Are Campaigns Getting Uglier, and Who Is to Blame? Negativity, Dramatization and Populism on Facebook in the 2014 and 2019 EP Election Campaigns

Figures & data

Table 1. Negative binomial regression (2014) (β coef).

	LIKES		SHARES		COMMENTS
Party Family
Left Socialist	−.097		.222	**	.033
Conservative	.007		−.201	**	.249	***
Nationalist	1.053	***	.826	***	1.302	***
Ecologist	.132	*	.489	***	.139
Ethnic Regional	−.123	**	−.119		−.432	***
Special Issue	.282	***	.176	*	.071
Party Size
Minor	−.484	***	−.524	***	−.911	***
Fringe	−.534	***	−.466	***	−1.113	***
Post characteristic
Negative Emotions	.181	**	.563	***	.177	*
Dramatization	.202	***	.182	*	.222	**
NegativeCampaigning	.052		.121		.041
Populist Content	.450	***	.437	***	.494	***
Posts per day	−.059	***	−.081	***	−.057	***
Day of campaign	−.037	***	−.036	***	−.032	***
Community size	.638	***	.560	***	.699	***
CONSTANT	−.566	**	−1.720	***	−4.110	***
N	6100		6100		6100
dispers_s	1.1977		1.2106		1.1643
dispers_p	1.4276		1.8499		1.8107

* p < 0.10, ** p < 0.05, *** p < 0.01.

Facebook affordances are not indicated due to space limit, full model is available in Appendix.

Table 2. Negative binomial regression (2019) (β coef).

	LIKES		SHARES		COMMENTS
Party family
Left Socialist	.211	***	.691	***	.089
Conservative	.419	***	.425	***	.215	***
Nationalist	1.236	***	1.308	***	.970	***
Ecologist	−.124	**	.277	***	−.270	***
Ethnic Regional	.043		.086		.091
SpecialIssue	−.333	***	.560	***	−.900	***
Party size
Minor	−.075	*	−.315	***	−.354	***
Fringe	−.213	***	−.745	***	−.917	***
Post characteristic
Negative Emotions	.208	***	.580	***	.229	***
Dramatization	.160	***	.219	***	.204	***
Negative Campaigning	.008		.243	***	.258	***
Populist Content	−.018		.041		−.048
Posts per day	.027	***	−.001		.036	***
Day of campaign	.003		−.003		.005
Community size	.618	***	.471	***	.724	***
CONSTANT	−2.227	***	−.818	**	−4.949	***
N	8193		8193		8193
dispers_s	1.1813		1.2174		1.2170
dispers_p	1.8844		2.7822		5.0635

* p < 0.10, ** p < 0.05, *** p < 0.01.

Facebook affordances are not indicated due to space limit, full model is available in Appendix.

Figure 1. Odds ratio, logistic regression of party family effect on content character.

Figure 2. User engagement (likes, shares, comments) by party family in 2014 and 2019.

Each bubble represents a Facebook post, which is located according to its number of likes and shares. The size of the bubble reflects the number of comments – the more comments, the larger the bubble.