Do parties perceive their voter potentials correctly? Reconsidering the spatial logic of electoral competition

Lichteblau, Josephine; Giebler, Heiko; Wagner, Aiko Article — Accepted Manuscript (Postprint) Do parties perceive their voter potentials correctly? Reconsidering the spatial logic of electoral competition Electoral Studies Provided in Cooperation with: WZB Berlin Social Science Center Suggested Citation: Lichteblau, Josephine; Giebler, Heiko; Wagner, Aiko (2020) : Do parties perceive their voter potentials correctly? Reconsidering the spatial logic of electoral competition, Electoral Studies, ISSN 1873-6890, Elsevier, Amsterdam, Vol. 65, pp. --, https://doi.org/10.1016/j.electstud.2020.102126 This Version is available at: https://hdl.handle.net/10419/218839 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. https://creativecommons.org/licenses/by-nc-nd/4.0/ 1 SUPPLEMENTARY MATERIAL for Lichteblau, Josephine,* Giebler, Heiko and Wagner, Aiko (2020): Do parties perceive their voter potentials correctly? Reconsidering the spatial logic of electoral competition. Electoral Studies. * Corresponding author, [email protected] 2 A) Additional tables and figures Left/Right Policy issues Combined model Perceived left-right proximity 0.25 *** 0.18 *** (0.02) (0.02) Perceived proximity immigration 0.18 *** 0.13 *** (0.02) (0.02) Perceived proximity economy/welfare 0.04 *** - 0.01 (0.02) (0.02) Perceived proximity nuclear power stations 0.10 *** 0.08 *** (0.02) (0.02) BIC 3454.86 3484.84 3418.06 Pseudo R² 0.05 0.05 0.07 N 6,094 6,094 6,094 TABLE A1: Results of conditional logit models predicting vote choice on ideological and issue proximities Note: Own calculations based on Roßteutscher et al. (2018a; 2018b). Standard errors are presented in brackets. *** = p < 0.001, ** = p < 0.01, * = p < 0.05. 3 Combined model (AME) Perceived left-right proximity 0.04 (0.00) *** Perceived proximity immigration 0.02 (0.0) *** Perceived proximity economy/welfare 0.00 (0.00) Perceived proximity nuclear power stations 0.01 (0.00) *** TABLE A2: Average marginal effects of ideological and policy issue proximities on vote choice Note: Own calculations based on Roßteutscher et al. (2018a; 2018b). Standard errors are presented in brackets. *** = p < 0.001, ** = p < 0.01, * = p < 0.05. In order to validate our claim that the left-right scheme still serves as a crucial heuristic for voters, we calculated three conditional logit models regressing vote choice on perceived ideological and policy issue proximities based on data of the post-election cross-section survey of the GLES 2017. For the ideological proximity, we calculated absolute distances between the respondents’ positions and the perceived party positions with regards to the left-right dimension, both measured on an 11-point scale and recoded so that high values represent high proximities. The same procedure was applied for three policy issues that are also measured on an 11-point scale. These issues refer to a) the redistribution of income (from “lower taxes/less government spending on health, education and social benefits” to “more government spending on health, education and benefits/higher taxes”), b) the restriction of immigration (from “immigration laws should be relaxed” to “immigration laws should be more restrictive”) and c) the future of nuclear power stations (from “more nuclear power stations should be build” to “all should be closed down today”). The first model represents the full model including all four proximity measures, the second model is the policy issue model only including the policy issue proximities and in the third model ideological proximity is the sole predictor of vote choice. The results suggest, firstly, that ideological congruence does not only exert an independent but also – 4 according to the average marginal effects (AMEs) – the strongest effect of all four proximity measures on vote choice. Secondly, comparing the BICs of the ideological and the policy issues model, we see that the former actually fits the data better than the latter. Thus, we can conclude that voters still use the left-right dimension as a heuristic when making their vote choice. TABLE A3: Summary Statistics of variables from the main models Variable Mean Std. Dev. Min. Max. Distribution Perceived voter potential (PVP) 5.78 2.22 1.54 10.16 Actual voter potential (AVP) 4.34 1.81 1.14 7.66 Parties' perceived leftright proximities (PLRP) 7.08 2.02 2.56 10.73 Voters' perceived leftright-proximities (VLRP) 8.38 1.35 4.95 10.06 Voter potential misperception (PVP minus AVP) 1.45 2.60 -1.45 8.59 Differences in leftright perceptions (PLRP minus VLRP) -1.30 1.30 -4.22 0.89 5 Note: Own calculations based on Roßteutscher et al. (2018a; 2018b). FIGURE A1: Perceived voter potential (PVP) and actual voter potential (AVP) Note: Own calculations based on Roßteutscher et al. (2018a; 2018b). 