The paper
Yphtach Lelkes and Sean J. Westwood (2017). The Limits of Partisan Prejudice. The Journal of Politics 79(2), 485–501.
doi:10.1086/688223 (opens in a new tab) · Replication data: doi:10.7910/DVN/D5WS0Y (opens in a new tab)
Main text
p. 489, Study 1
Published
Participants who read the article that criticized the opposing party recommended inclusion 64.74% of the time, while only 25.41% of participants recommended sharing the article criticizing their own party.
Correction
Participants who read the article that criticized the opposing party recommended inclusion 76.52% of the time, while only 37.67% of participants recommended sharing the article criticizing their own party.
Basis In the archived Study 1 data, 202 of the 264 participants who read the opposition-criticism article recommended it, and 110 of the 292 who read the own-party-criticism article did. The printed 64.74% is the share of recommenders who had read the opposition-criticism article (202 of 312), and 25.41% is the share of non-recommenders who had read it (62 of 244).
p. 491, Study 2.1
Published
Roughly 54% of players excluded the opposing partisan player and opted for the less educated Independent, while 94% of players selected the co-partisan to join their team.
Correction
Roughly 43% of players excluded the opposing partisan player and opted for the less educated Independent, while 94% of players selected the co-partisan to join their team.
Basis In the archived Study 2.1 data, 260 of the 611 participants (42.55%) left the opposing partisan off their team. The printed figure is close to the share of co-partisan selectors who included the opposing partisan (316 of 576, 54.86%).
p. 492, Study 2.2
Published
Indeed, the most commonly selected trait is partisanship, with 43.3% of participants attributing their exclusion to their party affiliation (table 3).
Correction
Indeed, the most commonly selected trait is partisanship, with 44.4% of participants attributing their exclusion to their party affiliation (table 3).
Basis The archived Study 2.2 file has 157 rows, 4 of them empty, and the printed percentages divide by all 157. The corrected figure divides by the 153 participants the text reports.
pp. 492-493, Study 2.2
Published
The percent of individuals reporting their partisanship as a reason for exclusion is relatively stable across levels of affective polarization and is similar for those at the lowest level of affective polarization (42.9%) and those at the highest level of affective polarization (48.5%). Although many people identified that partisanship had an effect on their exclusion, those at the lowest levels of polarization were also likely to attribute their exclusion to age and, to a lesser extent, their level of education and martial status.
Correction
The percent of individuals reporting their partisanship as a reason for exclusion does not differ significantly between those in the lowest quartile of affective polarization (39.0%) and those in the highest quartile (55.3%; p = .22). Although many people identified that partisanship had an effect on their exclusion, those at the lowest levels of polarization were also likely to attribute their exclusion to age and, to a lesser extent, their level of education and marital status.
Basis Quartiles and test as in the corrected Table 3 (p. 493).
p. 493, Table 3
Published
Table 3, “Respondent-Identified Trait(s) Responsible for Exclusion”: “Overall (%)” 33.1, 22.3, 43.3, 8.3; “Q1 of Affective Polarization (%)” 45.7, 28.6, 42.9, 11.4; “Q4 of Affective Polarization (%)” 30.3, 18.2, 48.5, 3.0; “Significance (Q1 − Q4)” p = .47, .82, .39, .70 (rows “Age”, “Level of education”, “Political party”, “Martial status”).
Correction
Corrected Table 3.
Corrected Table 3: respondent-identified trait(s) responsible for exclusion. Overall: all 153 participants. Q1 and Q4: lowest and highest quartiles of affective polarization (n = 41 and 38). Two-sample tests of proportions, as in the original. Reason Overall (%) Q1 of Affective Polarization (%) Q4 of Affective Polarization (%) Significance (Q1 − Q4) Age 34.0 41.5 28.9 p = .35 Level of education 22.9 26.8 21.1 p = .74 Political party 44.4 39.0 55.3 p = .22 Marital status 8.5 9.8 5.3 p = .74 Basis The archived isolation_exp.R prints the p-values for education, party, marital status and “none of these” against the rows Age through Marital status. With the printed groups, the p-values are Age .29, Level of education .47, Political party .82 and Marital status .39. The printed Q1 and Q4 columns are the lowest and highest of four equal-width bins of the affective polarization score (n = 35 and 33); the corrected table uses quartiles and divides the overall column by the 153 participants. No difference between Q1 and Q4 is significant under either grouping.
p. 494, Study 3.1
Published
Only a small minority of participants agreed with the police action (i.e., gave responses above the midpoint), and this proportion is not significantly different when the protesters were in-party members or when the protesters were out-party members (15% vs. 20%, respectively). Participants tended to disagree with the police actions, with a small and insignificant difference (b = −.05, 95% confidence interval [−.12, 0.01]) in disagreement depending on whether participants were asked about the in-party or the out-party. The mean proposed fine toward the protesters is significantly lower in the in-party condition ($1,713.75) than in the out-party condition ($3,000.21), a difference of $1,286.46 (95% confidence interval [$553.72, $2,019.19]) or 57.12%.
