Table 3.

Results of the best-fit negative binomial generalised linear mixed model explaining readership engagement.

Selected model: glmer.nb (Number of comments ~ Depiction of bats in the article + Political orientation + Country + (1|Year))
AIC	BIC	logLik	Deviance	df.resid
1590.7	1631.9	−785.3	1570.7	447
Scaled residuals:
Min	1Q	Median	3Q	Max
−0.4293	−0.4083	−0.3597	−0.1728	7.5924
Random effects:
Groups name	Variance	Std.Dev
Year (Intercept)	1.911	1.382
Number of obs:	457	groups: Year	29
Fixed effects
	Estimate	SE	z value	Pr( >|z|)
(Intercept)	1.41	0.733	1.925	0.054
Framing of the article: threat vs non-threat	1.209	0.306	3.950	< 0.0001
Political orientation: right-leaning vs left-leaning	−1.618	0.414	−3.911	< 0.0001
Political orientation: tabloid vs left-leaning	−1.335	0.399	−3.344	0.0008
Country: Germany vs France	−1.162	0.486	−2.391	0.0168
Country: Italy vs France	−1.329	0.449	−2.959	0.003
Country: Spain vs France	−0.138	0.740	−0.187	0.852
Country: UK vs France	−0.087	0.531	−0.163	0.870

Significant results with p < 0.01 are bolded.