Affect, Not Ideology: A Social Identity Perspective on Polarization

• p. 15 Table 2, 2008 block, row 'Positive – Negative', United States columns

  –1.31

• p. 15 Table 2, 2008 block, row 'Positive – Negative', United Kingdom columns

  –.87

• p. 15 Table 2, 2008, Positive – Negative, United States Out-party

  2.63

• p. 15 Table 2, 2008, Positive – Negative row

  –1.31 2.63 3.94*** –.87 1.80 2.67***

Judge's note: Read the upright image of Table 2. From the two rows above: US in-party 2.86 - .22 = 2.64 and out-party .62 - 1.93 = -1.31; UK in-party 1.93 - .13 = 1.80 and out-party .42 - 1.29 = -.87. The printed row has each pair the other way round. The printed In – Out values (3.94, 2.67) are positive and follow only from in = 2.63/1.80 and out = -1.31/-.87. The 1960 block has the same row in the header's order (.46 in, .27 out). U.S. – UK: in-party 2.64 - 1.80 = .84 (printed -.44), out-party -1.31 - (-.87) = -.44 (printed .84). No note or footnote defines the row differently.

Second check: another model agreed

Report a problem with this finding

This finding is

wrong right unclear or overstated already corrected (erratum, later version)

I am an author of this paper Send

• p. 15 Table 2, 1960 block, row 'Positive – Negative', Difference (U.S. – UK), Out-party

  .38***

• p. 15 Table 2, 1960, Positive – Negative, Difference (U.S.–UK) Out-party

  .13*** .38*** –.15***

Judge's note: Read the upright image of Table 2. US out-party .27, UK out-party -.01, difference .28. The same result follows from the two rows above in that column (.23 - (-.05) = .28). The neighbouring cells in the row reproduce (.46 - .33 = .13; .19 - .34 = -.15), and the US and UK cells are themselves consistent (.35 - .08 = .27; .12 - .13 = -.01). Rounding of unrounded inputs cannot move the result by .10.

Second check: another model agreed

Report a problem with this finding

This finding is

wrong right unclear or overstated already corrected (erratum, later version)

I am an author of this paper Send

• p. 15 Table 2, 1960 Intelligent, Difference In-party / Out-party

  –.07**

• p. 15 Table 2, 1960 Selfish, Difference In-party / Out-party

  .08***

• p. 15 Table 2, 2008 Intelligent, Difference In-party / Out-party

  –.24***

• p. 15 Table 2, 2008 Selfish and Closed-minded, Difference Out-party

  –.19***

Judge's note: Read page 16 and the upright image of Table 2. The sentence defines in-party bias as a comparison with the opposition, and the previous page and the top of p. 16 use the In – Out values for the same quantity ("0.12 in the United States and 0.22 in Britain"). In Table 2, US positive traits: 1960 in .47, out .35, In – Out .12; 2008 in 2.86, out .62, In – Out 2.24. Footnote 12 already acknowledges that the two years used different numbers of traits, so that part is explained. This cluster also carries a second finding about the signs in the Difference (U.S. – UK) columns of the single-trait rows; that defect is real and is judged under C020.

Second check: another model agreed

Report a problem with this finding

This finding is

wrong right unclear or overstated already corrected (erratum, later version)

I am an author of this paper Send

• p. 15 Table 2, 1960 Intelligent, Difference (U.S. – UK), In-party and Out-party

  –.07**

• p. 15 Table 2, 2008 Intelligent, Difference (U.S. – UK), In-party

  –.24***

• p. 15 Table 2, 2008 Selfish, Difference, Out-party

  –.19***

Judge's note: Read the upright image of Table 2; the minus signs are printed. 1960 Intelligent: .33 - .26 = +.07 (printed -.07), .27 - .09 = +.18 (printed -.17). 1960 Selfish: .04 - .02 = +.02 (printed -.02), .21 - .29 = -.08 (printed .08). 2008 Intelligent: .62 - .38 = +.24 (printed -.24), .14 - .10 = +.04 (printed -.04). 2008 Selfish: .47 - .28 = +.19 (printed -.19). 2008 Closed-minded: .08 - .05 = +.03 (printed -.03), .62 - .47 = +.15 (printed -.14). By contrast the In – Out column follows the header (.06 - .17 = -.11 printed -.11; -.43 - (-.25) = -.18 printed -.18) and so do the average rows (.47 - .34 = .13 printed .13; 2.86 - 1.93 = .93 printed .93). No note explains a different convention for these cells.

