Table 2.

Odds ratios obtained by logistic regression.

Variable name	Comm.	Conf. 1	Conf. 2	Conf. 1+2
Clinton sup.	0.942***	0.918***	0.936***	0.911***
Cross-group	1.195***	1.172***	1.191***	1.165***
Clinton sup., cross-group	1.064***	1.091***	1.070***	1.102***
Avg. score (author)		1.166***		1.174***
Avg. score (target)		1.151***		1.167***
Diff. avg. score		1.213***		1.228***
Diff. frac. positive			0.497***	0.498***
Frac. positive (author)			1.260***	1.247***
Frac. positive (target)			1.221***	1.195***

Each column corresponds to a model with a specific set of variables: we have community features in the first column; then, we add to the model the first, the second, and both sets of confounding variables. All the variables shown here have a statistically significant ($p<0.001$) impact on the likelihood of writing a comment.

* $p<0.05$, *** $p<0.001$.