Law Review
Northwestern Law
Northwestern University Law Review : Colloquy : 2007 : Chang-Villazor

"Testing the 'Model Minority Myth'":  A Case of Weak Empiricism

By Robert S. Chang[*] and Rose Cuison Villazor[**]

[Editor's Note:  This piece is a reply to Miranda Oshige McGowan and James Lindgren, Testing the "Model Minority Myth", 100 Nw. U. L. Rev. 331 (2006) (link).]

I.  Introduction

As the legal and political wars rage over affirmative action, the role played by Asian Americans is increasingly disputed.  Should Asian Americans be included in affirmative action programs?  Do such programs harm Asian Americans because spaces that should be given to them are instead given to less qualified Blacks and Latinos?  Does Asian American success indicate that affirmative action polices are unnecessary? At stake in each of these questions is the notion that Asian Americans are the "model minority."

The model minority designation often depicts Asian Americans as achieving success through cultural values and hard work,[1] presenting them as a blueprint for others who want to achieve similar success.  These positive attributes of Asian Americans are used to illustrate their ability to overcome discrimination and to juxtapose them with other racial groups, whose failure to overcome discrimination is then blamed on a lack of these cultural traits and an ethic of hard work.[2]  This juxtaposition may then be used to imply that social policies such as affirmative action, welfare, or reparations might be wrongly directed.

Critical discussion of the model minority designation is now standard fare in Asian American jurisprudence.[3]  A number of scholars have critiqued the veracity of the success stereotype and the validity of indicia such as household income or education as a measure of ongoing discrimination against Asian Americans.[4]  They have also posited that the model minority designation has had a negative impact on those Asian Americans who do not fit the success model.  Additionally, they claim the stereotype has harmed other racial minorities and poor whites who have not achieved similar success.

Professors McGowan and Lindgren make an important contribution by trying to test the causal connection between the model minority stereotype and these posited harms.  They set out to test this empirically through statistical analysis of survey data, concluding that their "results do not confirm the Model Minority Hypothesis."[5]  Instead, they contended that their research demonstrated "slightly more evidence conflicting with the Hypothesis than confirming it."[6]  They concluded that "we must all be careful to present our generalizations, not as essences or necessities, but as conclusions that are true only to the extent that they fit the world and untrue to the extent that they do not fit what they claim to capture."[7]  This cautionary note is directed to "Asian [sic] critical scholars"[8] who critique the model minority designation, suggesting that we are making a claim that does not fit the world.

We turn this cautionary note back on McGowan and Lindgren.  Their conclusion is only as good as their test design, which depends on their assumptions and the validity or accuracy of their underlying data.

II.  The McGowan/Lindgren Hypothesis and its Underlying Assumptions

McGowan and Lindgren do not dispute the "positive" side of the model minority stereotype.  Instead, they question its link to the harms posited by Asian American critical scholars.  McGowan and Lindgren catalogue the harms posited by Asian American scholars into five categories: the stereotype (1) amalgamates all Asian Americans into one group, masking social and economic disparities among them; (2) "blinds Americans to the persistence of discrimination against Asian Americans"; (3) "reinforces the American dream and implicitly blames other minority groups for their problems"; (4) "divides Asian Americans from other minority groups"; and (5) illustrates the new "Yellow Peril."[9]

Ultimately, McGowan and Lindgren's working hypothesis might be best described as follows: racial harms are not causally associated with the model minority designation if individual non-Hispanic whites do not hold both the positive and negative attitudes discussed above.[10]  We argue that their analysis is informed by three unstated assumptions: (1) that attitudes as expressed conform to "true" attitudes; (2) that context is irrelevant; and (3) that discrimination conforms to what has been described as the "perpetrator perspective."[11]

III.  The Gap Between Expressed Attitudes and "True" Attitudes

McGowan and Lindgren's raw data about racial attitudes held by non-Hispanic whites comes from the General Social Survey ("GSS"), which is conducted by the National Opinion Research Center ("NORC") at the University of Chicago.  The GSS is widely used in the social sciences.  It has a number of beneficial characteristics, including its high quality survey technique and ease of use.  Furthermore, the GSS, a survey begun in 1972, at first annually and then biennially since 1994, allows for longitudinal analyses of a number of different characteristics, including responses and trends in responses to questions about racial attitudes.[12]  However, researchers should be cautious about what the GSS measures and what kinds of conclusions are justified from its data.  This is because the GSS measures attitudes as they are expressed.[13]  However, there is a gap between attitudes as expressed and people's "true" attitudes.  We will discuss two problems with such survey data: (1) the falsification problem; and (2) the response bias problem.

