Online Appendix belonging too: “Rethinking the Link Between Self-Reported Personality Traits and Political Preferences”

Research on personality and political preferences generally assumes unidirectional causal influence of the former on the latter. However, there are reasons to believe that citizens might adopt what they perceive as politically congruent psychological attributes, or at least be motivated to view themselves as having these attributes. We test this hypothesis in a series of studies. Results of preregistered panel analyses in three countries suggest reciprocal causal influences between self-reported personality traits and political preferences. In two two-wave survey experiments, a subtle political prime at the beginning of a survey resulted in self-reported personality traits that were more aligned with political preferences gauged in a previous assessment. We discuss how concurrent assessment within the context of a political survey might overestimate the causal influence of personality traits on political preferences, and how political polarization might be exacerbated by political opponents adopting different personality characteristics or self-perceptions thereof.

The LISS panel is administered by CentERdata (Tilburg University, The Netherlands), see https://www. lissdata.nl/. The panel is based on a true probability sample of households drawn from the population register. Households that could not otherwise participate are provided with a computer and Internet connection.

Personality
Please use the rating scale below to describe how accurately each statement describes you. (very inaccurate, moderately inaccurate, neither inaccurate nor accurate, moderately accurate, very accurate) Openness: Measured with the following items: Have a rich vocabulary; Have difficulty understanding abstract ideas. (reverse coded); Have a vivid imagination; Am not interested in abstract ideas. (reverse coded); Have excellent ideas; Do not have a good imagination. (reverse coded); Am quick to understand things; Use difficult words; Spend time reflecting on things; Am full of ideas. Conscientiousness: Measured with the following items: Am always prepared; Leave my belongings around. (reverse coded); Pay attention to details; Make a mess of things (reverse coded); Get chores done right away; Often forget to put things back in their proper place (reverse coded); Like order; Shirk my duties (reverse coded); Follow a schedule; Am exacting in my work. Neuroticism: Measured with the following items: Get stressed out easily; Am relaxed most of the time (reverse coded); Worry about things; Seldom feel blue (reverse coded); Am easily disturbed; Get upset easily; Change my mood a lot; Have frequent mood swings; Get irritated easily; Often feel blue. Agreeableness Measured with the following items: Feel little concern for others (reverse coded); Am interested in people; Insult people (reverse coded); Sympathize with others' feelings; Am not interested in other people's problems (reverse coded); Have a soft heart; Am not really interested in others (reverse coded); Take time out for others; Feel others' emotions; Make people feel at ease. *Extraversion Measured with the following items: Am the life of the party; Don't talk a lot (Reverse coded); Feel comfortable around people; Keep in the background (reverse coded); Start conversations; Have little to say (reverse coded); Talk to a lot of different people at parties; Don't like to draw attention to myself (reverse coded); Don't mind being the center of attention; Am quiet around strangers (reverse coded).

Political Preferences
Left-Right In politics, a distinction is often made between "the left" and "the right". Where would you place yourself on the scale below, where 0 means left and 10 means right?
Euthanasia Some people believe that euthanasia should always be forbidden. Others feel that euthanasia should be permitted if the patient expresses that wish. Still others hold an opinion that lies somewhere in between. Where would you place yourself on a scale from 1 to 5, where 1 means that euthanasia should be forbidden and 5 means that euthanasia should be permitted?
Income Differences Some people believe that differences in income should increase in our country. Others feel that they should decrease. Still others hold an opinion that lies somewhere in between. Where would you place yourself on a scale from 1 to 5, where 1 means that differences in income should increase and 5 means that these should decrease? Immigration In the Netherlands, some people believe that immigrants are entitled to live here while retaining their own culture. Others feel that they should adapt entirely to Dutch culture. Where would you place yourself on a scale of 1 to 5, where 1 means that immigrants can retain their own culture and 5 means that they should adapt entirely?
EU Some people and political parties feel that European unification should go a step further. Others think that European unification has already gone too far. Where would you place yourself on a scale from 1 to 5, where 1 means that European unification should go further and 0 means that it has already gone too far?
1.2 Results belonging to Figure 2 in the main text

