Does digital campaigning matter, and if so, how? Testing a broadcast versus network effects model of candidates Twitter use

Figures & data

Figure 1. Method for deriving topic congruence.

Table 1. Twitter use overall and by party.

	All	Lab	Con	LD	Green	UKIP	NI & Nat*	Other
Frequency
Mean	196	209	129	217	168	284	200	252
SD	319	233	175	345	269	491	220	597
Max	4330	1614	1525	3020	3395	2833	1058	4330
Min	1	1	1	1	1	1	1	1
Congruence
Mean	0.843	0.827	0.854	0.817	0.854	0.830	0.877	0.897
SD	0.163	0.163	0.149	0.170	0.161	0.175	0.175	0.131
Max	1	1	1	1	1	1	1	1
Min	0.283	0.283	0.383	0.283	0.333	0.300	0.283	0.400
Interactivity
Mean	0.299	0.294	0.261	0.321	0.282	0.378	0.265	0.351
SD	0.241	0.222	0.26	0.225	0.215	0.297	0.241	0.256
Max	1	1	1	1	1	1	1	1
Min	0	0	0	0	0	0	0	0
Engagement**
Mean	4059	10877	1686	734.6	866.5	1308	4290	1622
SD	66130	124778	9828	4189	8338	3681	11467	7384
Max	2560002	2560002	155542	60411	122441	19678	87605	56704
Min	0	0	0	0	0	0	2	0
Online popularity
Mean	7425	13606	9273	3286	2428	2064	9105	4517
SD	29623	67333	26936	22197	17207	4296	27043	23091
Max	1257824	1257824	355811	329889	240717	31028	271531	265233
Min	3	15	14	21	8	5	13	3
Use Twitter (%)	89.0	92.5	86.2	88.9	91.0	79.0	87.3	91.1
Total N	1858	484	380	370	264	105	101	154

*Combines Sinn Fein, SDLP, UUP, Alliance Party, DUP, SNP, and Plaid Cymru.

**The Ns to calculate this variable are slightly lower (1852) than the other variables due to data on likes and retweets being collected at a different time point using the Twitter Academic API.

Figure 2. Distribution of issue candidate-constituency issue congruence scores.

Table 2. OLS regression predicting number of Candidate votes.

	Model 1			Model 2			Model 3
	Estimate	S.E.	p	Estimate	S.E.	p	Estimate	S.E.	p
Candidate Twitter use
Frequency	−0.035	0.010	0.000	−0.075	0.012	0.000	0.140	0.159	0.379
Interactivity	0.017	0.013	0.183	0.019	0.013	0.148	0.015	0.028	0.585
Issue congruence	0.011	0.013	0.371	0.009	0.012	0.455	−0.026	0.035	0.452
Twitter Engagement (ref. cat. = 1st decile − 0 to 10)
2nd decile − 11 to 34				0.094	0.053	0.074	0.120	0.122	0.328
3rd decile − 35 to 68				0.081	0.052	0.118	−0.034	0.098	0.731
4th decile − 69 to 125				0.169	0.052	0.001	0.076	0.099	0.444
5th decile − 126 to 208				0.140	0.053	0.009	0.027	0.099	0.786
6th decile − 209 to 337				0.146	0.055	0.008	0.054	0.100	0.589
7th decile − 338 to 514				0.144	0.057	0.012	0.044	0.101	0.661
8th decile − 515 to 1,039				0.296	0.060	0.000	0.198	0.104	0.056
9th decile − 1,040 to 3,092				0.353	0.064	0.000	0.333	0.109	0.002
10th decile − 3,093 to 2,560,002				0.307	0.089	0.001	0.211	0.130	0.105
Interaction between Engagement and Frequency
2nd decile * Frequency							0.080	0.230	0.728
3rd decile * Frequency							−0.257	0.163	0.114
4th decile * Frequency							−0.035	0.170	0.836
5th decile * Frequency							−0.275	0.180	0.125
6th decile * Frequency							−0.197	0.166	0.236
7th decile * Frequency							−0.220	0.163	0.177
8th decile * Frequency							−0.219	0.162	0.178
9th decile * Frequency							−0.246	0.161	0.128
10th decile * Frequency							−0.211	0.160	0.186
Interaction between Engagement and Interactivity
2nd decile * Interactivity							−0.042	0.046	0.363
3rd decile * Interactivity							0.007	0.046	0.880
4th decile * Interactivity							0.098	0.046	0.032
5th decile * Interactivity							−0.034	0.054	0.520
6th decile * Interactivity							−0.017	0.053	0.753
7th decile * Interactivity							0.021	0.059	0.728
8th decile * Interactivity							0.054	0.056	0.337
9th decile * Interactivity							−0.088	0.059	0.141
10th decile * Interactivity							0.072	0.068	0.289
Interaction between Engagement and Issue congruence
2nd decile * Issue congruence							0.077	0.047	0.098
3rd decile * Issue congruence							0.065	0.050	0.193
4th decile * Issue congruence							−0.020	0.049	0.687
5th decile * Issue congruence							−0.024	0.052	0.647
6th decile * Issue congruence							0.128	0.051	0.013
7th decile * Issue congruence							−0.019	0.049	0.691
8th decile * Issue congruence							0.012	0.059	0.835
9th decile * Issue congruence							0.231	0.061	0.000
10th decile * Issue congruence							0.020	0.081	0.808
Constant	9.384	0.081	0.000	9.271	0.088	0.000	9.385	0.122	0.000
N candidates	1246			1246			1246
N constituencies	592			592			592
Conditional R^2	0.935			0.941			0.943

Note: the models also control for candidate gender, party, Cabinet or Shadow Cabinet membership, marginality of constituency, votes for the party in the 2015 General Election, budget, constituency broadband availability, total followers, region, and total valid 2017 constituency vote. See Appendix 4 for full results.

Figure 3. Predicted vote by levels of engagement (based on Model 2 in table 2).

Figure 4. Predicted vote by levels of engagement and issue congruence in tweets (based on Model 3 in table 2).