Fig 1.
Overview of research design and procedures.
Table 1.
The distribution of professors and share of women by field.
Fig 2.
Mean possibilities of comment sentences pertaining to topic dimensions by genders.
(A) In five-star reviews. (B) In one-star reviews.
Table 2.
Dimensions and coverage of topics.
Fig 3.
Professors’ percentage across the numbers of received reviews by professor gender and field.
(A) In five-star reviews. (B) In one-star reviews.
Table 3.
Average word counts of comments per review by field and gender.
Fig 4.
Rating distributions and means of by professor gender and field.
Horizontal Lines in violins denote the mean values.
Table 4.
Distributions of five- and one-star review numbers and percentages by professor gender and field.
Table 5.
Regression results for ratings by field.
Fig 5.
Dunning score and popularity of the top 30 gender-distinct adjectives in RMP comments by gender and rating.
The popularity of a word is the percentage of professors whose comments contain the word. (A) In five-star reviews for women. (B) In five-star reviews for men. (C) In one-star reviews for women. (D) In one-star reviews for men.
Fig 6.
Dunning score and popularity of the top 30 gender-distinct non-adjectives in RMP comments by gender and rating.
The popularity of a word is the percentage of professors whose comments contain the word. (A) In five-star reviews for women. (B) In five-star reviews for men. (C) In one-star reviews for women. (D) In one-star reviews for men.
Fig 7.
Regression analysis of predicted possibilities from topic modeling in (A) five-star and (B) one-star reviews by professor-level topic dimensions. An orange mark denotes a positive coefficient of gender (1 = women), meaning that the topic is more likely to appear in comments for women professors. A blue mark denotes a negative coefficient of gender, indicating that the topic is more likely to appear in comments for men professors. Only statistically significant results are shown. prof. = professor; instr. = instructor.
Fig 8.
Regression analysis of predicted possibilities from topic modeling in (A) five-star and (B) one-star reviews by course-level topic dimensions. Subject dimension is excluded. Figure encodings have the same meanings as Fig 7. Only statistically significant results are shown. ppt. = powerpoint; hw. = homework; attd. = attendance.
Fig 9.
Regression analysis of predicted sentiment scores by professor-level topic dimensions.
Orange and blue marks denote positive and negative coefficients of gender (1 = women), respectively. Only gender-significant topics where gender is a significant contributing factor in the regression analysis are shown. prof. = professor; instr. = instructor. (A) Regression analysis of five-star reviews based on positive sentiment scores. (B) Regression analysis of one-star reviews based on negative sentiment scores.
Fig 10.
Regression analysis of predicted sentiment scores by course-level topic dimensions.
Figure encodings have the same meanings as Fig 9. Only gender-significant topics are shown. Subject dimension is excluded. hw. = homework; attd. = attendance. (A) Regression analysis of five-star reviews based on positive sentiment scores. (B) Regression analysis of one-star reviews based on negative sentiment scores.