Cross-lagged Coefficients from RI-CLPM models
The Table below provides the results from the Cross-lagged coefficients from the RI-CPLM models. The full SEM output can be derived from the replication files. In each of the three waves (2015, 2017 and 2019) personality was measured with the following 10 questions. First participants read "To get started, we will ask some questions concerning your personality." Followed by "To what extent do the following statements apply to you?". Respondents evaluated themselves on a scale with the following options: (1) "is not at all similar to me", (2) "is not similar to me", (3) "is just a bit similar to me", (4) is fairly similar to me, (5) "is similar to me", (6) "is very similar to me". In parentheses we indicate whether the item is reverse coded and to which trait the item belongs: Openness was measured with the items: only little artistic interest (reverse coded); have a lively fantasy. Conscientiousness was measured with the items: rather lazy (reverse coded); accomplish tasks effectively. Neuroticism was measured with wit the items: relaxed (reverse coded); easily becomes nervous. Agreeableness: was measured with the items: easily trust people; tend to criticize others (reverse coded). Extraversion was measured with the items: communicative and sociable; rather reserved (reverse coded)

Political Preferences
Left-right In politics people sometimes talk of "left"and "right". Where would you place yourself on this scale, where 0 means "left"and 10 means "right"?
EU "Some say European unification should be pushed further. Others say it already has gone too far. What is your opinion? Should European unification be pushed further or has it gone too far?" Ratings on a scale ranging from "further promote" (1) to "gone too far" (7).

Results belonging to Figure 3 in the main text
The results belonging to the pooled models for left-right ideological self-placement and EU attitudes are provided below. Complete output can be derived using the replication materials. [!h]

GSS panel (United States)
For detailed information about the GSS [Smith, 2015], see http://gss.norc.org/Documents/codebook/ Panel%20Codebook.pdf. The GSS data can be found on the APSR Dataverse page.

Personality
To measure authoritarianism, participants were asked the following question: "If you had to choose, which thing on this list would you pick as the most important for a child to learn to prepare him or her for life?" Respondents were asked to rank 5 traits as "first", "second", "third", "fourth" and "fifth", namely: To obey (OBEY; Authoritarianism); to be well like popular; To think for himself or herself (THINKSELF; Authoritarianism, reverse scored); To work hard; To help others when they need help.
We used rankings to variables OBEY ("to obey" ) and TIHNKSELF ("To think for himself or herself") as indicators of authoritarianism [Stenner, 2005, Feldman andStenner, 1997]. We reverse scored obey, so that first = 5, and took the average of the two measures as our index of authoritarianism. Note that respondents who did not provide an answer to both questions were set to missing.

Political Preference
Abortion scale: We used responses to a series of yes/no questions that gauge under which circumstances a respondent thinks abortion should be legal. Respondents read: "Please tell me whether or not you think it should be possible for a pregnant woman to obtain a legal abortion if..." and the statements below are followed. Respondents answer with "yes" or "no". In particular we rely upon the items: If there is a strong chance of serious defect in the baby? (BDEFECT); If she is married and does not want any more children? (ABNOMORE); If the woman's own health is seriously endangered by the pregnancy? (ABHLTH); If the family has a very low income and cannot afford any more children? (ABPOOR); If she became pregnant as a result of rape? (ABRAPE); If she is not married and does not want to marry the man? (ABSINGLE); The woman wants it for any reason? (ABANY) We coded "No" (1) and "yes" (0). These responses were linearly combined into a scale ranging from support for abortion to opposition to abortion.
LGBT scale: Consists of the items: Suppose this admitted homosexual wanted to make a speech in your community. Should he be allowed to speak, or not? (SPKHOMO). Scored with "Yes, allowed to speak" or "Not allowed".; Should such a person be allowed to teach in a college or university, or not? (COLHOMO). Scored with "Yes, allowed to teach" or "Not allowed".; Some people in your community suggested that a book he wrote in favor of homosexuality should be taken out of your public library, would you favor removing this book, or not? (LIBHOMO). Scored with "Remove" or "Not remove" For item SKPHOMO and COLHOMO we coded "not allowed" as 1 and "yes,..." as 0. For LIBHOMO we code "remove" as 1 and "not remove" as 0. These responses were linearly combined into a scale.