Immediately before the 2006 election, polls about Proposal 2 in Michigan, which would ban affirmative action, indicated that it was too close to call.[14]  It ended up passing by an overwhelming margin, approximately 58% to 42%.  People lie to pollsters, especially when it comes to candidates or issues involving race.[15]  We call this dynamic the falsification problem.  Although falsification might be subsumed under response bias, we separate it here to describe conscious lying to pollsters or interviewers.  The problem of falsification appears to extend beyond election polls.  One researcher comments that "[i]n spite of the exquisite efforts of social psychology, its home discipline, opinion polling even when done by the best-presumably the NES and the GSS-seems beleaguered by intractable problems of response falsification."[16]

Furthermore, any survey is subject to the problem of response bias, which can arise in a number of ways.  First, bias can arise from question wording.  This type of bias can be seen in survey responses to similar questions about residential integration.  Compare the following:

Residential Choice, 1 alternative plus scale (NORC):  "Here are some opinions other people have expressed in connection with black-white relations.  Which statement . . . comes closest to how you feel?  White people have a right to keep blacks out of their neighborhoods if they want to, and blacks should respect that right."  [Respondents are . . . asked to choose one of four responses: strongly agree, slightly agree, slightly disagree, strongly disagree.]

Residential Choice, 2 alternatives (ISR):  "Which of these choices would you agree with more:  (1) White people have a right to keep blacks out of their neighborhoods if they want to.  (2) Black people have a right to live wherever they can afford to, just like everybody else." [17]

Schuman et al. found that the second question produces a response from white respondents more favorable to black rights by an average margin of 25% compared to the first question.[18]

Second, bias can arise from question context.  This type of bias is determined by what questions are asked prior to the question being analyzed.  The GSS is probably less susceptible to this kind of bias because of the NORC's emphasis on keeping the questionnaire consistent from year to year, but there are variations because some items are rotated in and out of the survey.[19]

Third, bias can arise from an interaction between the race of the interviewer and that of the respondent.[20]

Finally, the privacy of the interview can impact responses, with differences seen in comparing face-to-face with virtual with mailed surveys.[21]

The GSS, conducted in face-to-face interviews,[22] is susceptible to all of these biases, casting doubt on whether expressed attitudes represent those truly held.  This is especially true because highly educated white respondents tend to respond to face-to-face surveys with responses that skew toward the socially desirable, reflecting society's norms about racial attitudes.[23]  Therefore, what tends to be expressed does not provide good data about "true" attitudes.

The predicted result of this tendency towards falsification and response bias is consistent with what McGowan and Lindgren found: non-Hispanic white respondents who expressed the positive attitudes toward Asian Americans associated with the model minority designation tended not to simultaneously express negative attitudes toward Asian Americans and other racial minorities.  This asymmetry between the expression of positive and negative stereotypes about Asian Americans accords with society's norms.  This calls into question what kinds of generalizations or conclusions may be drawn from McGowan and Lindgren's data.  The data does not really tell us what non-Hispanic whites feel, nor how they will act on their convictions in any specific situation, which brings us to our next point.

IV.  Context Is Everything

A second problematic assumption is that McGowan and Lindgren's analysis treats the expressed attitudes in the GSS as if context were irrelevant.   However, social psychology teaches us that context plays a tremendous role in affecting how attitudes manifest themselves.[24] Consider the following study:

The researchers asked participants to choose one of two rooms in which to watch a movie.  In the condition with no ambiguity, the researchers offered the same movie in two rooms.  A person with a disability sat in one of the rooms while the other room was empty. Most participants chose to sit next to the person in a wheelchair.  In the attributional ambiguity condition, the researchers offered two different movies: one that the person with a disability watched and another that played in an empty room.  In this case, most participants chose to sit in the empty room regardless of the movie that played there.[25]

The results from the first situation accord with social norms about how one should act towards those with different abilities.  Though participants might prefer not to sit next to the person in the wheelchair, the fear of appearing prejudiced led them to overcome that preference.  The results from the second situation, though, show that the force of the social norms is insufficient to overcome the participants' preferences when they can justify or rationalize their behavior as being based on movie choice and not prejudice.  Thus, the lack of any attributional ambiguity in the GSS context may mean that conclusions drawn from the expressed attitudes are likely to tell us little about real world discrimination.