Second check: another model agreed

Report a problem with this finding

This finding is

wrong right unclear or overstated already corrected (erratum, later version)

I am an author of this paper Send

• p. 16 Trait Ratings, paragraph beginning 'The dramatic strengthening'

  in-party bias increased to 0.41 among Conservative identifiers (an increase of 200 percent)

• p. 16 top of page, 1960 results

  for Conservatives, the in-party bias was 0.25

• p. 16 top of page, 1960 values

  for Conservatives, the in-party bias was 0.25; for Labor, it was 0.18

• p. 16

  (an increase of 200 percent)

Only the authors can say what is right here.

Judge's note: Read page 16 and Table 2. (0.41 - 0.25)/0.25 = 0.64; a 200 percent increase from 0.25 would be 0.75. The 2008 by-party values (0.41, 0.22) are not in Table 2, whose 2008 UK In – Out for positive traits is 1.52 on a count of five traits, so 0.41 appears to be on a different (possibly per-trait) scale than the 1960 figure of 0.25 on two traits. If 0.25 were halved to a per-trait 0.125, the rise to 0.41 would be about 228 percent, which may be what was meant, but the paper does not say so. As printed the percentage does not follow from the stated numbers; the authors would need to say which scale and baseline were used.

Second check: another model agreed

Report a problem with this finding

This finding is

wrong right unclear or overstated already corrected (erratum, later version)

I am an author of this paper Send

• p. 18 last paragraph of main text

  b = .20, p < .05 for Democrats in 1988 and 2004; b = .12, p < .05 and b = .18, p < .05 for Republicans in 1988 and 2004

• p. 19 Table 3, row 'Economic attitudes', 2004 columns

  .19*

• p. 19 Table 3, Economic attitudes row

  .20* .12* .19* .19*

• p. 19 Table 3, Economic attitudes row

  Economic attitudes .20* .12* .19* .19*

Only the authors can say what is right here.

Judge's note: Read pages 18 and 19. Table 3, Economic attitudes: Democrats 1988 .20, Republicans 1988 .12, Democrats 2004 .19, Republicans 2004 .19. The text gives .20 for Democrats in both years and .18 for Republicans in 2004, so both 2004 values differ from the table by .01. The text says the results are those "presented in table 3", and the strong-identifier coefficients quoted in the next sentence (.18, .26, .17, .22) match the table exactly, so the same model is meant. Which of text or table is right only the authors can say. The differences are small and do not change significance markers.

Second check: another model agreed

Report a problem with this finding

This finding is

wrong right unclear or overstated already corrected (erratum, later version)

I am an author of this paper Send

• p. 26 References, Hetherington entry

  Hetherington, Marc. 2002. “Resurgent Mass Partisanship: The Role of Elite Polarization.”

• p. 26 References, McCarty, Poole, and Rosenthal entry

  Congress.” American Political Science Review 95:673–88.

• p. 1 Introduction

  Hetherington 2002

• p. 26 Reference list

  Hetherington, Marc. 2002. “Resurgent Mass Partisanship: The Role of Elite Polarization.” American Political Science Review 95(3):619–31.

Judge's note: Read the page 26 reference list from the image. McCarty, Poole and Rosenthal 2001 is American Political Science Review 95:673–88; Poole, Rosenthal and Koford 1991 and Richardson 1991 are both American Political Science Review volume 85. On the list's own evidence volume 95 is 2001, and issue 3, pages 619–31, comes before pages 673–88 of the same volume. The year 2002 on the Hetherington entry, and the matching in-text citation on page 1 (confirmed in the text layer, page image not opened), therefore look off by one. The inconsistency is internal to the list; that the year rather than the volume is the wrong element is inferred from the other three entries.

Second check: another model agreed

Report a problem with this finding

This finding is

wrong right unclear or overstated already corrected (erratum, later version)

I am an author of this paper Send

• p. 7 Measures, third paragraph

  The thermometer ratings were also part of the 2004 Issues Survey

• p. 6 Data Sets

  2004 Blair Center Election Study:

• p. 5 Data Sets

  a 2004 Blair Center Election Study

• p. 5 Data Sets

  We take advantage of six different survey data sets

Only the authors can say what is right here.