Conservatism:
We use the item POLVIEWS: "We hear a lot of talk these days about liberals and conservatives. I'm going to show you a seven-point scale on which the political views that people might hold are arranged from extremely liberal-point 1-to extremely conservative-point 7. Where would you place yourself on this scale?" PARTYID: We use the item PARTYID: "Generally speaking, do you usually think of yourself as a Republican, Democrat, Independent, or what?" Respondents answer on the following scale: "Strong Democrat" (1); "Not very strong Democrat" (2); "Independent, close to Democrat" (3); "Independent (Neither, no response)" (4); "Independent, close to Republican" (5); "Not very strong Republican" (6); "Strong Republican" (7).
Social welfare attitudes scale: The GSS included four items that we combined to create a social welfare attitudes scale: Is it the government's responsibility to improve the standard of living of all Americans? (HELPPOOR); Is it the responsibility of the government in Washington to see to it that people have help in paying for doctors and hospital bills? (HELPSICK); Should the government try to solve our nations problems (HELPNOT); Should the government reduce the income differences between rich and poor? (EQWLTH).
These items were linearly combined into a scale following our pre-analysis plan.

Results belonging to Figure 4 in the main text
The results belonging to the pooled models are provided for abortion attitudes, LGBT attitudes, conservatism, partisanship and welfare attitudes.

CLPM and RI-CLPM
Though use of cross-lagged panel models [Finkel, 1995] is quite common and we presently pre-registered the plan to use them, it is important to note that this procedure has limitations. In particular, the traditional CLPM does not distinguish within-person and between-person variance, and is ultimately susceptible to omitted variable bias. Simply including individual-fixed effects can severely downwardly bias estimates, and even lead to sign reversals, when the number of waves of data is small [Nickell, 1981]. To ameliorate this issue, several models have been suggested [e.g., Allison et al., 2017, Hamaker et al., 2015 with the Random Intercept-Cross Lagged Panel Model (RI-CLPM) being perhaps the most popular. The RI-CLPM decomposes observed scores into "grand means, stable between components and fluctuating within components" [Mulder and Hamaker, 2020, p. 2]. However, this approach also has serious drawbacks when the number of waves is small (t<10). First, "if the true causal effect of X on Y is contemporaneous rather than lagged, [these models] substantially underestimate the true coefficient size ..." [Leszczensky and Wolbring, 2019, p. 7]. For instance, if changing ones political attitudes affects ones reported personality in less than two years (the shortest time period between measurements in the three datasets we analyze), than the relationship between the variables will appear substantially smaller (possibly to the point of the direction flipping) than it actually is. Vaisey and Miles [2017, p. 62] argue that in 3-wave panel datasets that collect measurements infrequently (as the present ones do), such downwardly biased estimates "will likely be the rule rather than the exception." [Leszczensky et al., 2018]. While Vaisey and Miles [2017] focus on Allison's [2009] model, which is similar to the RI-CLPM, our own simulations demonstrate that a contemporaneous reciprocal relationship between two variables is severely biased when using the RI-CLPM.
We simulate two sets of three-wave longitudinal models. In the first, we explicitly set the contemporaneous relationship between the two sets of variables to zero. In the second, we allow for contemporaneous effects. We set the true relationship between lagged X and Y to =.2. The autocorrelation between lagged Y and Y was set to =.7, to approximately match the high stability of personality we see in our data (increasing autocorrelation or also setting the autocorrelation in the X variable to be high makes RI-CLPM results even worse). True correlations for all other variables are randomly set to lie between .1 and .3. We then simulate 4000 datasets: one for RI-CLPM with no contemporaneous effect; one for RI-CLPM with a contemporaneous effect; one for RI-CLPM with a contemporaneous effect randomized to be between .1 and .3; and one for CLPM with a contemporaneous effect randomized to be between .1 and .3. We plot the estimated coefficients between lagged X and Y in Figure 1.
While RI-CLPM and CLPM models with no contemporaneous results recover, on average, the true relationship between lagged X and Y, the RI-CLPM model estimates are severely negatively biased when there is a contemporaneous effect. On average, the estimate is in the wrong direction. The CLPM model, on the hand, returns coefficients that are biased but far less so-the CLPM model is off by 50 percent while the RI-CLPM model is off by 100 percent. q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q Allison [2015, p.1] suggests a test for this bias: "estimate models that include both contemporaneous and lagged predictors. If a one-year lag is the correct specification, then the contemporaneous effect should be small and not statistically significant. If, on the other hand, the contemporaneous effect is large and significant, it should raise serious doubts about validity of the method and the kinds of conclusions that can be drawn." That positive and significant contemporaneous relationships between the outcome measure and the predictor appear frequently in our data raises serious concern about the validity of this method with the present data. To illustrate this, we plot the contemporaneous relationship between authoritarianism and political attitudes from the GSS in Figure 2 where four of the five contemporaneous effects are positive. Another concern is statistical power. The RI-CLPM model typically has less statistical power than does the traditional cross-lagged panel model, especially when comparing variables that are relatively stable [Allison, 2009, Hill et al., 2020. This increases the possibility of Type II error. Underpowered studies have a lower likelihood of replicating [Asendorpf et al., 2013].
Given the here discussed limitations of the RI-CLPM, we believe the traditional cross-lagged panel model estimates are more credible than the RI-CLPM models. Regardless, as our discussion above illustrates, drawing causal conclusions from longitudinal survey data is a challenging and fraught endeavor. We presently report panel data results using the traditional, and preregistered, CLPM and include the RI-CLPM models in this Online Appendix (Netherlands: SI 1.4; Germany: SI 2.3; United States SI 3.4). The RI-CLPM analyses suggest an absence of causal influences between personality and politics in either direction.