V.  The Inadequacy of the Perpetrator Perspective

By limiting their analysis to the attitudes held by non-Hispanic whites, McGowan and Lindgren appear to be testing whether non-Hispanic whites, as the dominant group, are, or could be, the perpetrators who cause the harms posited by Asian-American critical scholars.  In addition to ignoring the roles played by Asian Americans and other racial groups,[26] this analysis rests on a problematic assumption about what constitutes discrimination.  This approach, relying on the "perpetrator perspective,"[27] holds that discrimination exists only when an individual's intent to discriminate is evident.

We believe, with many other scholars,[28]that the perpetrator perspective does not capture much of the discrimination that takes place in the real world.  It ignores the possibility of discriminatory impact resulting from apparently neutral policies and ignores the role that cognitive bias plays in discrimination.[29]  One of the lessons from social psychology is that there are unconscious attitudes that can be resistant to the conscious will, so that we act in ways that we do not always intend.[30]  McGowan and Lindgren's analysis, grounded in the perpetrator perspective, does not consider the implications of the growing body of scholarship on unconscious racism and cognitive bias.

VI.  Conclusion

Professors McGowan and Lindgren's article is timely in light of the debate over affirmative action and the role that Asian Americans play in that process.  Unfortunately, they limit the scope of their analysis to the results of surveys of non-Hispanic whites produced from face-to-face interviews about their racial attitudes.  From this, they make claims about the model minority stereotype in the real world.  Their data and conclusions are likely to be used by those who seek to end affirmative action and who seek to use affirmative action as a wedge issue to create divisions among Asian Americans and between Asian Americans and other racial minorities.

Though they support their claims with graphs and statistical analyses, consistent with the recent empirical turn in legal scholarship, close scrutiny of their analysis reveals the unreliability of their findings.  The real world is a place where people lie, where people are unaware of their biases, and where conscious and unconscious biases may not be clear or manifest themselves outside of particular contexts or situations.  The result is that the work of Asian American scholars on the model minority myth probably says much more about the state of the real world than do McGowan and Lindgren.


* Sturm Distinguished Visiting Professor of Law, University of Denver Sturm College of Law.  Professor of Law and J. Rex Dibble Fellow, Loyola Law School, Loyola Marymount University.  A.B. Princeton University; M.A., J.D. Duke University.

** Assistant Professor of Law, Southern Methodist University Dedman School of Law.  B.A. University of Texas, J.D. American University, LL.M. Columbia University.      

1. Nicholas D. Kristof, The Model Students, N.Y. Times, May 14, 2006, at 4-13 (link).  See also Timothy Egan, Little Asia on the Hill, N.Y. Times, Jan. 7, 2007, at 4A24 (link).

2. See Robert S. Chang, Toward an Asian American Legal Scholarship:  Critical Race Theory, Post-Structuralism, and Narrative Space, 81 Cal. L. Rev. 1241, 1260 (1993), 1 Asian L.J. 1, 20 (1994).

3. See, e.g., Yamamoto et al., Race, Rights and Reparation:  Law and the Japenese American Internment 267-69 (2001) (discussing and criticizing the model minority myth); Pat Chew, Asian Americans:  The "Reticent" Minority and their Paradoxes, 36 Wm. & Mary L. Rev. 1 passim (1994) (same); Frank Wu, Neither Black nor White:  Asian Americans and Affirmative Action, 15 B.C. Third World L.J. 225, passim (1995) (same); Gabriel J. Chin et al., Beyond Self-Interest:  Asian Pacific Americans Toward a Community of Justice, A Policy Analysis of Affirmative Action, 4 UCLA Asian Pac. Am. L.J. 129, 148-151 (1996) (same) (link); Natsu Taylor Saito, Model Minority, Yellow Peril:  Functions of "Foreignness" in the Construction of Asian American Legal Identity, 4 Asian L.J. 71 passim (1997) (same).

4. See sources cited supra note 3.

5. Miranda Oshige McGowan & James Lindgren, Testing the "Model Minority Myth", 100 Nw. U. L. Rev. 331, 374 (2006) (link).

6. Id.

7. Id. at 377.

8. McGowan and Lindgren denote the work they are critiquing as that of "Asian critical scholars," justifying this usage in a lengthy footnote.  Id. at 332, n.3.  We object to the elision of the "American" in "Asian American," which is the prevailing usage.  Their usage perpetuates the view of Asian Americans as foreign. 

9. Id. at 336-43.

10. They justify limiting their analysis to non-Hispanic whites "[b]ecause most descriptions of the stereotype focus on whites' views."  McGowan & Lindgren, supra note 5, at 347.  But see infra note 26.

