Language Bias in Health Research: External Factors That Influence Latent Language Patterns

Abstract

Background: Concerns with problematic research are primarily attributed to statistics and methods used to support data. Language, as an extended component of problematic research in published work, is rarely given the same attention despite language's equally important role in shaping the discussion and framings of presented data. Purpose: This study uses a topic modeling approach to study language as a predictor of potential bias among collected publication histories of several health research areas. Methods: We applied Latent Dirichlet Allocation (LDA) topic models to dissect publication histories disaggregated by three factors commonly cited as language influencers: (1) time, to study ADHD pharmacotherapy; (2) funding source, to study sugar consumption; and (3) nation of origin, to study Pediatric Highly-Active Anti-Retroviral Therapy (P-HAART). Results: We found that, for each factor, there were notable differences in language among each corpus when disaggregated by each factor. For time, article content changed to reflect new trends and research practices for the commonly prescribed ADHD medication, Ritalin. For funding source, industry and federally funded studies had differing foci, despite testing the same hypothesis. For nation of origin, regulatory structures between the United States and Europe seemingly influenced the direction of research. Conclusion: This work presents two contributions to ethics research: (1) language and language framing should be studied as carefully as numeric data among studies of rigor, reproducibility, and transparency; and (2) the scientific community should continue to apply topic models as mediums to answer hypothesis-driven research questions.

Identifying accurate pro‐choice and pro‐life identity labels in Spanish: Social media insights and implications for comparative survey research

“…When that text is run through a text consolidation or topic‐modeling algorithm, a computer can eliminate information until only core themes remain. Topic models have been widely applied across social science disciplines, including tracking COVID‐19 discourse online, 28 comparing religious texts, 32 mapping product marketing, 33 and predicting research article trajectories 25 . For this study, we used the Latent Dirichlet Allocation (LDA) algorithm, the most commonly used and validated topic modeling algorithm in the literature 34 .…”

“…Topic models have been widely applied across social science disciplines, including tracking COVID‐19 discourse online, 28 comparing religious texts, 32 mapping product marketing, 33 and predicting research article trajectories. 25 For this study, we used the Latent Dirichlet Allocation (LDA) algorithm, the most commonly used and validated topic modeling algorithm in the literature. 34 See Figure 1 for a conceptual explanation of topic modeling.…”

“…Significant pluralities of the global population regularly use at least one social media platform, 23 with a high representation of Spanish speakers across platforms. Advances in big data and Natural Language Processing (NLP) methods have also made it possible to extract these data to draw deeper inferences about various phenomena, 24 , 25 , 26 including abortion views disclosed online. 27 Thus, these tools could be leveraged to identify Spanish‐language abortion identity labels, their uses, and which label(s) accurately convey a similar meaning to the English‐language terms pro‐choice and pro‐life.…”

Neutral or Framed? A Sentiment Analysis of 2019 Abortion Laws

“…As with other NLP methodologies, which include topic modeling (Blei et al, 2003) and machine learning classification (Kotsiantis, 2007), the theoretical suppositions of this class of methods argue that words can, themselves, be treated and studied as individual datapoints. When analyzed via algorithms, these words can be further leveraged to draw inferences about the collection of words, themselves, or generalized outward to other human phenomena (Valdez & Goodson, 2020).…”

On Mining Words: The Utility of Topic Models in Health Education Research and Practice

“…While the authors found three primary topics across all inventories tested, they found that one (Experience of Transcendence) was relatively rare in early inventories but increasingly common over time. In a public health context, Valdez and Goodson (2020) archived the history of research surrounding the medication Ritalin, from 1970 to 2016. Briefly, their analysis found that early articles related to the prescription of the drug to treat attention-deficit/hyperactivity disorder in boys.…”