Question wording wave 1
After the introduction, where respondents reported their Amazon Mechanical Turk Worker ID, respondents answered the following questions.
Partisanship (2 questions) Participants were asked to indicate whether they 'usually think of themselves as "a Republican, a Democrat, an Independent, or something else." Democrats and Republicans were routed to a question where they were asked "would you call yourself a strong [Democratic/Republican] or not very strong [Democratic/Republican]". Respondents who reported that they are Independent or something else were asked, "Do you think of yourself as closer to the Republican Party or the Democratic Party?", with response options of "Closer to the Republican Party", "Closer to the Democratic Party" or "neither".
Liberal-conservatism self-placement "Here is a seven-point scale on which the political views that people might hold are arranged from extremely liberal to extremely conservative. Where would you place yourself on this scale?" Response options were: Extremely liberal (1); Liberal (2); Slightly liberal (3); Moderate; middle of the road (4); Slightly conservative (5); Conservative (6); Extremely conservative (7).
Cultural conservatism Composite of the following items: • Abortion: "Some people feel that the government should not restrict a woman's right to an abortion, and that women should be able to have them whenever they choose. Others feel that abortion is wrong and should not be allowed. Which is closer to the way you feel?" Scored on a scale with the following categories: Strongly feel government should not restrict abortion (1); Feel government should not restrict abortion (2); I feel equally close to both views (3); Feel abortion is wrong and should not be allowed (4); Strongly feel abortion is wrong and should not be allowed (5).
• Deporting immigrants: "Do you support or oppose sending unauthorized immigrants now living in the United States back to their home country?" Scored on a scale with the following categories: Strongly support (1); Somewhat support (2); Neither support nor oppose (3); Somewhat oppose (4); Strongly oppose (5).
• Affirmative action for African American and Hispanic students: "Do you support or oppose race-based affirmative action programs designed to increase the number of African American and Hispanic students on college campuses?" Scored on a scale with the following categories: Strongly support (1); Somewhat support (2); Neither support nor oppose (3); Somewhat oppose (4); Strongly oppose (5).
Economic conservatism A composite of the following 2 items: • Government's role in providing a good standard of living: "Some people feel the government in Washington should see to it that every person has a job and a good standard of living. Others think the government should just let each person get ahead on their own. Which is closer to the way you feel?" Scored on a scale with the following categories: Strongly feel government should see to jobs and standard of living (1); Somewhat feel government should see to jobs and standard of living (2); I feel equally close to both views (3); Somewhat feel government should let each person get ahead on own (4); Strongly feel government should let each person get ahead on own (5).
• Role of government in providing health care: "Do you think it is the responsibility of the federal government to make sure all Americans have healthcare coverage, or is that not the responsibility of the federal government?" Scored on a scale with the following categories: Strongly believe it is the responsibility of the federal government (1); Somewhat believe it is the responsibility of the federal government (2); I feel equally close to both views (3); Somewhat believe it is not the responsibility of the federal government (4); Strongly believe it is not the responsibility of the federal government (5).
Presidential approval "Do you approve or disapprove of of the way Donald Trump is handling his job as president?" Scored on a scale with the following categories: Strongly approve (1); Somewhat approve (2); Neither approve or disapprove (3); Somewhat disapprove (4); Strongly disapprove (5).
Attention check wave 1 "We are interested in what issues people believe to be the most important issue facing the country. To show that you've read this much, please mark both the taxes and none of the above boxes below. That's right, just select these two options only." Options included "economy", "immigration", "taxes", "poverty", "none of the above" and "all of the above." Other measures in the survey not part of this study The survey also included two questions about religiosity
• Latino: "Are you Spanish, Hispanic or Latino?" with response options of "Yes" and "No".
• Race-ethnic group: "Which of these best describes your racial-ethnic group?" with response options of "White", "Black or African American", "American Indian or Alaska Native", "Asian or Asian-American", "Native Hawaiian or Pacific Islander", "Mixed race", or Other.
• Age: "What is your age?" • Education: "What is the highest level of education that you have already completed?" Participants could answer with: Did not finish high school; High school degree or equivalent; 2-year College degree (Associate's degree); 4-year College degree (Bachelor's degree); Graduate degree (such as Master's, Ph.D., JD, MD, etc.).
For one-third of the respondents this was followed by a survey experiment unrelated to this study.