Correction
Only a small minority of participants agreed with the police action (i.e., gave responses above the midpoint), and this proportion is not significantly different when the protesters were in-party members or when the protesters were out-party members (16.7% vs. 22.0%, respectively; p = .23). Participants tended to disagree with the police actions, with a small but significant difference (b = −.061, 95% confidence interval [−.119, −.002]) in disagreement depending on whether participants were asked about the in-party or the out-party: disagreement was lower when the protesters were out-party members. The mean proposed fine toward the protesters is significantly lower in the in-party condition ($1,740.88) than in the out-party condition ($3,051.98), a difference of $1,311.10 (95% confidence interval [$601.55, $2,020.66]), or 42.96% of the out-party mean.
Basis The printed percentages, condition difference and fine means do not reproduce from the archived Study 3.1 data, which hold the 395 participants the text reports and from which the regression slopes in this section reproduce. The values above are computed on those participants (367 of them proposed a fine); b is the out-party minus in-party difference in disagreement (p = .042), and the fine interval is a Welch interval. The printed 57.12% is the ratio of the two printed means (1,713.75/3,000.21).
p. 494, Study 3.1
Published
The differences between the slopes of affective polarization for both conditions is substantively larger than zero (b = .37, 95% confidence interval [.20, .54]).
Correction
The differences between the slopes of affective polarization for both conditions is substantively larger than zero (b = .37, 95% confidence interval [.05, .69]).
Basis The estimate is the interaction of affective polarization and condition in the model behind figure 3 (SE = .162, p = .023). The printed interval spans about one standard error on either side of the estimate.
p. 494, Study 3.1
Published
We find no evidence that affective polarization predicts intent to discriminate against members of the opposition in this study.
Correction
Pooled across parties, we find no evidence that affective polarization predicts intent to discriminate against members of the opposition in this study. Among Republicans, however, more affectively polarized participants agreed more with the police action (b = .46, 95% confidence interval [.07, .86]) and proposed larger fines (log fine, b = 4.38, 95% confidence interval [.21, 8.54]) when the protesters were Democrats (see appendix 4.3 and 4.5).
Basis The agreement estimate is printed in SI Table A18. The fine estimate is printed in SI Table A20 under the heading Democrats, In-Party; see the correction to Table A20 (SI p. 14). Among Democrats, neither out-party relationship is significant (p = .09 for agreement and .52 for the fine).
p. 494, footnote 18
Published
18. We do see differences between Republicans and Democrats in this experiment (see appendix 4.6). However, because differences in response by party ID do not replicate in the next study and because the total N is small when the sample is split by party ID, we are not able to make strong conclusions about these results.
Correction
- We do see differences between Republicans and Democrats in this experiment (see appendix 4.3 and 4.5): among Republicans, affective polarization is positively related to agreeing with the police action and to the size of the fine when the protesters are Democrats. However, because differences in response by party ID do not replicate in the next study and because the total N is small when the sample is split by party ID, we are not able to make strong conclusions about these results.
Basis The appendix has no section 4.6. The Study 3.1 models by party are in sections 4.3 (Table A18) and 4.5 (Table A20).
p. 495, Study 3.2
Published
Regardless of whether the target was a member of the out-party or the in-party, participants adopted the normatively “correct” position and favored an investigation of possibly illegal activities (see fig. 4).
Correction
Regardless of whether the target was a member of the out-party or the in-party, participants adopted the normatively “correct” position and favored an investigation of possibly illegal activities (see fig. 5).
Basis Figure 5 shows support for the investigation; figure 4 shows the Study 3.1 fines.
p. 496, Study 3.2
Published
While there is in-party favoritism, it does not overcome the desire to investigate the wrong-doing of the middle man, with the mean always above the midpoint on the disagree-agree scaler in the in-party condition ($1,713.75) than in the out-party condition ($3,000.21), a difference of $1,286.46 (95% confidence interval [$553.72, $2,019.19]) or 57.12%.
Correction
While there is in-party favoritism, it does not overcome the desire to investigate the wrong-doing of the middle man, with the mean always above the midpoint on the disagree-agree scale in the in-party condition (M = .61).
