Revisiting positional choice: Survey evidence from Germany

Behringer, Jan; Endres, Lukas; van Treeck, Till Working Paper Revisiting positional choice: Survey evidence from Germany ifso working paper, No. 33 Provided in Cooperation with: University of Duisburg-Essen, Institute for Socioeconomics (ifso) Suggested Citation: Behringer, Jan; Endres, Lukas; van Treeck, Till (2024) : Revisiting positional choice: Survey evidence from Germany, ifso working paper, No. 33, University of Duisburg-Essen, Institute for Socio-Economics (ifso), Duisburg This Version is available at: https://hdl.handle.net/10419/285374 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/4.0/ uni-due.de/soziooekonomie/wp ifso working paper Jan Behringer Lukas Endres Till van Treeck Revisiting Positional Choice: Survey Evidence from Germany 2024 no.33 Revisiting Positional Choice: Survey Evidence from Germany * Jan Behringer Lukas Endres Till van Treeck March 11, 2024 Abstract We conduct a survey to test for concerns about relative standing based on a large sample of the German population. Our survey approach asks respondents to choose between two hypothetical states of the world, in which they receive either a larger relative endowment or a larger absolute endowment that leaves them worse off in comparison to others. We receive the highest shares of positional answers for household income and one’s children’s IQ. We show that respondents with a larger social comparison orientation are significantly more likely to indicate positional preferences. This finding is robust despite some limited evidence that unintended alternative explanations based on feelings of material deprivation, feeling overworked, respondents’ work ethic and environmental concerns affect respondents’ choice. JEL Classification: C90, D63 Keywords: Social Comparisons; Positionality; Survey Data * Jan Behringer, Macroeconomic Policy Institute (IMK), Georg-Glock-Straße 18, 40474 D¨usseldorf, Germany. E-Mail: [email protected]. Lukas Endres, Macroeconomic Policy Institute (IMK), Georg-Glock-Straße 18, 40474 D¨usseldorf, Germany and the Institute for Socio-Economics, University of Duisburg-Essen. E-Mail: [email protected]. Till van Treeck, Institute for Socio-Economics, University of Duisburg-Essen, Lotharstraße 65, 47057 Duisburg, Germany. E-Mail: [email protected]. We would like to thank participants at the 23rd FMM Conference in Berlin, the PK Conference in Lille, and the PhD-Workshop in D¨usseldorf for helpful comments and suggestions. 1 Introduction Status-oriented individuals are argued to evaluate their own situation in reference to their social environment. A series of empirical contributions quantified the degree to which concerns about relative standing matter across different items, referred to as their positionality (Frank, 2005). This is typically measured by means of a survey approach based on preferences for an improved relative standing at the expense of absolute endowments (Alpizar et al., 2005; Carlsson et al., 2007; Hillesheim and Mechtel, 2013; Solnick and Hemenway, 1998; Solnick and Hemenway, 2005; Solnick et al., 2007). In this paper, we follow up on this research by conducting a survey among a large sample of the German population to test for concerns about relative standing regarding income, leisure time, housing wealth, debt, own children’s intelligence quotient (IQ), and grade point average (GPA). We investigate the behavioural mechanism underlying positional choice by combining our results with self-reported information on the importance of social comparisons and other sentiment variables proxying alternative explanations. It has long been argued that individual satisfaction with the own current situation and hence the utility derived from individual endowments are context dependent. Thus, inequality in endowments implies negative externalities for those being worse off in relative terms at any given absolute endowment. Various empirical contributions have highlighted the relevance of concerns about relative standing to individual well-being (Luttmer, 2005) or satisfaction (Card et al., 2012) and economic behaviour, such as consumptionsaving decisions of private households (Behringer, Endres and van Treeck, 2023; Bertrand and Morse, 2016; Charles et al., 2009; Frank et al., 2014), or labour supply (Behringer, Gonzalez-Granda and van Treeck, 2023; Bowles and Park, 2005; Oh et al., 2012). Our identification strategy follows the literature that measures relative concerns, by asking survey participants to choose between two hypothetical states of the world in which they either receive a