Question wording wave 2
At the start of the study, participants were randomly assigned to the politics treatment or the internet treatment. Here we provide the item wording of both treatments.

Politics treatment
After entering their Amazon Mechanical Turk Worker ID, participants received the following instructions: "Thank you for taking our survey. This survey will ask you questions about your political views. Please answer honestly. Your responses will be kept completely anonymous." This was followed by 6 questions about their political views listed below in the order in which they were presented.
• Partisanship (2 questions) Participants were asked to indicate whether they 'usually think of themselves "as a Republican, a Democrat, an Independent, or something else." Democrats and Republicans were routed to a question where they were asked "would you call yourself a strong [Democratic/Republican] or not very strong [Democratic/Republican]". Respondents who indicated "Independent" or "something else" were asked, "Do you think of yourself as closer to the Republican Party or the Democratic Party?" with response options of "Closer to the Republican Party", "Closer to the Democratic Party", or "neither".
• Liberal-conservatism self-placement (1 question): "Here is a seven-point scale on which the political views that people might hold are arranged from extremely liberal to extremely conservative. Where would you place yourself on this scale?" with answer categories: Extremely liberal (1); Liberal (2); Slightly liberal (3); Moderate; middle of the road (4); Slightly conservative (5); Conservative (6); Extremely conservative (7).
• Health care (1 item): "Do you think it is the responsibility of the federal government to make sure all Americans have healthcare coverage, or is that not the responsibility of the federal government?" with answer categories: Strongly believe it is the responsibility of the federal government (1); Somewhat believe it is the responsibility of the federal government (2); I feel equally close to both views (3); Somewhat believe it is not the responsibility of the federal government (4); Strongly believe it is not the responsibility of the federal government (5).
• Abortion (1 item): "Some people feel that the government should not restrict a woman's right to an abortion, and that women should be able to have them whenever they choose. Others feel that abortion is wrong and should not be allowed. Which is closer to the way you feel?" with answer categories: Strongly feel government should not restrict abortion (1); Feel government should not restrict abortion (2); I feel equally close to both views (3); Feel abortion is wrong and should not be allowed (4); Strongly feel abortion is wrong and should not be allowed (5).
Internet treatment After entering their Amazon Mechanical Turk Worker ID, participants received the following instructions: "Thank you for taking our survey. This survey will ask you questions about your Internet browsing preferences. Please answer honestly. Your responses will be kept completely anonymous." This was followed by 6 questions about their internet browsing preferences and behavior listed below in the order in which they were presented.
• Browsing preferences (2 questions): Participants were asked to indicate whether they "prefer browsing the Internet on a smartphone, a tablet (such as an iPad), a laptop computer, or a desktop computer." Based on their answer to the first question, participants were asked "Would you say you strongly or only somewhat prefer browsing the Internet on a [smartphone / tablet / laptop computer / desktop computer], compared to other devices?" which they could answer with "Strongly prefer" or "Somewhat prefer".
• Time spend on the internet (1 question): "On average, how many hours per day do you spend on the Internet?" with the following answer categories: Less than one hour a day (1); 1-2 hours a day (2); 2-3 hours a day (3); 3-4 hours a day (4); and More than 4 hours a day (5).
• Time spent on social networks (1 question): "On average, how many hours per day do you spend on social networks, such as Facebook and Instagram?" with answer categories Less than one hour a day (1); 1-2 hours a day (2); 2-3 hours a day (3); 3-4 hours a day (4); More than 4 hours a day (5).
