Dynamics of the social construction of knowledge: an empirical study of Zhihu in China

EPJ Data Science

Table 1 HLM regression for information accumulation

	Model 1		Model 2
	Estimates	Errors	Estimates	Errors
Fixed Effects
Intercept	0.28	0.002	0.46	0.01
Marketplace of Ideas
Gini-Answers Length			−0.11^∗∗∗	0.01
Gini-Comments			−0.11^∗∗∗	0.01
Participation Features
Presence of Active Users^a
Highly Active Users			−0.01^∗∗∗	0.003
Active Users			−0.01^∗∗∗	0.002
Anonymous			0.001	0.001
Temporal Features
Time-Interval between Answers^b			−0.01^∗∗∗	0.004
Order of Answers			−1.44^∗∗∗	0.08
Discourse Features
Sentiment			0.0002	0.001
Readability^c			−0.01^∗∗	0.003
Control Variables
Question Type: Opinion^d,e			0.003	0.004
Question Type: Fact^d,e			−0.04^∗∗∗	0.01
Number of Answers per question^b,d			−0.004^∗	0.002
Random Effects
Intercept T₀₀	0.07	0.16	0.07	0.15
Model Fit Statistics	37,451		39,705

Notes. *** p<0.001; ** p<0.01; * p<0.05
^aHighly active users are users whose answering frequencies rank in the top 1%. Active users are users whose answering frequencies rank in top 2%–10%. Ordinary users are users whose answering frequencies rank in bottom 90%. The comparison group is ordinary users.
^bVariables are log transformed.
^cReadability is an inverse measure. The higher the readability score of a text is, the higher the complexity of the text is.
^dVariables at level 2 (i.e., the question level).
^eThe comparison group is Experience-based questions.