larger relative endowment or a higher absolute endowment that leaves them worse off in comparison to others. Any given item is argued to be more positional the larger the share of individuals stating a preference for a relative versus absolute advantage. Prior surveys of this kind consistently find that individuals express 1 preferences for an improved relative standing, even at the cost of being worse off in absolute terms, across a broad range of items. We conduct the, to the best of our knowledge, first large scale survey of this kind based on a sample of the German population comprising 1,552 individuals. Except for a survey by Carlsson et al. (2007) among 335 individuals from Sweden, all other studies are either based on small classroom surveys or convenience samples. Our findings regarding the positionality of individual items line up well with results from the prior literature. We receive the highest share of positional answers for monthly net income and own children’s IQ, followed by housing wealth and own children’s GPA. Our survey respondents least often chose the positional state of the world for leisure time and outstanding debt, which in turn exhibit the highest shares of responses indicating a preference for a more advantageous absolute endowment despite a relative disadvantage. We next investigate potential drivers of positional choice. While our survey design has been applied frequently in the prior literature, the underlying behavioural mechanism has not been empirically verified. We fill this gap by validating whether positional preferences do indeed hinge on comparisons with others. For this purpose, we use self-reported information on the importance of comparisons with others to construct a Social Comparison Index (SCI). Applying this measure in multinomial logit regressions, we show that respondents who report to put a lot of emphasis on comparisons with others are significantly more likely to choose the state of the world with smaller absolute, yet larger relative endowments, holding individual sociodemographic characteristics constant. The magnitude of the effect is closely aligned with the overall share of positional answers for the respective item. This finding is highly consistent with an explanation of individual preferences for higher relative endowments based on concerns about relative standing, thereby validating the applied survey design. We control for and investigate a series of alternative mechanisms that might explain respondents’ preference for either state of the world, irrespective of concerns about relative standing. Feelings of material deprivation might drive preferences for larger absolute endowments regarding material items or working hours. In turn, feeling overworked could 2 be reflected in a preference for lower absolute, but higher relative working hours. A respondents’ work ethic and the associated belief in own agency in determining own outcomes could affect individual preferences over working hours, while a respondents’ stance towards the compatibility of economic growth with the environment might motivate preferences for states of the world with absolutely lower income and wealth. We find that in some cases our proxy variables for the respective competing explanations are significantly related to individual choice of endowments, likely biasing positionality interpretations of observed response shares. However, as the estimated coefficients on our SCI are largely unaffected by the inclusion of any of these measures, we conclude that a substantial part of the variation in positional choice across items and individuals can be ascribed to concerns about relative standing. The remainder of this paper is structured as follows. In Section 2, we describe our survey design and data. Section 3 presents the results from our survey and our empirical analysis of positional choice. Section 4 concludes. 2 Survey Design and Data In December 2022 we conducted an online survey among a large sample of 1,552 individuals.1Survey participants were required to reside in Germany and be between 18 and 75 years old. Our methodological approach to measure the positionality of various items follows the previously established design that asks respondents to choose from two hypothetical states of the world (e.g., Solnick and Hemenway, 1998). One state offers survey respondents a lower absolute, yet higher relative endowment of a given item. In the other state the endowment is higher in absolute terms, yet the respondent would be worse off relative to others. Survey participants also have the option to indicate that they do not have a preference for either state. Our survey includes questions on income, leisure, housing wealth, debt, an own child’s IQ, and GPA. Using the income item as an example, the survey question reads as follows: 1Our initial sample included 1,608 individuals. To ensure data quality we exclude 56 observations of survey participants who spent less than half of the median time to complete the full survey. 