• Online shopping (1 question): "How often do you shop online?" with response categories: More than once a day (1); About once a day (2); About once a week (3); About once every two weeks (4); About once a month (5); Less than once a month (6); Never (7).
• Video games (1 question): "How often do you play video games online?" with response categories: More than once a day (1); About once a day (2); About once a week (3); About once every two weeks (4); About once a month (5); Less than once a month (6); Never (7).
After the politics or internet treatment all participants were routed to the same personality inventory.
Personality Participants received the following instructions: "Here are a number of characteristics that may or may not apply to you. For example, do you agree that you are someone who likes to spend time with others? Please click a button to the right of each statement to indicate the extent to which you agree or disagree with that statement. Note that disagreement is on the left side of the rating scale and agreement is on the right side of the rating scale. I am someone who . . . " Below the items for each trait are listed. Items were rated with this scale: "Disagree strongly", "Disagree a little", "Neutral; No opinion", "Agree a little", and "Agree strongly". The order in which the items were presented was varied randomly across participants.
Openness was measured with the items: "Is fascinated by art, music, or literature"; "Has little interest in abstract ideas"; "Is original. comes up with new ideas". Conscientiousness was measured with the items: "Tends to be disorganized" (reverse coded); "Has difficulty getting started on tasks" (reverse coded); "Is reliable, can always be counted on". Extraversion was measured with the items:"tends to be quiet" (reverse coded); "Is dominant, acts as a leader"; "Is full of energy". Agreeableness was measured with the items: "Is compassionate, has a soft heart"; "Is sometimes rude to others"; "Assumes the best about people". Neuroticism was measured with the items: "Worries a lot"; "Tends to feel depressed, blue"; "Is emotionally stable, not easily upset" (reverse coded).

Measures wave 1
We asked standard questions about demographics (age, gender, race, education). Political preferences were measured using partisanship, presidential approval, conservatism and cultural and economic ideology. Partisanship was measured using the standard 7-category partisan identification question asked by the American National Election Study ranging from strong Democrat (1) to strong Republican (7). Presidential approval with a 7-point scale ranging from strong disapproval of President Trump (1) to strong approval of President Trump (7). Conservatism on a 7-point scale ranging from "extremely liberal" (1) to "extremely conservative" (7). Cultural conservatism was measured using four items: (1) abortion, (2) same-sex marriage, (3) deporting undocumented immigrants, (4) affirmative action for African American's and Hispanic students. Economic conservatism was measured using two items: (1) government's role in providing a good living and (2) role of government in providing healthcare. The item wording of all items be found in OA 4.1.1 and the actual survey can be found on our OSF page: https://osf.io/kqexp/.
For the test of our pre-registered hypotheses, we created an overarching general ideology dimension consisting of our measures of partisanship, presidential approval, social and economic policy preferences. We scaled this variable so that it ranged from the most left-wing (liberal; 0) to the most right-wing (conservative; 1) political preferences -for descriptive statistics and psychometric properties of this and other ideology dimensions see Supplementary Information 5.1 and 5.2 on Dataverse. The cultural conservatism scale was created by summing the four cultural conservatism items from wave one and recoding this into a scale ranging from the most left position on the scale (0) to the most conservative position on the scale (1). Following the same procedures, we create an economic conservatism scale. Additionally, we created a symbolic ideology dimension as a composite of party identification, presidential approval, and conservatism. Figure 5

