Social Pressure and Voter Turnout: Evidence from a Large-Scale Field Experiment

• p. 6 Study Population, left column

  The 180,002 households were then divided into 10,000

• p. 6 Table 1, N row

  99,999

• p. 6 Table 1, N row

  20,002

• p. 2 Introduction, last paragraph (left column)

  approximately 80,000 households

Judge's note: Only the cell arithmetic is confirmed. The paper does not explain where the two extra households went or why group sizes are off by one or two; it does hedge with "approximately 20,000 households" and footnote 8's "approximately 180,000 households", so this is a small unexplained imprecision. The other part of the cluster is not a defect: "approximately 80,000 households were sent one of four mailings" on page 2 refers to the treated households (80,003 in Table 1), not the whole 180,002 sample. Checked in the page images of pages 2, 5 and 6.

Second check: another model agreed

• p. 9 Discussion, third paragraph

  The present study is the first since Gosnell (1927) to examine whether these social pressures to vote can be used to encourage people to cast ballots.

• p. 9 Discussion, second paragraph

  but none have introduced social surveil- lance

• p. 5 Prior Experimental Research on Shaming and Voting

  These two recent experiments, which together had treatment groups ap- proximately 10% of the size of the current study, found borderline statistically significant evidence that social pressure increases turnout.

Judge's note: Both passages read in the page images (PDF pages 5 and 9). Read literally the Discussion sentences overlook the authors' own pilots, which the paper itself calls "prior studies". The likely intent is that the pilots are part of the same research programme, or that only published studies are counted (the Discussion paragraph refers to "previously published experimental studies" and the pilots are a conference paper). A loose wording point, not a factual error about the data.

Second check: another model agreed

• p. 9 Discussion, first paragraph

  8% range have never before been observed, even for

• p. 10 Table 4, 1999 New Haven, Civic duty message, Turnout rate row

  38.9

• p. 10 Table 4, 1999 New Haven, Civic duty message, Turnout rate row (Four and Six mailings)

  44.6

• p. 9 Discussion, first paragraph

  Results in the 5% to 8% range have never before been observed, even for a nine-piece mail program.

Only the authors can say what is right here.

Judge's note: Checked the row in the page image and crop: 38.9 (None), 39.2 (Two), 44.6 (Four), 45.8 (Six), 42.2 (Eight), with N of 278 and 295 for the four- and six-mailing cells. The two differences fall inside the range the sentence says was never observed. The paragraph opens with "Consider the results for a single mailing" and the authors may mean statistically reliable or per-piece effects (the cells are small, and the Eight cell is only 3.3 points above control), but the sentence carries no such qualification and explicitly extends to multi-piece programs. No explanation found elsewhere in the article. A wording overstatement rather than a wrong number; the authors can say what qualification was meant.

Second check: another model agreed

• p. 9 Discussion, first paragraph

  though one mail piece more commonly produces a re-

• p. 10 Table 4, column 'One'

  90.6

• p. 11 Table 4 continued, Queens County rows, column 'One'

  54.2

• p. 9 Discussion, first paragraph

  are rare but not unprecedented, though one mail piece more commonly produces a result less than 1 percentage point

Only the authors can say what is right here.

Judge's note: Read the One column on PDF pages 10 and 11 from the page images: 41.7, 42.0, 44.2 against 42.2 (1998 New Haven); 90.6 against 89.0 (Pennsylvania prime Republicans); 37.3 against 35.4 (Los Angeles Asian Americans); 53.7, 54.2, 52.1 against 52.2 (Queens County). Differences are -0.5, -0.2, +2.0, +1.6, +1.9, +1.5, +2.0, -0.1. The sentence is introduced as a reading of the single-mailing results in Table 4, and on the raw cells "rare" and "more commonly less than 1" do not hold. The authors may be thinking of per-piece effects across the multi-mailing programs or of precision-weighted estimates, which the text does not say; hence a question for the authors rather than a correction.

Second check: another model agreed

• p. 4 The Calculus of Voting, first paragraph

  1957; Riker and Ordershook 1968; see also Aldrich

• p. 4 The Calculus of Voting, third paragraph

  include these psychic benefits (Riker and Ordeshook

• p. 16 References

  Riker, William H., and Peter C. Ordershook. 1968.

• p. 2 First paragraph

  1954) and formal (Downs 1957; Riker and Ordeshook

Judge's note: Both spellings are printed in the page images of pages 2, 4 and 16. The inconsistency is internal to the paper. Which spelling is correct (Ordeshook) rests on outside knowledge, but one of the two must be wrong.

