re-analysis of the Commission on Community Interrelations experiments

Answering an Antisemitic Remark, 1945 to 1949

A project of the Lab for the Global Study of Antisemitism, Anne Tanenbaum Centre for Jewish Studies, University of Toronto. Ron Levi and Sol Goldberg.

X in

6 publications · 1,130 checks on the transcription · 47 data tables · 3 attitude experiments, 1946 · 7 preference rounds, 1945–1946 · 1 replication, 1949

Selltiz, Citron, Harding, Rosahn and Wormser (1950): the preference series

Levi & Goldberg, transcribed 2026-10-14 from the page images of

Selltiz, C., A. F. Citron, J. Harding, O. Rosahn and M. H. Wormser. 1950. "The Acceptability of Answers to Anti-Semitic Remarks." International Journal of Opinion and Attitude Research 4: 353-390. U of T Scan and Deliver copy. PDF: (held by the authors)

Page numbers are the journal's. Every number was read from the page image; the OCR text served only to find pages. No cell was illegible. The authors published aggregates only.

Check before use: python3 scripts/verify_selltiz.py runs 90 checks: printed totals and percentages against the choice counts, the running text's counts and significance statements, and the two tables the article shares with Citron's dissertation (Chapter Two) against the pref_* transcription. All pass. Six printed figures cannot be reproduced and are pinned as known discrepancies (section 4).

1. The rounds

The article is the second in the series that begins with Citron, Chein and Harding (1950). It reports seven rounds of short plays, each with an anti-Semitic remark and one or more answers, followed by a question on which answer was best. Replies of Jewish subjects were dropped.

Round	Place	Incident	Answers compared	Preference measure	Choosers
I	New York, organized groups	four incidents fitted to each group	no answer, mere negation, persuasive, aggressive	check ballot	43
II	New York, theological students and a YMCA group	bus	2 manners x situational logic or democratic ideology, plus no answer; five versions	check ballot	48
III	New York, off-the-street	bus	three versions, no answer always third	interview and check ballot	107
IV	New York, church, PTA, YWCA, union auxiliary	veterans' center	3 manners x democratic ideology, individual differences, mental hygiene; three versions	interview (discussion ballot in two groups)	47
V	New York, union, church and Y, off-the-street	employment bureau	2 manners x American tradition or individual differences; four presentations A-D	interview and discussion ballot	532
VI	Peoria, Illinois, church, YWCA, college class	employment bureau	as Round V	face sheet	67, plus 15 social workers
VII	Times Square and Newark, off-the-street	employment bureau with a game between versions	as Round V; answerer Stevenson or Levy	answer sheet	503

The only date printed for any round is 26 November 1945, on a Round I discussion record (p. 358).

Crosswalk to the dissertation. Round V is the dissertation's Round IV: Table 5 equals Table Two cell for cell, and Table 6 equals Table Eight. Round VI is the Peoria footnote of the dissertation (p. 41: 67 subjects, 78 per cent low manner; Table 7 gives 52 of 67). Round VII is the dissertation's Rounds V and VI and the Experiment III of Citron, Chein and Harding (1950): the same 520 subjects (428 New York, 92 Newark). The dissertation names three pilot rounds and reports no data from them; the article has four, and neither source says how they correspond.

2. Files

File	Rows	One row per
`sel1950_choices.csv`	102	round x audience x presentation x position: how many chose that answer
`sel1950_tables.csv`	80	printed total or percentage in Tables 1-8 that is not a choice count
`sel1950_text_numbers.csv`	119	number reported only in the running text
`sel1950_design.csv`	36	design fact, with the crosswalk to the dissertation
`sel1950_instruments.csv`	32	attitude item, preference question, interview question or ballot statement
`sel1950_scripts.csv`	65	line of the bus (Appendix A) and veterans' bureau (Appendix B) scripts

`sel1950_choices.csv`

round (I to VII), table (source table, or text), page, audience (as the table labels it), instrument (check_ballot, interview, discussion_ballot, interview_and_check_ballot, interview_and_discussion_ballot, face_sheet, mc_question_answer_sheet), presentation (A to D in Rounds V to VII, as lettered in Tables 5, 7 and 8), position (order in which the audience saw the answer; blank for Round I, whose order varied unsystematically), content (situational_logic, democratic_ideology, american_tradition, individual_differences, mental_hygiene, no_answer; Round I persuasive_reply, aggressive_reply, mere_negation), manner (low, medium, high, blank for no answer and Round I), answer (the key used in answers.csv, for the four Round V to VII answers), n_choosing, n_cases (choosers in that audience and presentation), notes.

In Rounds V to VII each audience saw two answers that differ in both manner and content (A: low American tradition, then high individual differences; B: the reverse order; C: low individual differences, then high American tradition; D: the reverse order). A choice for one is a choice against the other.

Other files

sel1950_tables.csv follows pref_tables.csv: table, page, round, row_label, column_label, value (as printed), unit, is_printed_total, notes. Tables 3 and 6 are held here in full because they are percentages, not choice counts.

sel1950_text_numbers.csv: id, round, page, statistic, value (as printed, which is sometimes a list such as "383, 149"), unit, refers_to, quote, notes.

3. What changes for the analyses

The dissertation printed the Experiment III preference vote only as calm against indignant totals (Table Twelve). Table 8 here prints it by presentation and answerer, so the content of the answer can be separated from its manner and position in the same audiences whose attitude change the 1950 article reports. Script 13_same_audience.py uses it. In those 503 people, content had no detectable effect on preference (logit 0.00, SE 0.11), while calm American values was followed by less prejudice than calm individual differences (d = 0.37, p = .002).

4. Known discrepancies

Each was checked on the page image. The verifier prints all six.

Table 2: no answer in the five groups that saw low democratic ideology first is printed 38 per cent; 20 of 55 is 36.4, and the printed column sums to 101.
p. 374: the text gives 155 Round V interviewees; the group counts on the same page (49, 53, 56) sum to 158.
p. 383: low manner presented first received "63 per cent" of New York church and Y choices; Table 5 gives 40 of 73, 55 per cent.
Round VII against dissertation Table Twelve: Stevenson calm choices 157 of 293 here, 155 of 293 there; Levy calm choices 141 of 210 here, 144 of 212 there. The article's Newark column sums to 89, against 91 Newark people in the dissertation.

5. Cautions

Rounds I to IV varied incident, answers and order from group to group; the authors call their conclusions tentative. Round IV's latin square broke when one church group saw a series identical to another's.
Round V pools interviews and discussion ballots after finding one significant difference between them in twelve cells.
In Round VII a four-minute game separated the two versions, and 78 per cent of choices went to the second answer (72 per cent in Round V).
Round VI's actors were amateurs, and each of the four Peoria groups saw one presentation, so presentation and group are the same.