Basis The fine figures belong to Study 3.1 (p. 494) and were repeated here. Study 3.2 asked no fine question.
p. 498, Discussion
Published
On the one hand, the task was no less behavioral than those that asked respondents to make scholarship allocation decisions (Iyengar and Westwood 2015; Munro, Weih, and Tsai 2010)
Correction
On the one hand, the task was no less behavioral than those that asked respondents to make scholarship allocation or admissions decisions (Iyengar and Westwood 2015; Munro, Lasane, and Leary 2010)
Basis Munro, Weih, and Tsai (2010) studies attributions about the behavior of political in-group and out-group members. The study of evaluations of college admissions applications is Munro, Lasane, and Leary (2010), which is in the reference list.
Supplementary information
SI contents (p. 1) and section headings 4.2 (p. 11), 4.4 (p. 13) and 5.2 (p. 17)
Published
4.2 Model used to generate Figure 4 [...] 4.4 Model used to generate Figure 5 [...] 5.2 Model used to generate Figure 6
Correction
4.2 Model used to generate Figure 3 [...] 4.4 Model used to generate Figure 4 [...] 5.2 Model used to generate Figure 5
Basis The article has five figures. Sections 4.2, 4.4 and 5.2 hold the models for figure 3 (agreement with the police action), figure 4 (Study 3.1 fine) and figure 5 (support for the investigation).
SI pp. 2-3, Tables A1-A4 (Study 1 demographics)
Published
Tables A1-A4 (percent): Table A1, “Non-White” 26.61, “White” 72.68; Table A2, “Male” 45.36, “Female” 53.93; Table A3, “HS or Less” 19.46, “College+” 39.82, “Some College” 40.00; Table A4, “Democrat” 53.93, “Republican” 45.36.
Correction
Corrected Tables A1-A4, computed on the 556 Study 1 participants.
Corrected Tables A1-A4: Study 1 participants (N = 556). Percentages of all participants and counts. Table Category Percent All A1: Distribution of Race Non-White 26.80 149 A1: Distribution of Race White 73.20 407 A2: Distribution of Gender Male 45.68 254 A2: Distribution of Gender Female 54.32 302 A3: Distribution of Education HS or Less 19.60 109 A3: Distribution of Education College+ 40.11 223 A3: Distribution of Education Some College 40.29 224 A4: Distribution of Party ID Democrat 54.32 302 A4: Distribution of Party ID Republican 45.68 254 Basis The printed percentages divide by the 560 rows of the archived file, 4 of which have no party or demographic data, so each table sums to about 99.3%. The counts are unchanged.
SI p. 13, Table A19
Published
Table A19 column headings: “In-Party”, “Out-Party”, “In-Party”, “Out-Party”.
Correction
Column headings: In-Party, In-Party, Out-Party, Out-Party. The estimates are unchanged.
Table A19 with corrected column headings: affective polarization and amount of fine (log). Columns 1-2: in-party condition, without and with controls; columns 3-4: out-party condition. Standard errors in parentheses; * p<0.05; ** p<0.01; * p<0.001. In-Party In-Party Out-Party Out-Party Affective Polarization −5.765* (1.424) −6.056* (1.411) 1.356 (1.762) 0.733 (1.566) Female 0.209 (0.521) −1.424** (0.523) White −1.237 (0.680) −0.895 (0.703) Income: 30-59k 1.788* (0.688) −0.475 (0.692) Income: 60-79k 1.491* (0.736) −1.223 (0.738) Income: 80k+ 1.243 (0.724) −0.581 (0.744) Education: College+ −1.304 (0.729) 0.894 (0.714) Education: Some College −0.319 (0.711) 0.942 (0.669) Republican 1.147* (0.543) 3.675* (0.560) Intercept 8.331* (1.057) 8.526* (1.372) 4.013** (1.343) 4.016** (1.518) Observations 192 192 175 175 Adjusted R2 0.075 0.122 −0.002 0.240 Basis In exp3_1_results.R, the columns are the in-party models without and with controls, then the out-party models without and with controls. The observations (192, 192, 175, 175) are the in-party and out-party fine samples.
SI p. 14, Table A20
Published
Table A20, columns headed “Republicans In-Party” (0.370, N = 82), “Republicans Out-Party” (−8.853, N = 110), “Democrats In-Party” (4.376, N = 82) and “Democrats Out-Party” (−1.468, N = 93).
Correction
Corrected Table A20, with each model under its own heading. The Republican out-party estimate is 4.376 and the Democrat in-party estimate is −8.853.