Second check: none (judge only)

• p. 3 Left column, second and third paragraphs

  studied to date. Third, our experimental results also

• p. 3 Introduction, 'Our study makes several important contributions' paragraphs

  First, we provide strong statistical evidence

• p. 3 Introduction, next paragraph

  Third, our experimental results also

• p. 3 Introduction, left column, contributions paragraph

  by far the most cost-effective voter mobilization tactic studied to date. Third, our experimental results also address

Judge's note: Read the left column of PDF page 3 in the page image. The only "second" on the page is "The second aspect of norms" in the right column, which belongs to a different list. The enumeration skips its second ordinal.

Second check: none (judge only)

• p. 3 Left column, second paragraph

  see Gerber and Rugers 2007 for an application to

• p. 16 References

  Gerber, Alan S., and Todd Rogers. 2007.

• p. 3 Introduction, left column

  Gerber and Rugers 2007

• p. 3

  see Gerber and Rugers 2007 for an application

Judge's note: The page 3 image prints "Rugers"; the page 16 image prints "Todd Rogers". There is no other Gerber 2007 entry, so the same work is meant. Graded as a misspelling rather than a wrong citation because the work is still identifiable.

Second check: none (judge only)

• p. 4 Right column, paragraph before equation (2)

  But what is D? The convention model in equation

• p. 4 The Calculus of Voting

  The convention model in equation

Judge's note: Seen in the page 4 image, right column. Wrong word form.

Second check: none (judge only)

• p. 5 Footnote 6

  Those with a high probability of voting absentee voting were de-

Judge's note: Seen in footnote 6 in the page 5 image.

Second check: none (judge only)

• p. 5 Study Population, last paragraph on page

  had a greater than a 60% probability of choosing the

Judge's note: Seen in the page 5 image, right column. The parallel clause just before reads "had over a 60% probability".

Second check: none (judge only)

• p. 7 Treatments, 'Self' paragraph

  primary election “we intend to mail an updated chart,”

• p. 14 Appendix A, Self mailing

  We intend to mail you an

• p. 7 Results, description of Self mailing

  we intend to mail an updated chart,”

• p. 7 Treatments, description of the "Self" mailing

  we intend to mail an updated chart

Judge's note: Page 7 image shows the quotation in the Self paragraph. Page 14 image (Self mailing) has "We intend to mail you an updated chart"; page 15 image (Neighbors mailing) has "we intend to mail an updated chart". The quotation is not verbatim for the Self letter (one word, "you", omitted). Meaning unchanged.

Second check: none (judge only)

• p. 7 Results, second paragraph

  Nickerson 2006b)

• p. 16 References

  Nickerson, David W. 2006b.

• p. 7 Results, paragraph on magnitude of effects

  Nickerson 2006b

Judge's note: Page 7 and page 16 images show the "b" suffix. The only other Nickerson entry is "Forthcoming", and a text search finds no "2006a". The citation still resolves to one entry, so the suffix is a harmless leftover.

Second check: none (judge only)

• p. 7 Results, last paragraph before cost discussion

  on the order of 3 percentage-points or less

Judge's note: Seen in the page 7 image, right column. The same page has "1.8 percentage points" and "4.9 percentage-point increase". A cosmetic inconsistency with the paper's own usage.

Second check: none (judge only)

• p. 8 Equation (8)

  + λ1V1i + λ1V1i + · · · + λ5V5i + ui.

• p. 8 Equation (8)

  λ1 V1i + λ1 V1i + · · · + λ5 V5i

• p. 8 Equation (8)

  + λ1V1i + λ1V1i + · · · + λ5V5i + ui

Judge's note: Seen in the page image and the crop: the term λ1V1i is printed twice before the ellipsis, and the series ends at λ5V5i. The surrounding text says the model controls for voting in five recent elections, so the terms should run 1 to 5. A typographical slip in the notation with an evident fix; it does not affect any result.

Second check: none (judge only)

• p. 8 Right column, first paragraph

  are significant at p < .0001.

• p. 8 Table 3 note

  p < .001.

Judge's note: Both statements are on PDF page 8 and are as quoted. They are compatible (a value below .0001 is also below .001), so this is only a difference in the stated threshold between text and table note, not a contradiction. Either form could be legitimate.

Second check: none (judge only)

• p. 10 Table 4, first row under 1998 New Haven, Civic duty message

  Turnout ratev

Judge's note: The stray "v" is printed in the page image, so it is not an extraction artefact. The table notes define only a, b, c and 1 to 5, so it is not a footnote marker.

Second check: none (judge only)

• p. 11 Table 4 note a

  The control group is the same both New Haven experiments in 1998.

• p. 11 Table 4 note b

  The control group is the same both New Haven experiments in 1999.

• p. 10 Table 4, 1998 New Haven block

  Neighborhood solidarity message

Judge's note: Notes read in the crop of PDF page 11; both lack a preposition. On PDF page 10 the 1998 New Haven block has three arms (civic duty, neighborhood solidarity, close turnout), each with control 42.2 and N = 11,596, so "both" fits only the two-arm 1999 block. Cosmetic; the meaning is clear.