in the main text
The results belonging to the test using policy preferences in Figure 5 are presented in Table 18. Partisanship (2 questions) Participants were asked to indicate whether they usually think of themselves "as a Republican, a Democrat, an Independent, or something else." Democrats and Republicans were routed to a question where they were asked "would you call yourself a strong [Democratic/Republican] or not very strong [Democratic/Republican]". People reporting "Independent" or "something else" were asked, "Do you think of yourself as closer to the Republican Party or the Democratic Party?" with response options of "Closer to the Republican Party", "Closer to the Democratic Party", and "neither".
Liberal-conservatism self-placement "Here is a seven-point scale on which the political views that people might hold are arranged from extremely liberal to extremely conservative. Where would you place yourself on this scale?" scored on a likert-type scale with categories: Extremely liberal (1); Liberal (2); Slightly liberal (3); Moderate; middle of the road (4); Slightly conservative (5); Conservative (6); Extremely conservative (7).

Experimental treatments
Politics Condition The survey was introduced with the following instructions: "Thank you for taking our survey. This survey will ask you questions about your political views. Please answer honestly. Your responses will be kept completely anonymous." • Partisanship (2 questions) Participants will be asked "Generally speaking, do you usually think of yourself as a Republican, a Democrat, an Independent, or something else?" Democrats and Republicans will be routed to a question asking, "Would you call yourself a strong [Democratic/Republican] or not very strong [Democratic/Republican]" which they will answer by selecting either "Strong [Democratic/Republican]" or "Not very strong [Democratic/Republican]". People reporting that they are Independent, or something else, will be asked, "Do you think of yourself as closer to the Republican Party or the Democratic Party?" which they will answer by selecting"Closer to the Republican Party", "Closer to the Democratic Party", or "neither".
• Conservatism (1 question): "Here is a seven-point scale on which the political views that people might hold are arranged from extremely liberal to extremely conservative. Where would you place yourself on this scale?" with answer categories: Extremely liberal; Liberal; Slightly liberal; Moderate; middle of the road; Slightly conservative; Conservative; Extremely conservative.
• Immigration (1 item): "As you may know, thousands of refugees have left Honduras and other Central American countries to escape the situation there. Would you approve or disapprove of allowing these refugees to come into this country?" with answer categories: Strongly disapprove; Somewhat disapprove; Neither approve nor disapprove; Somewhat approve; Strongly approve.
• Abortion (1 item): "Some people feel that the government should not restrict a woman's right to an abortion, and that women should be able to have them whenever they choose. Others feel that abortion is wrong and should not be allowed. Which is closer to the way you feel?" with answer categories: Strongly feel government should not restrict abortion; Feel government should not restrict abortion; I feel equally close to both views; Feel abortion is wrong and should not be allowed; Strongly feel abortion is wrong and should not be allowed.
• Presidential approval (1 item): " Do you approve or disapprove of of the way Donald Trump is handling his job as president?" with answer categories: Strongly approve; Somewhat approve; Neither approve nor disapprove; Somewhat disapprove; Strongly disapprove.
• What do you like: "These days, Democrats and Republicans in Washington represent very different policy platforms. Regardless of whether you prefer the Democrats or the Republicans, please take a few minutes to think about why you choose to prefer that party, as opposed to the other party. In the space below, please write 3 or 4 things you especially like about the political party that you like best." This question had an open-ended response.
• What do you dislike: " Now please take another few minutes to consider your biggest criticisms against the opposing political party. What do you dislike about the other party? In the space below, please write 3 or 4 things you especially dislike about the party you do not prefer." This question had an open-ended response.