3 Imagine that you can choose between two possible worlds (World A and World B). Both worlds are the same except for one difference, which is described below. Which world would you prefer to live in? In both worlds, prices for all goods are the same and at the same level as they are today. ”Others” refers to average other persons in society. (1) A: Your monthly net income as a single person is 3,000 euros; others receive 2,000 euros. (2) B: Your monthly net income as a single person is 4,000 euros; others receive 6,000 euros. (3) I do not have a preference. Full texts for the other survey items are provided in the Appendix. We chose hypothetical endowments for each survey item to most closely reflect the reality of an average person living in Germany, thereby providing relatable examples. For example, the offered values for items such as income or housing wealth are within close range of, yet slightly above the average in Germany. This should also prevent that survey participants avoid picking world A due to loss aversion. Our survey additionally asks respondents to indicate the extent to which they agree with several statements about comparisons with others. We rely on questions that were specifically designed to measure individual differences in social comparison orientation (SCO) and have been shown to effectively capture comparison behaviour (Gibbons and Buunk, 1999). In particular, we ask survey participants about the extent to which they agree with the statements that they frequently compare their popularity with others, that they rarely compare themselves with others and that they never evaluate their situation in life in comparison to others.2Respondents can choose between answers ranging from (1) Disagree completely to (7) Agree completely. We average responses to the three 2Our questions on social comparisons are either translations or close adaptations of those designed by Gibbons and Buunk (1999) and read as follows: (a) “I often compare my social recognition and popularity with other people.”; (b) “I am not the type of person who compares often with others.”; (c) “I never consider my situation in life relative to that of other people.” 4 questions to construct our SCI.3We use this measure to validate the proposed mechanism of positional choice. In addition, we include questions that measure respondents’ sentiment and attitudes towards a variety of topics, which might be relevant to their decision in the main survey irrespective of concerns about relative standing. It could pose a violation of the positionality interpretation of our survey results if respondents’ choice is strongly influenced by any of these. In particular, we ask survey participants about their satisfaction with their household income and leisure time, and their stance towards implications of economic growth for the natural environment.4We also ask about the extent to which they perceive personal economic success and income to be determined by hard work or luck.5 The exact wording of and summary statistics for our separate survey questions on social comparisons, work ethic, and environmental concerns are provided in the Appendix. We utilise these additional questions to measure alternative explanations that could influence respondents’ choice of hypothetical endowments. Specifically, our measure of income satisfaction proxies (perceived) material deprivation. Perceiving the own income as insufficient to meet (material) needs might induce a non-positional response, as it provides larger absolute endowments for items such as income or housing wealth. In turn, satisfaction with leisure time is supposed to capture feelings of being overworked, which could have the opposite effect to income satisfaction. Besides the effect on world choice regarding working hours, individuals might for example prefer higher (absolute) net wealth as a prerequisite to working less. Answers to the questions whether hard work or luck determine personal success reflect respondents’ work ethic. Strongly believing in own agency in achieving desired outcomes might go hand in hand with the desire to work 3To maintain a consistent ordering of the responses, we use inverse values for the latter two statements. 