. 2023 Jan 10;10:1045279. doi: 10.3389/fpubh.2022.1045279

Table 4.

Selected features for modeling.

	L

	SCLIWC	Weibo-5BML	D	B	Total
NFA total	Qmark (r = −0.11)**, money, dash (r = −0.07), death (r = −0.08)**, apostrophe, ingest, inclusive, religion, achieve, we, certain, sexual, love, multiFun	NA	NA	Self-description length, bi-followers/friends (r = −0.07), general emoticons-total words ratio(r = 0.10)*, posts posted at night/total posts, posts contain URL/total posts(r = 0.08)	19
NFA approach	PPron (r = 0.12)*, Qmark (r = −0.12), I (r = 0.12), sad, work, social (r = 0.12), see(r = 0.10), adverb (r = 0.10), percept (r = 0.10), apostrophe, ingest (r = 0.07), pronoun (r = 0.10), friend, time (r = 0.07), achieve	happiness	Age(r = −0.07)*	Self-description length, posts about 50–140 words/total posts	19
NFA avoidance	Religion, friend, death (r = 0.09)**, dash, exclusive, SpecArt, sexual, ingest, certain, period, Qmark, achieve	NA	NA	General emoticons-total words ratio(r = −0.06), bi-followers/followers, bi-followers/friends (r = 0.08)	15

L, linguistic features; D, demographic features; B, behavioral features; NA means no features in this class were selected; *p < 0.05, **p < 0.01, ***p < 0.001(the Pearson correlation coefficient between selected features and questionnaire scores); Total, the total number of selected features for each dimension.