11. Alan D. Freeman, Legitimizing Racial Discrimination Through Antidiscrimination Law:  A Critical Review of Supreme Court Decisions, 62 Minn. L. Rev. 1049, 1052 (1978).

12. See NORC, Projects:  GSS Study Description, (last visited Feb. 19, 2007) (link).

13. See infra note 22.

14. See Editorial, Affirmative Action in Michigan, Chi. Trib. 22 , Nov. 6, 2006 ("Polls say the outcome is too close to call.").

15. See Kevin Lanning, The Social Psychology of the 2004 U.S. Presidential Election, 5 Analyses of Soc. Issues & Pub. Pol'y 145, 148 (2005) (describing the "tendency for polls to over-predict results for African American candidates" and the "pattern in which greater support is indicated for African American candidates by the public act of talking to a pollster than by the more private act of voting").

16. William M. Epstein, Response Bias in Opinion Polls and American Social Welfare, 43 Soc. Sci. J. 99, 106 (2006) (link).

17. Howard Schuman et al., Racial Attitudes in America 78-79 (1997).

18. Id. at 79.

19. Id. at 82.

20. See, e.g., Shirley Hatchett & Howard Schuman, White Respondent and Race-of-Interviewer Effects, 39 Pub. Opinion Q. 523 (1975-76); Cynthia Webster, Hispanic and Anglo Interviewer and Respondent Ethnicity and Gender:  The Impact on Survey Response Quality, 33 J. Marketing Res. 62, 70 (February 1996).

21. See, e.g., Maria Krysan, Privacy and the Expression of White Racial Attitudes:  A Comparison Across Three Contexts, 62 Pub. Opinion Q. 506 (1998); Maria Krysan & Mick P. Couper, Race in the Live and the Virtual Interview:  Racial Deference, Social Desirability, and Activation Effects in Attitude Surveys, 66 Soc. Psych. Q. 364, 365, 375-78 (2003).

22. See Schuman et al., supra note 17, at 59 ("All NORC data are based on face-to-face interviewing.").

23. Id. at 2-3 (discussing the concern that expressed racial attitudes "may represent little more than the superficial verbalization of socially approved norms" and conceding that their book, instead of being called Racial Attitudes, "might well be called Racial Norms in America").

24. See generally Jon Hanson & David Yosifon, The Situational Character:  A Critical Realist Perspective on the Human Animal, 93 Geo. L.J. 1 (2004) (link).

25. Vanessa Lynn Ewing et al., Student Prejudice Against Gay Male and Lesbian Lecturers, 143 J. Soc. Psych.  569, 572 (2003) (discussing M. L. Snyder et al., Avoidance of the Handicapped:  An Attributional Ambiguity Analysis, 37 J. Personality & Soc. Psych. 2297 (1979)).

26. While the dominant group's attitudes and perceptions are important, it is equally important to understand how other racial groups, including Asian Americans themselves, perpetuate problematic stereotypes and attitudes.  See Eric Yamamoto, Rethinking Alliances: Agency, Responsibility and Interracial Justice, 3 UCLA Asian Pac. Am. L.J. 33, 38 (1995).

27. See Freeman, supra note 11, at 1052-53.

28. See, e.g., Alan Freeman, Antidiscrimination Law:  The View from 1989, 64 Tul. L. Rev. 1407, 1410-11 (1990) (arguing that the intentional discrimination requirement perpetuates inequality); Linda Hamilton Krieger, The Content of Our Categories:  A Cognitive Bias Approach to Discrimination and Equal Employment Opportunity, 47 Stan. L. Rev. 1161, 1165 (1995) (same) (link); Charles R. Lawrence, III, The Id, the Ego, and Equal Protection:  Reckoning with Unconscious Racism, 39 Stan. L. Rev. 317, 324 (1987) (same).

29. See, e.g., Gary Blasi, Advocacy Against the Stereotype:  Lessons from Cognitive Social Psychology, 49 UCLA L. Rev. 1241 (2002); Jerry Kang, Trojan Horses of Race, 118 HARV. L. REV. 1489 (2005) (link).  See generally Symposium on Behavioral Realism, 94 Cal. L. Rev. 945 (2006) (link).

30. See Hanson & Yosifon, supra note 24, at 36.


Copyright 2007 Northwestern University

Cite As:  101Nw. U. L. Rev. Colloquy 101 (2007),

Persistent URL:

(Download PDF)


(back to top)