6 FIGURE A2: Left-right placements of parties by respective party candidates and left-right party placements by other parties’ voters Note: Own calculations based on Roßteutscher et al. (2018a; 2018b). 7 B. Robustness-Checks To validate our findings, we conducted a number of (identical) robustness checks for both of our main models, the PVP model (Table 1 in the paper) and the PVP-minus-AVP model (Table 2 in the paper). We calculated several additional models, the results of which are presented in Tables B2 (PVP-models) and B3 (PVP-minus-AVP models). First of all, the number of cases to test our theoretical argument is small resulting in vulnerability to outliers and influential cases. Therefore, we conducted Jackknife tests for both of our main Models (Models M1 in Tables B2 and B3). Here, we used the jackknife routine implemented in STATA. We also looked for outliers from a theoretical perspective. With the AfD, our analyses include a party showing very large deviations between PVP and AVP. Furthermore, the AfD was a rather new party in 2017, which makes it more difficult for other parties to evaluate their voter potentials among this specific target party’s electorate. Therefore, we calculated two additional models for each main model: one in which we excluded all cases containing the AfD as the receiving party (Models M2) and one that excluded all cases containing the AfD as the target party (Models M3). Furthermore, we calculated four additional PVPand PVP-minus-AVP models each, for which PVPand PLRP-scores were aggregated across four different subsamples of individual candidates: experienced vs. inexperienced candidates (M4a and M4b) and MP’s vs. unelected candidates (M5a and M5b). Secondly, we applied our analytical models to a different context, namely to that of the German Federal Elections of 2013 (Models M6). Here, we used data from the GLES 2013 candidate survey (Rattinger et al., 2014) and post-election voter survey (Rattinger et al., 2017). All variables used for our main analyses of the 2017 elections are also part of the 2013 surveys and have the same coding. Therefore, the operationalization of our dependent and independent variables is identical to that described in the research design section of our paper. Furthermore, we calculated a model pooling the 2013 and 2017 observations (M7). Thirdly, to make sure that no omitted variable is at hand, we added a number of control variables to our main models, each at a time. We controlled for several party size related factors: the receiving and target parties’ size (Models M8 and M9), vote gains (Models M10 and M11) as well as their relative vote gains (Models M12). Additionally, we added a variable indicating whether the two members of the party pair belong to the same political camp to our main models 8 (Models M13). Descriptions of these control variables and some summary statistics are presented in Table B1. Finally, we calculated our main models using an alternative operationalization of our dependent variables. Here, we calculated our PVP and PVP-minus-AVP measures using the median of the PTWVand PTV-distribution (Models 14). FIGURE B1: Perceived voter potential (PVP) and actual voter potential (AVP) of experienced vs. inexperienced and elected vs. unelected candidates Note: Own calculations based on Roßteutscher et al. (2018a; 2018b). 15 References Bundeswahlleiter, 2017. Final Results. Statistisches Bundesamt, Wiesbaden. available on: https://www.bundeswahlleiter.de/en/bundestagswahlen/2017/ergebnisse.html. Rattinger, H., Roßteutscher, S., Schmitt-Beck, R., et al., 2014. Candidate Campaign Survey 2013, Survey and Electoral/Structural Data (GLES), Cologne. GESIS Data Archive. ZA5716 Data file Version 3.0.0. doi:10.4232/1.12043. Rattinger, H., Roßteutscher, S., Schmitt-Beck, R., et al., 2017. Post-election Cross Section (GLES 2013). GESIS Data Archive. Cologne. ZA5701 Data file Version 3.0.0. doi:10.4232/1.12809. Roßteutscher, S., Schmitt-Beck, R., Schoen, H., et al., 2018. Candidate Campaign Survey (GLES 2017). GESIS Data Archive. Cologne. ZA6814 Data file Version 3.0.0. doi:10.4232/1.13089. Roßteutscher, S., Schmitt-Beck, R., Schoen, H., et al., 2018. Post-election Cross Section (GLES 2017). GESIS Data Archive. Cologne. ZA6801 Data file Version 4.0.0. doi:10.4232/1.13138.