Stereotype Threat and the Intellectual Test Performance of African Americans
Journal of Personality and Social Psychology, 1995 · paper · Psychology
Introduces stereotype threat, the risk of confirming a negative group stereotype, and tests it in four studies. Black participants underperformed relative to White participants on a difficult verbal test when it was framed as diagnostic of ability, but not otherwise.
Checked: PDF from turtletalk.blog, fetched 5 Oct 2026. Page numbers are those of that PDF.
5 findings
Reference or label Study 1 accuracy means name a condition that is not in the design: "Black non-diagnostic-only, and Black diag-nostic-challenge conditions were, .519, .518, .561, .546, and .490"; the last cell should be Black non-diagnostic-challenge.
- p. 5 Study 1 Results, Accuracy
Black non-diagnostic-only, and Black diag-
- p. 5 Accuracy, list of adjusted means
Black non-diagnostic-only, and Black diag- nostic-challenge conditions
- p. 3 Study 1 design
the non-diagnostic-challenge condition
- p. 5 Accuracy
Black non-diagnostic-only, and Black diag-nostic-challenge conditions were
Judge's note: Page 5 image: the list reads "White diagnostic, White non-diagnostic-only, White non-diag-nostic-challenge, Black non-diagnostic-only, and Black diag-nostic-challenge". The design (page 3) has diagnostic, non-diagnostic-only and non-diagnostic-challenge conditions, and the Black diagnostic mean (.420) is given in the preceding sentence, so the fifth listed cell can only be Black non-diagnostic-challenge. The "non-" is absent in print, not lost at the line break. The values are unaffected; the label is wrong.
Second check: another model agreed
Report a problem with this finding
Reference or label "(Sarason, 1980)" is cited on PDF pages 4 and 5, but the reference list has only "Sarason, I. G. (1972). Experimental approaches to test anxiety"; there is no 1980 entry.
- p. 4 Study 1, Dependent Measures
used in test anxiety research (Sarason, 1980)
- p. 5 Study 1 Discussion
(Sarason, 1980). Thus to further test the reliability
- p. 15 References
Sarason, I. G. (1972). Experimental approaches to test anxiety
- p. 4 Dependent Measures, cognitive interference
(Sarason, 1980)
Only the authors can say what is right here.
Judge's note: Page 4 crop shows "used in test anxiety research (Sarason, 1980)"; page 5 image shows "(Sarason, 1980). Thus to further test". The page 15 crop and a text search of the reference list show a single Sarason entry, dated 1972, which is cited separately as "Sarason, 1972" on PDF pages 3 and 13. Either the 1980 entry is missing or the year is wrong in both places; only the authors can say which. The side remark in one reviewer note that Geen (1991) is listed but never cited is correct (the text cites only Geen, 1985) but is a separate point.
Second check: another model agreed
Report a problem with this finding
Reference or label Reference list entry "Geen, R. G. (1991). Social motivation. Annual Review of Psychology, 42, 377-399." is not cited anywhere in the text; every in-text citation is to Geen, 1985.
- p. 15 References
Geen. R. G. ( 1991). Social motivation. Annual Review of Psychology,
- p. 15 References
Geen. R. G. ( 1991). Social motivation. Annual Review of Psychology
- p. 15 References
Geen, R. G. (1991). Social motivation. Annual Review of Psychology
Only the authors can say what is right here.
Judge's note: Page 15 image shows both Geen entries (1985 and 1991). A search of the article text for "Geen" finds four in-text citations, all "Geen, 1985", and no citation of a 1991 Geen work. Either the entry is an orphan or one of the 1985 citations was meant to be 1991; only the authors can say.
Second check: another model agreed
Report a problem with this finding
Wording Study 3 states "Twelve of these fragments", "the 12 most common constructs" and "a fairly high number (12) of target fragments", but "The complete list was as follows" gives eleven: RACE, LAZY, BLACK, POOR, CLASS, BROTHER, WHITE, MINORITY, WELFARE, COLOR, TOKEN.