Judge's note: Data Sets (pp. 5-6) lists ANES, YouGov 2008, Almond and Verba, 2004 Blair Center Election Study, the 2010 eleven-nation study and AP-Yahoo! 2008. '2004 Issues Survey' occurs once (p. 7) and is not defined. The 2004 analysis on p. 20 uses thermometer differences from the Blair Center study ('the Blair Survey'), so that study is probably meant, but the paper does not say the two names are the same. The reviewer's side point that a seventh data set is described is not a defect: the Wisconsin Advertising Project is not a survey.

Second check: another model agreed

Report a problem with this finding

This finding is

wrong right unclear or overstated already corrected (erratum, later version)

I am an author of this paper Send

• p. 14 Trait Ratings, first paragraph

  are common to the 1960 and 2010 questionnaires

• p. 14 Trait Ratings, first sentence

  The trait ratings data are also available for 1960 and 2008

• p. 16 paragraph beginning 'Between 1960 and 2008'

  porters in 2010.

• p. 7 Stereotypes of Party Supporters

  This indicator is limited to the YouGov 2008 study

Judge's note: Read pages 14 and 16 and the upright image of Table 2. The sentence before the first quote says "available for 1960 and 2008", Table 2 is titled "1960, 2008" with source "YouGov/Polimetrix in 2008", the p. 16 paragraph opens "Between 1960 and 2008", footnote 12 says "The 2008 survey", and the Measures section (p. 7) says "This indicator is limited to the YouGov 2008 study". The 2010 survey is described only for the marriage question. No statement in the paper puts a trait battery in 2010.

Second check: another model agreed

Report a problem with this finding

This finding is

wrong right unclear or overstated already corrected (erratum, later version)

I am an author of this paper Send

• p. 10 paragraph below Figure 2

  while Democrats’ evaluations of conservatives remain in the 45–50 range over the entire period, Republicans’ ratings of lib-

• p. 10 Figure 2 caption and legend

  Figure 2. Thermometer Ratings of Ideological Identifiers.

Only the authors can say what is right here.

Judge's note: On the p. 10 image the figure is titled 'Thermometer Ratings of Ideological Identifiers', the y-axis is 'Mean Thermometer Rating of Ideological Groups', and the series in the 45-50 range is 'Liberal - Out Group' (liberals rating conservatives), with 'Conservative - Out Group' at about 41-44 below it. The text attributes these two series to Democrats and Republicans. Table A2 in the Online Appendix, on which the figure is based, is not among the files, so it cannot be ruled out that the raters were in fact partisans and the legend is the mislabelled part; either the text or the figure labels the raters wrongly.

Second check: another model agreed

Report a problem with this finding

This finding is

wrong right unclear or overstated already corrected (erratum, later version)

I am an author of this paper Send

• p. 5 Research Design, second paragraph

  including gender, race, and religion, as bases for affec-

• p. 11 paragraph citing figure 3

  As shown in figure 3, race and religion

• p. 12 Figure 3 legend

  Race (Whites/Blacks)

• p. 11 Figure 3 text

  race and religion are weaker forms of identity than partisanship

Only the authors can say what is right here.

Judge's note: Figure 3 (p. 12) has series for Democrats/Republicans, Democratic/Republican Party, Race (Whites/Blacks) and Religion (Catholic/Protestant) only; the text on p. 11 says 'race and religion are weaker forms of identity than partisanship'. Searched the text for 'gender': it appears otherwise only as a matching variable and a regression control. No gender comparison and no pointer to one in the appendix.

Second check: another model agreed

Report a problem with this finding

This finding is

wrong right unclear or overstated already corrected (erratum, later version)

I am an author of this paper Send

• p. 22 paragraph beginning 'We implemented a hierarchical linear model'

  from battleground-state residence, time of interview, gender, race, residence in the South, and education

• p. 22 Table 5, predictor list

  Strong identifier

• p. 22 footnote 27

  the political interest variable was used from the November 2007 wave

• p. 22 Second test of campaign effects

  battleground-state residence, time of interview, gender, race, residence in the South, and education

Only the authors can say what is right here.

Judge's note: Read page 22 from the page image. Table 5's rows are Battleground, Wave, Wave × Battleground, Republican, Strong identifier, Age, Female, Race: white, and three education rows. Its note says only "Intercept, random effects for DMA, and individual respondents not reported", so an unreported South or political-interest coefficient is not covered. Footnote 27 does mention age, but nothing in the text or footnote mentions party or strength of identification for this model. No explanation was found elsewhere in the article. Only the authors can say which specification was actually estimated.

Second check: another model agreed

Report a problem with this finding

This finding is

wrong right unclear or overstated already corrected (erratum, later version)

I am an author of this paper Send