Humanitarian GIS- Uncertainty & Ethics
5/2/21
VGI, or volunteered geographic information, includes sources like Twitter, OpenStreetMap, and iNaturalist. While they can be great sources of data, there can be a lot of uncertainty given the unstructured and unregulated nature of data creation. Data volume is inherently biased toward who has volunteered the data. This can lead to an incomplete picture of reality. For example, during disasters, Tweets are not just by victims of its effects but also by first responders, news sources, and others hearing the news and offering condolences and other information. Moreover, with disasters for example, it can be difficult to quantify when it started and ended given vulnerability and social context are so connected to disaster damage. Bots have also become common in social media as well. All of these complexities in Twitter data make it difficult to extrapolate from boiled-down, blanket-analyzed data. The same can be said for VGI data in general, as data should be interpreted differently based on who is uploading the data and why. For example, my Insect Biology class had to enter a lot of data entries into iNaturalist throughout the semester, making it seem like Middlebury, VT has a disproportionate number of insects. It would be important to consider that most posts of insects in Middlebury came from one particular demographic (college students) to prevent inaccurate extrapolation of the data.
Because VGI data is based on individuals uploading data, it can be difficult to ensure everyone is aware of how the data will be used. This is particularly important for personal or sensitive information, which comes up a lot on Twitter, particularly during disasters when help is most needed. However, if data cannot be analyzed due to privacy concerns, there would be increased uncertainty in the data sample as there is bias toward data without privacy concerns. Because privacy concerns are often connected to important information about how disasters are affecting an individual, excluding this data from a study may mean missing important variables and context. It is therefore important to make individuals aware of how their data may be used, and if individuals had published data in situations where privacy was less important than their wellbeing at the moment, researchers should be extra cautious by asking for permission again and providing an opportunity for individuals to remove their data from the study or ensure anonymity. I think ethics will always come back to individuals’ relationships with one another. If researchers have decent social skills, are aware of power dynamics, and are respectful of those they are studying, and use all of these to loop “subjects” into conversation about their studies, I think ethics would become clearer and more navigable. The impersonality of data collection and use, while important sometimes, can also lead to ethical violations given such a separation from reality.
Sources:
Crawford, K., and M. Finn. 2014. The limits of crisis data: analytical and ethical challenges of using social and mobile data to understand disasters. GeoJournal 80 (4):491–502. DOI:10.1007/s10708-014-9597-z