Corrected Table A20: affective polarization and amount of fine (log), by party of the participant, with each model under its own heading. Standard errors in parentheses; * p<0.05; ** p<0.01; * p<0.001. Republicans: In-Party Republicans: Out-Party Democrats: In-Party Democrats: Out-Party Affective Polarization 0.370 (2.643) 4.376* (2.126) −8.853* (1.588) −1.468 (2.284) Female 0.595 (0.794) −0.905 (0.674) −0.060 (0.668) −1.530 (0.784) White −3.270 (1.663) 1.676 (1.515) −1.153 (0.735) −1.294 (0.853) Income: 30-59k 1.776 (1.208) −0.212 (0.891) 1.322 (0.818) −0.241 (1.045) Income: 60-79k 1.078 (1.132) −0.767 (0.943) 0.994 (0.967) −1.490 (1.113) Income: 80k+ −0.391 (1.118) 0.647 (0.939) 2.141* (0.969) −1.481 (1.130) Education: College+ −1.636 (1.076) 1.687 (0.935) −1.264 (0.964) 0.517 (1.086) Education: Some College 0.182 (1.097) 1.336 (0.964) −0.855 (0.915) 1.238 (0.935) Intercept 7.363** (2.673) 1.293 (2.756) 10.891* (1.575) 6.146** (2.068) Observations 82 82 110 93 Adjusted R2 0.090 0.105 0.214 0.035 Basis exp3_1_results.R passes the models to stargazer in the order Republican in-party, Democrat in-party, Republican out-party, Democrat out-party, under headings in the order Republicans In-Party, Republicans Out-Party, Democrats In-Party, Democrats Out-Party. The observations confirm the order: 82 + 110 = 192 in-party and 82 + 93 = 175 out-party participants, as in Table A19.
SI p. 17, Table A25
Published
Table A25, note: “p<0.1; *p<0.05; *p<0.01”.
Correction
Corrected Table A25, with significance levels at the cutoffs of the other SI tables. The estimates are unchanged.
Corrected Table A25: affective polarization and support for investigation, with the significance levels used in the other SI tables. Standard errors in parentheses; * p<0.05; ** p<0.01; * p<0.001. In-Party In-Party Out-Party Out-Party Affective Polarization −0.407* (0.097) −0.402* (0.098) 0.083 (0.114) 0.047 (0.118) Male 0.021 (0.034) 0.031 (0.039) White −0.014 (0.038) −0.069 (0.048) Age −0.002 (0.001) 0.002 (0.001) Income: 30-59k 0.080 (0.044) −0.008 (0.050) Income: 60-79k −0.014 (0.045) −0.016 (0.054) Income: 80k+ 0.122* (0.054) 0.020 (0.057) College+ −0.024 (0.046) 0.087 (0.050) Some College −0.054 (0.042) −0.004 (0.049) Republican −0.013 (0.036) −0.007 (0.041) Intercept 0.897* (0.070) 0.953* (0.090) 0.675* (0.085) 0.631* (0.107) Observations 231 231 237 237 Adjusted R2 0.067 0.092 −0.002 0.005 SI p. 18, Table A26
Published
Table A26, column “Democrats Out-Party”: “Affective Polarization” −0.225 (0.199), “Observations” 104, identical to column “Republicans Out-Party”; note: “p<0.1; *p<0.05; *p<0.01”.
Correction
Corrected Table A26. The Democrat out-party estimate is .136 (95% confidence interval [−.14, .41], N = 133). Significance levels use the cutoffs of the other SI tables.
Corrected Table A26: affective polarization and support for investigation, by party of the participant; column 4 is the Democrat out-party model. Standard errors in parentheses; * p<0.05; ** p<0.01; * p<0.001. Republicans: In-Party Republicans: Out-Party Democrats: In-Party Democrats: Out-Party Affective Polarization −0.423* (0.167) −0.225 (0.199) −0.331** (0.125) 0.136 (0.142) Male −0.054 (0.059) 0.010 (0.060) 0.076 (0.042) 0.057 (0.050) White −0.024 (0.082) −0.241* (0.110) −0.015 (0.042) −0.036 (0.052) Age −0.0002 (0.002) 0.005** (0.002) −0.003* (0.001) 0.001 (0.002) Income: 30-59k 0.024 (0.079) −0.236** (0.080) 0.107* (0.053) 0.120 (0.064) Income: 60-79k 0.012 (0.081) −0.171* (0.081) −0.044 (0.054) 0.067 (0.070) Income: 80k+ 0.074 (0.084) −0.120 (0.088) 0.151* (0.075) 0.092 (0.072) College+ 0.056 (0.080) 0.036 (0.085) −0.054 (0.057) 0.127* (0.060) Some College −0.011 (0.074) −0.042 (0.082) −0.077 (0.051) 0.050 (0.060) Intercept 0.914* (0.151) 1.027* (0.189) 0.934* (0.111) 0.495* (0.131) Observations 102 104 129 133 Adjusted R2 0.025 0.104 0.159 0.026 Basis exp3_2_resuts.R calls stargazer(modelin1, modelout1, modelin2, modelout1), so the printed column 4 repeats the Republican out-party model. With the Democrat model in place, the out-party slopes (−.225 for Republicans, .136 for Democrats) do not differ significantly (p = .49 for the interaction in the out-party condition), as footnote 20 states.