literature on cultural ecosystem services–this paper makes statistical links between features, activities and values conveyed through Instagram. The diagrams provide insight to the lifestyle and emotions associated with different landscape features, some changeable with hydro development or removal, and informs our new work on conservation culturomics for social impact assessment. Yan continues to drive this work as an IDPhD student. Congratulations, Yan.
Yan Chen presenting her IDPhD work at NSF-funded workshop in Singapore, January 28-29, 2019.
Yan Chen is wrapping up a few days in Singapore for the NSF-funded Research Coordination Network (RCN) in Science, Engineering and Education for Sustainability (SEES) on Putting Sustainability into Convergence: Connecting Data, People, and Systems. This international workshop has been diverse in attendees and disciplines. Yan reflected, “The most discussed question is how people from different disciplines can collaborate. There are many scholars like me, as social scientists who are using sophisticated data analysis models; while others are engineers working on social issues. We both, at a certain degree, struggle in ‘cultural shocks’ between different disciplines.” It’s been a great opportunity for her to workshop with similarly cross-cutting folks. She described her session as discussing, “data sources, sizes, validity, sharing, proxies, and so on. …. [agreeing] that data or method cannot develop only on the technologies, but has to answer certain questions. For social scientists, finding a good mechanism of data sharing or archiving may be very useful. Also, how to cope with the rapidly developing technologies will be another challenge for us.” Thanks to SSHRC for supporting Yan’s trip, via Mike Smit’s Insight Grant, on which I’m a CI, Assessing the social impacts of hydroelectricity-driven landscape changing using text, images and archives: a Big Data approach.
Three books from my AAG trip to New Orleans
Bookshelf serendipity strikes again! I can’t resist telling this story, though it was awhile ago. Appropriate to the geography conference I was in town to attend, I bought a compelling atlas of New Orleans by Rebeccas Solnit and Snedeker, Unfathomable City, part of a series at UC Press, at a small bookstore behind the Cabildo. Each map and essay tells an idiosyncratic story about the place, including human and river channel migrations, social and landscape erosions, including social clubs, seafood and sex. While excellent, it wasn’t an easy one to tote in my bag for idle moments.
The ‘take a book, leave a book’ shelf at my hotel filled the gap with But What if We’re Wrong, by pop seer Chuck Klosterman (sorry Columns Hotel, I’ll leave one next time). Klosterman is talking about the same thing that I was at AAG to talk about: climax thinking. He asks how we can learn to make decisions anticipating the many ways that we might be wrong, so we don’t box ourselves in. Instead we denigrate the people who made decisions or assessments we reflect upon today as folly, but assume against evidence that we’re going to be right. There is something to be said about doubt.
Having finished both of the above, I needed another book for the flight home. During a layover in Toronto airport I bought Everybody Lies by Seth Stephens-Davidowicz. He is a computer scientist who thinks data mining has replaced social science, so a few sections (particularly the conclusion) grated, but this was a fun (and surprisingly dirty) introduction to how secondary online datasets like our Google queries help us learn what people are really thinking and feeling. I have since brought up his examples in a few social science contexts, like the energy incubator at Cornell. I’m loathe to hand it to students, given some of the icky content (people are, it turns out, gross), but as survey response rates drop, offensiphobia rules, and questions around sustainability cross the sociology/psychology boundary, such datasets may well be the only way we can really understand what kind of society we are really working with.