Second check: none (judge only)

• p. 11 Table 4 note 1

  Gerber and Green 2000, 1

• p. 11 Table 4 footnote 1

  Gerber and Green 2000, 1 and 1(b)

Judge's note: Read in the crop. The "1" is not a stray character: the sentence parses as "1(a) from Gerber and Green 2000; 1 and 1(b) from Gerber, Green, and Green 2003 and Gerber 2004", which matches the table, where the 1999 New Jersey, Connecticut and 2002 Pennsylvania blocks carry marker 1 and 1999 New Haven carries 1b. The defect is therefore only weak punctuation (a comma where a semicolon is needed, a semicolon before "and Gerber 2004"), a missing article in "from experiment", and marker style. Less serious than a garbled citation.

Second check: none (judge only)

• p. 16 References

  Knack, Stephen, and Martha E. Kropf. 1998.

Only the authors can say what is right here.

Judge's note: The entry is in the page image of PDF page 16. A search of the full article text finds "Kropf" only in the reference list; the text cites Knack (1992) and Knack 2001 only. PDF pages 12 to 15 are OCR text, so a citation there could in principle be missed by the search, but those pages are appendix material. The authors should either cite the work or remove the entry.

Second check: none (judge only)

• p. 16 References, Gerber and Rogers 2007

  the positive. Yale University manuscript.

Judge's note: Checked in the crop at high resolution: an opening quotation mark precedes "The Effect" and there is no closing mark after "the positive." Other article titles in the list are closed. Cosmetic.

Second check: none (judge only)

• p. 16 References

  Lerner, Jennifer S., and Philip E. Tetlock. 1999.

• p. 16 References

  Ringold, Debra Jones. 2002.

Judge's note: Order confirmed in the page images: on PDF page 16 the sequence is Ledyard, Lerner, Lepper, Lijphart and later Ramirez, Ringold, Riker; Rind opens PDF page 17. The rest of the list is alphabetical. Cosmetic only.

Second check: none (judge only)

• p. 16 References

  Hoffman, Elizabeth, Kevin A. McCabe, Vernon L. Smith. 1996.

Judge's note: Checked the crop and full page 16. The entry has no "and" before "Vernon L. Smith". Other three-author entries on the page do (e.g. "Green, Donald P., Alan S. Gerber, and David W. Nickerson. 2003."; "Kallgren, Carl A., Raymond R. Reno, and Robert B. Cialdini. 2000."), and the in-text citation on page 9 of the text reads "Hoffman, McCabe, and Smith 1996". Cosmetic.

Second check: none (judge only)

• p. 16 References, Asch 1958

  Eleanor E. Macoby

Only the authors can say what is right here.

Judge's note: Rests on outside knowledge: that the editor of Readings in Social Psychology (3rd ed.) is Eleanor E. Maccoby, with a double c. The page image does print "ed. Eleanor E. Macoby, Theodore M. Newcomb, and Eugene L. Hartley". The name appears nowhere else in the paper, so the correct spelling cannot be checked from the files.

Second check: none (judge only)

• p. 16 References, Lepper, Greene and Nisbett

  Nisbett. 1973 .

Judge's note: Visible in the crop and on full page 16: a clear gap between "1973" and the period, unlike every other entry on the page (e.g. "Tetlock. 1999.", "Lijphart, Arend. 1997."). The gap is larger than normal justification spacing and sits before punctuation. Cosmetic.

Second check: none (judge only)

• p. 16 References, Deci

  Psychology. 18 (April): 105–15.

Judge's note: Checked the crop and full page 16. The Deci entry has a period after the italic journal title before "18 (April)". The Lepper entry for the same journal reads "Journal of Personality and Social Psychology 28 (October): 129–37." with no period, as do the other journal entries (a few Annals entries use a comma, none a period). Cosmetic inconsistency.

Second check: none (judge only)

• p. 16 References, Nickerson Forthcoming

  American Political Science Review 102 (February)

Judge's note: Checked the crop and full page 16. In this entry the journal title is set upright, whereas the same title is italic in the Lijphart, Palfrey and Rosenthal, and Plutzer entries on the same page. The entry ends "102 (February)" with no period, while all other entries end with one. The missing page range is expected for a forthcoming article and is not a defect. Cosmetic.

Second check: none (judge only)

• p. 17 References, Trivedi 2005

  “The Effect of Identify-Based GOTV Direct

Only the authors can say what is right here.

Judge's note: Rests on outside knowledge: that the published title of Trivedi's 2005 Annals article uses "Identity-Based". The page 17 image does print "Identify-Based GOTV Direct Mail Appeals". "Identify-Based" is not a meaningful compound and the article concerns ethnic-identity appeals to Indian Americans, so a misprint is very likely, but the title is not quoted anywhere else in the files to check against.

Second check: none (judge only)