Skip to main content
. 2022 Nov 22;6(2):e40764. doi: 10.2196/40764

Table 4.

Latent Dirichlet allocation topic modeling for all comments, showing comment distribution, Facebook metrics, and sentiments from Linguistic Inquiry and Word Count (LIWC; N=51,835).

Topic number Topic name Facebook metricsa—comments, n (%) Sentiments from LIWCb



Word count, mean (SD) Positive emotion percentage, mean (SD) Negative emotion percentage, mean (SD)
1 Sharing personal experiences 14,000 (27) 33.72 (46.15) 4.29 (11.34) 3.41 (7.40)
2 Survivor stories 7026 (13.6) 18.95 (22.61) 5.93 (13.98) 2.18 (4.49)
3 Risk-reduction discussion 7080 (13.7) 16.39 (25.26) 4.94 (10.93) 3.39 (6.24)
4 Religious content 11,254 (21.7) 10.03 (13.90) 23.44 (22.05) 0.91 (5.12)
5 Asking medical questions 5964 (11.5) 9.33 (9.21) 4.87 (15.11) 6.93 (9.37)
6 Sharing appreciation and information 6511 (12.6) 5.71 (10.08) 23.87 (26.99) 0.89 (4.11)

aData collected in November 2021.

bPositive and negative emotions represent the percentage of words in a post that appear in the dictionary indicating positive and negative emotions.