Does affective polarisation increase turnout? Evidence from Germany, The Netherlands and Spain

Figures & data

Table 1. Key features of studies 1 to 3.

Study	Country	Data structure	AP measure	Turnout measure	Sophistication measure
Study 1	Germany	Aggregate	Parties	Self-report and actual turnout	Political interest
Study 2	Spain	Two-wave panel	Partisans	Self-report	Political interest and knowledge
Study 3	The Netherlands	Eleven-wave panel	Parties	Self-report	Political interest

Figure 1. Trends in key variables.

Table 2. ARIMA model predicting turnout.

	Main model
Affective polarisation	0.022***
	(0.003)
– First lag	−0.001
	(0.003)
– Second lag	0.006+
	(0.003)

Note: An ARIMA(2,0,0) model predicting turnout. Standard errors within parentheses. Controls for in-party sympathy (lag), ideological polarisation (contemporaneous, first lag, and second lag), salience of cultural and economic issues (both lagged), and time until next election (and its squared term). Full model in Online appendix A.

Source: Politbarometer.

Figure 2. Cumulative impulse response functions.

Note: Cumulative impact of unit change of X on Y, with 95% confidence intervals. Based on VAR-model presented in Table A2 in the Online appendices.

Source: Politbarometer.

Table 3. Predicting actual turnout at bundesland elections (key coefficients).

	Separate models	Single model
	Turnout	Turnout
Lagged affective polarisation (previous wave)	0.034+ (0.021)	0.032 (0.023)
Lead affective polarisation (subsequent wave)	0.010 (0.029)	−0.002 (0.032)

N = 100.

Note: Includes fixed effects for federal states and controls for election closeness, ideological polarisation of parties at the federal state level (standard deviation in left–right positions) and a linear time variable. Full model reported in Online appendix A.

⁺0.10; *0.05; **0.01; ***0.001.

Table 4. Lagged DV panel regression predicting (1) turnout and (2) affective polarisation.

	Turnout	Affective polarisation
Lagged turnout		0.027
Lagged affective polarisation	0.125***

Note: Models include controls for the lagged dependent variable; political interest (3 dummies); ideological divergence on immigration (lag), economy (lag), and the Catalan issue (lag); and perceived party polarisation on decentralisation. Standardised variables. Full model in Table B1 in the Online appendices.

Source: E-DEM.

Figure 3. Effect of affective polarisation (z) on turnout intention, by political interest.

Table 5. Lagged DV panel regression predicting (1) turnout and (2) affective polarisation.

	Turnout	Affective polarisation
Lagged turnout		0.153***
Lagged affective polarisation	0.192***

Note: Controlled for lagged dependent variable; political interest (3 categories); ideological divergence on left–right; and wave dummies. Random intercepts for respondents. Full table in Table C1 in the Online appendices.

Source: LISS.

Figure 4. Effect of AP on turnout (left) and vice versa (right).

Note: Based on standardised values of affective polarisation. Based on Table C1 in the Online appendices.

Figure 5. Interaction with political interest.