- p. 7 Study 3, Measures, Stereotype activation
Twelve of these fragments had as one possible solution a word reflecting
- p. 7 Study 3, Measures, Stereotype activation, end of page
The complete list was as follows:
- p. 8 Study 3, Measures, top of left column
We included a fairly high number (12) of target fragments
- p. 7 Measures, complete list
The complete list was as follows
Only the authors can say what is right here.
Judge's note: Counted on the images across the page break. Bottom of page 7: RACE, LAZY, BLACK, POOR, CLASS, BROTHER, WHITE and the stem "M I ___". Top of page 8: "(MINORITY); W E L ___ (WELFARE); C O ___ (COLOR); T O ___ (TOKEN)." Eleven entries, nothing lost in extraction. The self-doubt list has the seven it states, so the counting convention is not the cause. No statement in the paper explains a twelfth fragment. Whether a word was dropped from the list or the count is wrong is for the authors.
Second check: another model agreed
Report a problem with this finding
Wording Study 3, Indicating Race: "a significant Race X Condition interaction, F( 1, 61) = 6.60, p < .01" gives 1 numerator degree of freedom for the interaction of a 2 x 3 design; the other Study 3 interactions are F(2, 61).
- p. 8 Study 3 Results, Indicating Race
significant Race X Condition interaction, F( 1, 61) = 6.60, p
- p. 8 Study 3 Results, Stereotype Activation (same ANCOVA design)
Race X Condition interaction, F(2, 61 ) = 3.30
- p. 8 Study 3 Results, Indicating Race
F( 1, 61) = 6.60
- p. 8 Study 3 Results, Stereotype Activation
F(2, 61) = 3.30
Only the authors can say what is right here.
Judge's note: Page 8 image and crop confirm "F(1, 61) = 6.60". The analysis is described as "the standard ANCOVA", the 2 (race) x 3 (condition) design, and the condition effect in the same sentence has 2 degrees of freedom. The three other Race x Condition interactions on the page are F(2, 61) = 3.30, 3.34 and 4.14. A two-condition analysis would not have 61 error degrees of freedom. The printed pair also does not correspond under the declared test: F(1, 61) = 6.60 is p of about .013 (the .01 critical value is about 7.06), whereas F(2, 61) = 6.60 is about .003. A single-df contrast is conceivable but is not what the text describes. The period in "F(2.61)" is handled in C015.
Second check: another model agreed
Report a problem with this finding
12 minor (typos and the like; most checked by the judge only)
Reference or label Reference list: "Goffman, I. (1963). Stigma. New York: Simon & Shuster, Inc." gives the author's initial as "I." and spells the publisher "Shuster".
- p. 15 References
Goffman, I. (1963). Stigma. New York: Simon & Shuster, Inc.
- p. 15 References, Goffman (1963)
Simon & Shuster, Inc.
- p. 15
GofTman, I. ( 1963). Stigma. New York: Simon & Shuster, Inc.
Only the authors can say what is right here.
Judge's note: Rests on outside knowledge: that the author of Stigma (1963) is Erving Goffman (initial E.) and that the publisher's name is spelled Simon & Schuster. Page 15 image confirms the entry is printed with "I." and "Shuster"; neither is an extraction artefact (the "GofTman" in the text layer is). The facts relied on are not in the files, so this goes to the authors as a query.
Second check: another model agreed
Report a problem with this finding
Wording "F( 1, 107) = 10.47, p < .001" (Study 1, test bias, main effect of race): the statistic and the P value do not correspond under the declared test.
- p. 5 Study 1 Results, Self-Report Measures
F( 1, 107) = 10.47
- p. 5 Self-Report Measures, test bias
F( 1, 107) = 10.47, p < .001
Only the authors can say what is right here.
Judge's note: Page 5 image confirms the printed text. With 1 and 107 degrees of freedom, F = 10.47 corresponds to p of about .0016; p < .001 needs F of about 11.4 or more (t of about 3.39; 10.47 is t of about 3.24). This is an ANCOVA main effect, so the Bonferroni procedure of footnote 2 does not apply, and it would raise p in any case. No one-sided test is declared. Not an abstract result; S3.
