More than Words? An Analysis of the Sentiment Polarity of Anatomy Lectures in Medical and Allied Health Education

Abstract

Previous research suggests that medical and allied health education can erode student empathy. The decline in student empathy seems to correlate with the “hidden curriculum” ‐ the attitudes and values conveyed, most often in an implicit, informal, and unintentional fashion, via the formal educational structures, practices, and culture of an educational institution. We typically think of education as simply the transmission of facts, but the words and images used to convey those facts can also shape attitudes, even if this impact is unintended. Our study examines variation in the sentiment polarity of lectures given within the anatomy curriculum at one institution. We tested the following hypotheses: 1) the sentiment polarity of first‐year anatomy lectures is significantly different from neutral and 2) the sentiment polarity of lectures is correlated with the percentage of total lecture time spent on clinical applications. Anatomy lectures (n = 120) recorded during the 2016–17 academic year across all programs (excluding veterinary medicine) at Midwestern University were anonymized and transcribed by work study students. We analyzed transcribed lectures using text mining and a published sentiment lexicon. We counted individual words that matched the lexicon with a positive or negative sentiment, in addition to total word count, and individual word frequency. Standard anatomical descriptors (i.e., “inferior”) were excluded from the analysis. The net sentiment score for each lecture was defined as the difference between the number of positive vs negative words. We quantified the clinical content of a lecture as the proportion of time (minutes) spent on clinical applications. We used a one‐sample Wilcox test to evaluate Hypothesis 1 and Spearman rank correlation to evaluate Hypothesis 2. We used the text mining package tidytext and ran all statistical tests in R. The median sentiment score across all lectures was −18.5, which was significantly different from zero (p < .001). Sentiment score was negatively correlated with the proportion of lecture time spent on clinical applications (rho = −0.316, p < .001). These results suggest that first‐year anatomy students might be at risk for empathy erosion especially in the context of clinical applications because of the relatively high frequency of words with negative sentiment. This study was approved by Midwestern IRB (#AZ 1111, Approved on 01/25/2018).