Internet condition
Internet treatment After entering their Amazon Mechanical Turk Worker ID, participants received the following instructions: "Thank you for taking our survey. This survey will ask you questions about your Internet browsing preferences. Please answer honestly. Your responses will be kept completely anonymous." This was followed by 6 questions about their internet browsing preferences and behavior listed below in the order in which they were presented.
• Browsing preferences (2 questions): Participants were asked to indicate whether they "prefer browsing the Internet on a smartphone, a tablet (such as an iPad), a laptop computer, or a desktop computer." Based on their answer to the first question, participants were asked "Would you say you strongly or only somewhat prefer browsing the Internet on a [smartphone / tablet / laptop computer / desktop computer], compared to other devices?" which they could answer with "Strongly prefer" or "Somewhat prefer".
• Time spend on the internet (1 question): "On average, how many hours per day do you spend on the Internet?" with answer categories: Less than one hour a day (1); 1-2 hours a day (2); 2-3 hours a day (3); 3-4 hours a day (4); and More than 4 hours a day (5).
• Time spent on social networks (1 question): "On average, how many hours per day do you spend on social networks, such as Facebook and Instagram?" with answer categories Less than one hour a day (1); 1-2 hours a day (2); 2-3 hours a day (3); 3-4 hours a day (4); More than 4 hours a day (5).
• Online shopping (1 question): "How often do you shop online?" with response categories: More than once a day (1); About once a day (2); About once a week (3); About once every two weeks (4); About once a month (5); Less than once a month (6); Never (7).
• Video games (1 question): "How often do you play video games online?" with response categories: More than once a day (1); About once a day (2); About once a week (3); About once every two weeks (4); About once a month (5); Less than once a month (6); Never (7).
• What do you like: "These days, people have very different browsing preferences. Regardless of whether you prefer to browse the Internet on a smartphone, a tablet (such as an iPad), a laptop computer, or a desktop, please take a few minutes to think about why you prefer this device, as opposed to the other devices. In the space below, please write 3 or 4 things you especially like about the device you like best." This question had an open-ended response.
• What do you dislike: "Now please take another few minutes to consider your biggest criticisms against the other devices you could use to browse on the internet. What do you dislike about these other devices? In the space below, please write 3 or 4 things you especially dislike about the devices you do not prefer to use to browse the internet." This question had an open-ended response.

Panel studies
The panel studies were not collected by us but the panel studies adhere to APSA's Principles and Guidance for Human Subject Research. In all three panel studies, informed consent was obtained and participants could withdraw their participation at anytime. The panelists were assured that their data would be kept anonymous and no identifiable information was collected. The panel studies do not contain deception.
In the Dutch LISS panel participants receive a compensation which is equivalent to a rate of 15 Euros per hour (based on how long panellists spend on average on that specific survey). In the German GESIS panel, participants receive 5 Euro for every completed survey that takes on average 15 minutes to complete. In the GSS panel participants received an -undisclosed -"token of our appreciation for your time and cooperation", see https://gss.norc.org/for-survey-participants.

Big Five and Authoritarianism Experiments
The original data collection we undertook for these studies adheres to APSA's Principles and Guidance for Human Subject Research.
We obtained informed consent, and respondents were told they could withdraw their data at anytime. Respondents were assured that their data would be kept anonymous. Respondents were contacted via their anonymous Mechanical Turk ID. We did not collect any identifiable information. IDs were removed before this data is publicly shared.
The experiments did not involve deception. Respondents were merely randomly assigned to different surveys.
Respondents, on average, took 10 minutes to complete the survey and were paid $1.50, which is equivalent to $9 an hour. This is above the national minimum wage, and considered high on Mechanical Turk (a 2018 study found that the average earnings on that platform was $2 per hour).

Individual Level Meta-Analysis
The table below provides provides the results belonging to the individual data meta-analysis that we ran pooling across the Big Five and Authoritarianism experiments. Note that the standard errors are clustered at the individual level and one-sided p-values are reported.