Unveiling Bias: A Systematic Review Of Computational Communication Literature - Info and Reading Options

The Internet Archive:

Bias is inescapable in the social world. Any one person has only a singular experience of life, and that experience becomes the lens through which they seek out, interact with, and interpret the world around them. Too often, bias grows insidious and malicious; enabling systems of oppression and reifying false evaluations as natural truth (Crenshaw, 2006; Jackson, Jr. & Weidman, 2005). Given technology’s capacity to amplify bias (boyd & Crawford, 2012; Buolamwini & Gebru, 2018), we conduct a systematic review of communication literature in computational social science (CSS). Informed by this review, our primary contribution is the development of a conceptual framework to help researchers interrogate and mitigate bias across key stages of the research process. We argue that mitigating bias in CSS faces at least two interconnected challenges. The first is that science itself is social, and researchers are subject to the same biases we aim to mitigate. Each researcher is rooted in a paradigm that influences their ontological, epistemological, and axiological perspective, ultimately shaping what questions they ask, what data they use, and how they analyze and interpret their findings (Shugars, 2024a; Webb Williams, 2024). This may result in bias in what populations or issues are studied, such as the ongoing disparity of research tending to examine Western, Educated, Industrial, Rich, Democratic (WEIRD) societies (Henrich et al., 2010) or the reliance on Twitter as a model organism of social platform research (Tufekci, 2014). The impacts of such bias become even more insidious when coupled with the second challenge: empirical approaches—and computational methods in particular—may serve to reinforce and enhance existing biases (Buolamwini & Gebru, 2018; D’Ignazio & Klein, 2020; Shugars, 2024b). Indeed, modern statistics is built largely on the shoulders of eugenicists, whose practices of cranial measurements and IQ tests provided a numerical veneer of objectivity to their racist ideology (Jackson, Jr. & Weidman, 2005). Critical scholars have long warned of the danger in interpreting subjective measurements as objective reality, arguing that our empiricist society is too quick to conflate numerical interpretations as natural truth. Unfortunately, there is reason to believe computational methods have only exacerbated this concern. Computational approaches often rely on large datasets of uncertain provenance, such as digital traces of online behavior (Lazer et al., 2021; Lazer & Ognyanova, 2024). This data is typically passively generated as people use digital services to navigate their daily lives. While such ready-made data can provide valuable insights, researchers rarely have access to full knowledge of how such datasets were constructed or what biases they may reflect (Lazer & Ognyanova, 2024; Salganik, 2019). Furthermore, the scale of such data typically requires standardization and reliance on algorithms, both of which can perpetuate existing inequalities. For example, the common preprocessing step of restricting a corpus to English-language text may disproportionately remove African American Vernacular English or other forms of speech deemed non-standard (Blodgett et al., 2016). Simultaneously, the use of black-box algorithms makes it challenging to determine whether and how bias is reflected in the resulting model (Shugars, 2024a). Finally, algorithmic output is reliant on human interpretation, which again raises the specter of human bias cloaked in the mantle of objective fact. Algorithmic-informed judgments are used in decisions about bail, insurance, and other life-changing matters, often with little thought to the bias baked in at every step of the process. References Blodgett, S. L., Green, L., & O’Connor, B. (2016). Demographic dialectal variation in social media: A case study of African-American English. In J. Su, K. Duh, & X. Carreras (Eds.), Proceedings of the 2016 Conference on Empirical Methods in Natural Language Processing (pp. 1119–1130). Association for Computational Linguistics. https://doi.org/10.18653/v1/D16-1120 boyd, d., & Crawford, K. (2012). Critical questions for big data: Provocations for a cultural, technological, and scholarly phenomenon. Information, Communication & Society, 15(5), 662–679. https://doi.org/10.1080/1369118X.2012.678878 Buolamwini, J., & Gebru, T. (2018). Gender shades: Intersectional accuracy disparities in commercial gender classification. Proceedings of the 1st Conference on Fairness, Accountability and Transparency, 77–91. https://proceedings.mlr.press/v81/buolamwini18a.html Crenshaw, K. W. (2006). Race, reform, and retrenchment: Transformation and legitimation in antidiscrimination law. In Law and social movements. Routledge. D’Ignazio, C., & Klein, L. F. (2020). Data feminism. MIT Press. Henrich, J., Heine, S. J., & Norenzayan, A. (2010). The weirdest people in the world? Behavioral and Brain Sciences, 33(2–3), 61–83. https://doi.org/10.1017/S0140525X0999152X Jackson, Jr., J. P., & Weidman, N. M. (2005). The origins of scientific racism. The Journal of Blacks in Higher Education, 50(1), 66–79. https://www.jstor.org/stable/25073379 Lazer, D., Hargittai, E., Freelon, D., Gonzalez-Bailon, S., Munger, K., Ognyanova, K., & Radford, J. (2021). Meaningful measures of human society in the twenty-first century. Nature, 595, 189–196. https://doi.org/10.1038/s41586-021-03660-7 Lazer, D., & Ognyanova, K. (2024). The future of computational social science. In J. M. Box-Steffensmeier, D. P. Christenson, & V. Sinclair-Chapman (Eds.), Oxford handbook of engaged methodological pluralism in political science (pp. 1–22). Oxford University Press. https://doi.org/10.1093/oxfordhb/9780192868282.013.50 Salganik, M. J. (2019). Bit by bit: Social research in the digital age. Princeton University Press. Shugars, S. (2024a). A matter of perspective: Computational social science and researcher choice. In J. M. Box-Steffensmeier, D. P. Christenson, & V. Sinclair-Chapman (Eds.), Oxford handbook of engaged methodological pluralism in political science (vol 1) (p. 0). Oxford University Press. https://doi.org/10.1093/oxfordhb Shugars, S. (2024b). Critical computational social science. EPJ Data Science, 13(1), 13. https://doi.org/10.1140/epjds/s13688-023-00433-2 Tufekci, Z. (2014, April 15). Big questions for social media big data: Representativeness, validity and other methodological pitfalls. Proceedings of the 8th International AAAI Conference on Weblogs and Social Media. ICWSM, Ann Arbor, MI. https://doi.org/10.48550/arXiv.1403.7400 Webb Williams, N. (2024). What type of data are images? In J. M. Box-Steffensmeier, D. P. Christenson, & V. Sinclair-Chapman (Eds.), Oxford handbook of engaged methodological pluralism in political science (vol 1) (p. 0). Oxford University Press. https://doi.org/10.1093/oxfordhb/9780192868282.013.47

Search for “Unveiling Bias: A Systematic Review Of Computational Communication Literature” downloads:

Visit our Downloads Search page to see if downloads are available.

Find “Unveiling Bias: A Systematic Review Of Computational Communication Literature” in Libraries Near You:

Read or borrow “Unveiling Bias: A Systematic Review Of Computational Communication Literature” from your local library.

• The WorldCat Libraries Catalog: Find a copy of “Unveiling Bias: A Systematic Review Of Computational Communication Literature” at a library near you.