4We measure environmental attitudes by asking respondents to state the degree to which they agree with the statement ”In order to protect the environment Germany needs economic growth.” Our survey additionally includes an alternative, antonymous statement. When using responses to the alternative questions for our analysis we obtain similar results (available on request). These questions stem from the International Social Survey Programme (ISSP) and have frequently been applied in environmental research to measure environmental concerns and perceptions about the compatibility of economic growth and the environment (e.g., Franzen and Vogl, 2013; Drews et al., 2018). 5Specifically, we ask survey participants about the degree to which they agree with the following statement: ”One has to work hard in order to succeed.”. We ask about an antonymous statement, as an alternative measure of work ethic, which yields similar results in our empirical analysis (available on request). 5 longer hours irrespective of how long others work. We also consider survey participants’ stance towards ecological implications of economic growth, as perceiving economic growth as harmful to the natural environment could be associated with preferences for absolutely lower material endowments.6Summary statistics for the final measures applied in our analysis and the SCI are provided in Panel A of Table 1. Lastly, our survey includes questions about sociodemographic characteristics. Specifically, we ask participants about their gender, age, employment status, highest educational attainment, number of children, net household income, and political party preferences. Panel B of Table 1 presents summary statistics of our sample. 3 Results Figure 1 shows response shares by survey item. Across items a substantial share of respondents indicated preferences for the positional state of the world. The fraction of positional answers ranges from 32.8% (net income) to 16.4% (debt). While the positional state was chosen most often for net income endowments, it is also at the top with respect to non-positional answers, making it one of the most polarising items. Particularly for lower income individuals the choice might be governed by feelings of material deprivation, driving up non-positional responses. Concerning the share of positional answers, net income is closely followed by the own child’s IQ, a child’s grades at school and housing wealth. However, these items received substantially lower shares of non-positional answers than the income scenario, primarily resulting from a larger share of respondents indicating indifference. Children’s IQ is the only survey item with a larger share of positional than non-positional answers. Leisure and especially outstanding debt have a substantially lower share of positional answers (25.3% and 16.4%) and the highest shares of non-positional answers (50.5% and 45.9%). In Table 2 we explore the behavioural mechanisms underlying positional choice. The prior literature has ascribed preferences for the positional state of the world to concerns 6While we ask respondents to indicate their preference under a ceteris paribus assumption, it is unlikely that they can completely disregard their own circumstances and beliefs when making their decision. 6 0.0 0.2 0.4 0.6 Income Leisure Wealth Debt IQ GPA World A World B Fraction Figure 1: Response Fractions by Survey Item 13 Table 1: Summary Statistics Mean Std. Dev. Min Max N Panel A: Sentiment Variables SCI 3.04 1.37 1 7 1,542 Satisfied with income 4.39 1.69 1 7 1,549 Satisfied with leisure 5.07 1.55 1 7 1,543 Work ethic 5.57 1.33 1 7 1,537 Growth environment 4.39 1.68 1 7 1,494 Panel B: Sociodemographic Characteristics Male 0.52 0.50 0 1 1,552 Age 52.41 14.00 18 75 1,552 High education 0.28 0.45 0 1 1,552 Employed 0.62 0.48 0 1 1,552 Children 0.30 0.71 0 5 1,535 Net household income Less than e2,000 0.26 0.44 0 1 1,552 e2,000-e2,999 0.24 0.43 0 1 1,552 e3,000-e3,999 0.19 0.39 0 1 1,552 e4,000-e4,999 0.18 0.38 0 1 1,552 e5,000 and more 0.13 0.34 0 1 1,552 Political party preference No preference 0.36 0.48 0 1 1,447 Left Party 0.05 0.22 0 1 1,447 SPD 0.16 0.36 0 1 1,447 Green Party 0.15 0.35 0 1 1,447 CDU/CSU 0.18 0.38 0 1 1,447 FDP 0.04 0.20 0 1 1,447 AfD 0.06 0.24 0 1 1,447 Other 0.01 0.09 0 1 1,447 14 Table 2: Drivers of Positional Choice (1) (2) (3) (4) (5) (6) Panel A: Income SCI 0.037*** 0.038*** 0.036*** 0.036*** 0.040*** 0.040*** (0.009) (0.009) (0.009) (0.009) (0.009) (0.010) Satisfied with income 0.015 0.019* (0.009) (0.010) Satisfied with leisure -0.007 -0.014 (0.008) (0.009) Work ethic 0.012 0.012 (0.010) (0.010) Growth environment 0.001 0.002 (0.008) (0.008) Respondent controls Yes Yes Yes Yes Yes Yes Observations 1425 1424 1419 1416 1382 1371 Pseudo R-squared 0.034 0.040 0.035 0.036 0.036 0.045 (1) (2) (3) (4) (5) (6) Panel B: Leisure SCI 0.017* 0.017 0.017 0.016 0.013 0.013 (0.009) (0.009) (0.009) (0.009) (0.009) (0.009) Satisfied with income -0.004 -0.005 (0.008) (0.009) Satisfied with leisure 0.002 0.001 (0.008) (0.009) Work ethic 0.023* 0.024* (0.010) (0.010) Growth environment -0.001 0.003 (0.007) (0.007) Respondent controls Yes Yes Yes Yes Yes Yes Observations 1425 1424 1419 1416 1382 1371 Pseudo R-squared 0.042 0.042 0.043 0.045 0.044 0.047 Notes: All columns report average marginal effects from multinomial logit regressions of the covariates on the likelihood that an individual chooses the positional outcome (i.e. World A). *** p<0.001, ** p<0.01, * p<0.05. 15 Table 2: Continued (1) (2) (3) (4) (5) (6) Panel C: Wealth SCI 0.025** 0.025** 0.025** 0.025** 0.025** 0.026** (0.009) (0.009) (0.009) (0.009) (0.009) (0.009) Satisfied with income 0.008 0.007 (0.008) (0.009) Satisfied with leisure -0.001 -0.000 (0.008) (0.009) Work ethic 0.002 0.002 (0.009) (0.010) Growth environment 0.006 0.006 (0.007) (0.008) Respondent controls Yes Yes Yes Yes Yes Yes Observations 1425 1424 1419 1416 1382 1371 Pseudo R-squared 0.041 0.041 0.041 0.041 0.042 0.042 (1) (2) (3) (4) (5) (6) Panel D: Debt SCI 0.013 0.013 0.010 0.014 0.012 0.009 (0.007) (0.007) (0.008) (0.007) (0.008) (0.008) Satisfied with income -0.002 0.009 (0.007) (0.008) Satisfied with leisure -0.016* -0.019* (0.007) (0.008) Work ethic -0.001 -0.001 (0.008) (0.008) Growth environment 0.005 0.004 (0.006) (0.006) Respondent controls Yes Yes Yes Yes Yes Yes Observations 1425 1424 1419 1416 1382 1371 Pseudo R-squared 0.048 0.049 0.051 0.049 0.053 0.056 Notes: All columns report average marginal effects from multinomial logit regressions of the covariates on the likelihood that an individual chooses the positional outcome (i.e. World A). *** p<0.001, ** p<0.01, * p<0.05. 16 Table 2: Continued (1) (2) (3) (4) (5) (6) Panel E: IQ SCI 0.040*** 0.039*** 0.038*** 0.040*** 0.035*** 0.035*** (0.009) (0.009) (0.009) (0.009) (0.009) (0.010) Satisfied with income -0.001 0.001 (0.009) (0.010) Satisfied with leisure -0.010 -0.009 (0.008) (0.009) Work ethic 0.013 0.013 (0.010) (0.010) Growth environment 0.006 0.008 (0.008) (0.008) Respondent controls Yes Yes Yes Yes Yes Yes Observations 1425 1424 1419 1416 1382 1371 Pseudo R-squared 0.047 0.048 0.048 0.048 0.048 0.052 (1) (2) (3) (4) (5) (6) Panel F: GPA SCI 0.028** 0.028** 0.027** 0.027** 0.026** 0.025** (0.009) (0.009) (0.009) (0.009) (0.009) (0.009) Satisfied with income -0.001 0.002 (0.009) (0.009) Satisfied with leisure -0.006 -0.007 (0.008) (0.009) Work ethic 0.012 0.007 (0.009) (0.010) Growth environment -0.003 -0.002 (0.007) (0.008) Respondent controls Yes Yes Yes Yes Yes Yes Observations 1425 1424 1419 1416 1382 1371 Pseudo R-squared 0.043 0.044 0.044 0.043 0.044 0.045 Notes: All columns report average marginal effects from multinomial logit regressions of the covariates on the likelihood that an individual chooses the positional outcome (i.e. World A). *** p<0.001, ** p<0.01, * p<0.05. 17 Appendix: Survey Questions The appendix shows English translations of our key survey questions regarding the survey on positional choice and additional sentiment questions which we relied on to construct variables for our empirical analysis in Section 3 in the paper. 1 Positional Choice Imagine that you can choose between two possible worlds (World A and World B). Both worlds are the same except for one difference, which is described below. Which world would you prefer to live in? In both worlds, prices for all goods are the same and at the same level as they are today. ”others” refers to average other persons in society. (a) Income (1) A: Your monthly net income as a single person is 3,000 euros; others receive 2,000 euros. (2) B: Your monthly net income as a single person is 4,000 euros; others receive 6,000 euros. (3) I do not have a preference. (b) Leisure (1) A: Your weekly working time is 38 hours; the weekly working time of others is 41 hours. (2) B: Your weekly working time is 35 hours; the weekly working time of others is 32 hours. (3) I do not have a preference. 