Second check: another model agreed
Report a problem with this finding
Wording Self-doubt fragments "again included at least two blank letter spaces in each fragment", but the fragment for HARD is printed "_A R D (HARD)", which has one blank.
- p. 8 Study 3 Measures, Self-doubt activation, fragment list
(HARD)
- p. 8 Study 3 Measures, Self-doubt activation
again included at least two blank letter spaces in each frag-
Judge's note: Crop of page 8: the fragment shows a single underscore followed by A R D, while the other six (LOSER, DUMB, SHAME, INFERIOR, FLUNK, WEAK) show two or more blanks. Independent of how the underscore is rendered, HARD has four letters and three are given, so only one can be blank. Either the fragment as printed differs from the one used or the "at least two" statement is loose.
Second check: another model agreed
Report a problem with this finding
Wording "These effects were qualified by a significant Race X Condition interaction, F(1, 39) = 4.1, p < .01": the F value and the P bound do not correspond under the declared test (F with 1 and 39 df).
- p. 9 Study 3 Results, Self-Handicapping
F(1, 39) = 4.1, p < .01
- p. 9 Study 3 Results, Self-Handicapping (same df, for comparison)
F( 1, 39) = 5.73, p < .03
- p. 9 Self-Handicapping, hours of sleep
interaction, F(1, 39) = 4.1, p < .01
- p. 9 Self-Handicapping, hours of sleep
F(\, 39) = 4.1, p < .01
Only the authors can say what is right here.
Judge's note: Page 9 image confirms the printed text: F(1, 39) = 4.1, p < .01. Under an F distribution with 1 and 39 df, 4.1 sits at about the .05 critical value (about 4.09); p < .01 needs F of about 7.3 or more. The same paragraph, with the same df, prints F = 5.73 with p < .03, F = 6.53 with p < .02, F = 7.26 with p < .02 and F = 3.58 with p < .07, so this pair is out of line with its neighbours. No one-sided test or correction is declared for these ANCOVA F tests (the Bonferroni note under Table 1 concerns comparisons of means and would not lower a P value). A rough back-calculation from the Table 1 means is in line with an F near 4.1, but cell sizes are unequal so it is approximate. Not a result in the abstract, so S3. Only the authors can say whether the F or the P bound is misprinted.
Second check: another model agreed
Report a problem with this finding
Wording The stereotype threat questionnaire is described as "each of eight statements", but "A MANOVA of the stereotype threat scale" is reported with "F(9, 31) = 8.80, p < .01", degrees of freedom that correspond to nine dependent variables.
- p. 11 Study 4 Results, Stereotype Threat and Academic Identification Measures
F(9, 31) = 8.80, p < .01
- p. 10 Study 4 Method, Dependent Measures
eight statements
- p. 10 Study 4 Method, Dependent Measures
nine further items
- p. 12 Study 4 Discussion, second paragraph
(as measured by the stereotype threat and stereotype avoidance measures)
Only the authors can say what is right here.
Judge's note: Page 10 image: "each of eight statements" for the stereotype threat questionnaire and "nine further items" for academic identification. Page 11 image: "A MANOVA of the stereotype threat scale ... F(9, 31) = 8.80, p < .01". For a two-level factor the numerator df of a MANOVA equals the number of dependent variables, and with 44 participants, four cells and one covariate (error df 39, as in the paper's F(1, 39) tests) nine variables give (9, 31) and eight would give (8, 32). The paper does not mention a ninth threat item or any other variable in that analysis; nine is the number of identification items. Whether the item count, the df or the scale named is misreported cannot be told from the paper, so this is a question for the authors. The second point raised in this cluster (a "stereotype avoidance" measure in the Study 4 discussion) is judged under C021.
Second check: another model agreed
Report a problem with this finding
Typo Missing article: "their grades average two thirds of letter grade lower than those of graduating Whites" should read "two thirds of a letter grade".