18 (c) Wealth (1) A: You live in a property that, like all other properties in your neighbourhood, is worth 300,000 euros; in other neighbourhoods, properties are worth 200,000 euros. (2) B: You live in a property that, like all other properties in your neighbourhood, is worth 400,000 euros; in other neighbourhoods, properties are worth 600,000 euros. (3) I do not have a preference. (d) Debt (1) A: Your level of debt is 400,000 euros; others are 600,000 euros in debt. (2) B: Your level of debt is 300,000 euros; others are 200,000 euros in debt. (3) I do not have a preference. (e) IQ (1) A: Your child’s intelligence quotient (IQ) is 100; the average of other parents’ children is 90. (2) B: Your child’s intelligence quotient (IQ) is 110; the average of other parents’ children is 120. (3) I do not have a preference. (f) GPA (1) A: Your child has a grade point average of 2.0 in school; other parents’ children have a grade point average of 2.5. (2) B: Your child has a grade point average of 1.7 in school; other parents’ children have a grade point average of 1.2. (3) I do not have a preference. 19 2 Sentiment Questions Social Comparison Orientation Most people compare themselves with others from time to time. There is nothing particularly ”good” or ”bad” about this kind of comparison, and some people do it more often than others. What about you: To which degree do you agree with the individual statements? (a) I often compare how I am doing socially (e.g., social recognition, popularity) with other people. (b) I am not the type of person who compares often with others. (c) I never consider my situation in life relative to that of other people. Work Ethic To which degree do you agree with the individual statements? (a) What a person achieves in life is above all a question of fate or luck. (b) One has to work hard in order to succeed. Growth Environment The following statements describe different attitudes towards the environment. To which degree do you agree with the individual statements? (a) In order to protect the environment Germany needs economic growth. (b) Economic growth always harms the environment. Respondents were asked to answer all sentiment questions according to the following scale: (1) 1 = Disagree completely (2) 2 - 6 (3) 7 = Agree completely 20 Table A1: Summary Statistics for the Sentiment Questions Mean Std. Dev. Min Max N Social Comparison Orientation Popularity 2.68 1.66 1 7 1,533 Type of person 4.99 1.77 1 7 1,536 Situation in life 4.58 1.76 1 7 1,532 Work Ethic Achievements fate or luck 3.53 1.62 1 7 1,531 Work hard to succeed 5.57 1.33 1 7 1,537 Growth Environment Growth protects environment 4.39 1.68 1 7 1,494 Growth harms environment 3.70 1.72 1 7 1,525 21 Table A2: Positional Choice and Respondent Characteristics (1) (2) (3) (4) (5) (6) Income Leisure Wealth Debt IQ GPA SCI 0.037*** 0.017* 0.025** 0.013 0.040*** 0.028** (0.009) (0.009) (0.009) (0.007) (0.009) (0.009) Male -0.025 -0.028 -0.044 -0.035 0.015 -0.004 (0.025) (0.023) (0.024) (0.020) (0.025) (0.024) Age >49 -0.020 0.003 -0.067* -0.044 -0.043 -0.089** (0.029) (0.027) (0.029) (0.024) (0.029) (0.029) High education -0.010 -0.022 -0.055* -0.024 -0.033 0.009 (0.028) (0.026) (0.027) (0.022) (0.028) (0.027) Employed 0.034 0.071** -0.025 0.012 -0.015 -0.010 (0.028) (0.026) (0.028) (0.023) (0.028) (0.028) Children -0.002 0.024 0.009 0.007 0.046** 0.009 (0.019) (0.017) (0.017) (0.014) (0.018) (0.017) Net household income (Base = Less than e2,000) e2,000-e2,999 -0.013 -0.020 0.015 0.013 -0.011 -0.013 (0.037) (0.033) (0.034) (0.028) (0.035) (0.034) e3,000-e3,999 -0.030 0.017 -0.008 0.010 0.041 0.020 (0.040) (0.038) (0.036) (0.030) (0.039) (0.038) e4,000-e4,999 -0.043 -0.007 0.086* 0.051 0.033 0.031 (0.040) (0.037) (0.039) (0.033) (0.040) (0.039) e5,000 and more -0.083 -0.016 0.046 0.023 0.013 0.019 (0.044) (0.040) (0.044) (0.035) (0.044) (0.043) Political party preference (Base = No preference) Left Party -0.085 0.001 -0.141** -0.029 -0.115* -0.117* (0.053) (0.054) (0.046) (0.044) (0.050) (0.048) SPD 0.042 0.017 -0.009 -0.002 -0.009 0.053 (0.038) (0.036) (0.037) (0.031) (0.038) (0.038) Green Party -0.009 -0.048 -0.000 -0.048 -0.075* -0.045 (0.038) (0.033) (0.038) (0.029) (0.037) (0.036) CDU/CSU 0.124** 0.073* 0.057 0.031 0.102** 0.084* (0.038) (0.036) (0.037) (0.031) (0.038) (0.037) FDP 0.014 0.048 -0.049 -0.036 0.112 0.028 (0.065) (0.064) (0.060) (0.048) (0.068) (0.062) AfD 0.108 0.041 0.023 0.008 0.049 0.004 (0.056) (0.051) (0.052) (0.045) (0.056) (0.052) Other 0.062 0.130 -0.106 0.008 -0.063 -0.023 (0.139) (0.143) (0.114) (0.114) (0.137) (0.121) Observations 1425 1425 1425 1425 1425 1425 Pseudo R-squared 0.034 0.042 0.041 0.048 0.047 0.043 Notes: All columns report average marginal effects from multinomial logit regressions of the covariates on the likelihood that an individual chooses the positional outcome (i.e. World A). *** p<0.001, ** p<0.01, * p<0.05. 22