- p. 2 Introduction, left column
their grades average two thirds of
Judge's note: Read on the page 2 image, left column: the line ends "two thirds of" and the next begins "letter grade lower", with no "a". Not an extraction loss.
Second check: none (judge only)
Report a problem with this finding
Typo One Study 1 condition is named four ways: "non-diagnostic-plus-challenge condition", "non-diagnostic-challenge condition", "nondiagnostic-challenge condition" and "the challenge condition"; the non-diagnostic-only condition is likewise shortened to "the nondiagnostic condition".
- p. 3 Study 1 introduction
non-diagnostic-plus-challenge condition
- p. 4 Results, Test Performance
the challenge condition, t( 107) = 2.63
Judge's note: Page 3 image shows "non-diagnos-tic-plus-challenge condition" and, in the next sentence, "non-diagnostic-challenge condition". Page 4 image shows "the nondiagnostic condition, t(107) = 2.88 ... or the challenge condition, t(107) = 2.63" and "nondiagnostic-challenge condition". The meaning is recoverable in each place, so this is a consistency point only.
Second check: none (judge only)
Report a problem with this finding
Typo Punctuation slips in statistics: "F(2.61) = 3.40, p < .04" (page 8; should be F(2, 61)) and "t(107) = 2.40, p < .02., while test description" (page 5; stray period after .02).
- p. 8 Study 3 Results, Indicating Race
F(2.61) = 3.40, p < .04
- p. 5 Study 1 Results, Perceived Performance
.02., while test description had no effect
- p. 8 Study 3 Results, Indicating Race
F(2.61) = 3.40
- p. 5 Perceived Performance, last sentence
p < .02., while test description had no effect
Judge's note: Crop of page 8 shows "F(2.61) = 3.40" with a period and no space, where the neighbouring tests read "F(1, 61)" with a comma. Crop of page 5 shows ".02., while". Both are in print, not extraction artefacts.
Second check: none (judge only)
Report a problem with this finding
Typo Missing unit: "they spent 94 s answering each of these items in contrast to 71 for Black participants in the nondiagnostic condition"; the other values in the sentence carry "s" ("73 s", "71 s").
- p. 6 Results and Discussion, time on first five items
in contrast to 71 for Black participants in the nondiagnostic con-
Judge's note: Page 6 image, right column: "94 s", then "71 for Black participants", then "73 s for Whites" and "71 s for Whites". The unit is absent in print on the first comparison value only.
Second check: none (judge only)
Report a problem with this finding
Typo Missing word: "more than participants in any of other conditions, all ps < .05" should read "any of the other conditions".
- p. 8 Study 3 Results, Stereotype Activation
more than participants in any of other condi-
Judge's note: Crop of page 8, right column, Stereotype Activation: "more than participants in any of other condi-tions". The parallel sentence under Self-Doubt Activation has "any of the other conditions".
Second check: none (judge only)
Report a problem with this finding
Typo "It did the same for their self doubts." is unhyphenated, while the paper elsewhere writes "self-doubts" (e.g. "Seven word fragments reflecting self-doubts", "self-doubts associated").
- p. 8 Self-Doubt Activation
It did the same for their self doubts.
Judge's note: Crop C016-a shows "self doubts" with a space and no hyphen in the first sentence under the Self-Doubt Activation heading; the same crop shows "self-doubts" hyphenated in the method paragraph, and the text has the hyphenated form at four other places. Cosmetic inconsistency.
Second check: none (judge only)
Report a problem with this finding
Typo "then, separately, that it causes in these participants to be threatened by the stereotype" is ungrammatical (superfluous "in", or a missing noun).
- p. 10 Study 3 Discussion, continued, left column
then, separately, that it causes in these participants to be threat-
- p. 10 Discussion before Study 4, left column
that it causes in these participants to be threat-
Judge's note: Page 10 image, left column, second paragraph: the wording is printed exactly as quoted. The clause mixes "causes in these participants [a sense of threat]" with "causes these participants to be threatened". Cosmetic.
Second check: none (judge only)