Volume 3, No. 11
September/October 1999
Admissions Tests
Blaming the Tests, Not Bridging the Gap
By Jon Sanders
Two recent developments concerning the racial gap in standardized-test scores signal a troubling willingness by the government and the Educational Testing Service (ETS) to ignore it. The first is the release by the U.S. Department of Education's Office of Civil Rights (OCR) of proposed guidelines to discourage universities from using standardized tests (such as the ACT and SAT) in admissions. The second is the ETS's new "Strivers" program, which would weigh test-takers' backgrounds including their gender, race and ethnic background in their performance on the SAT.
Neither proposal would work toward answering the question of why the gap exists in the first place. Both seem intent instead of keeping some means of race-preferential, "affirmative-action" discrimination in place in collegiate admissions decisions.
The OCR's guidelines
The first preemptive strike against a post-preferences America was made by the OCR in proposing its guidelines. The method of the OCR's proposal is a tried-and-true practice known as killing the messenger. The messenger, in this case, is standardized testing, and the message is the education gap (whites consistently outscore blacks on standardized tests). This fact the OCR was apparently prepared to accept as long as universities were able to use racial preferences in their admissions to pave over the gap (an uncharitable observer might note they have been unable to extend the ruse to graduation rates). But as preferences fall across the country they're already illegal in Texas, Louisiana, Mississippi, California and Washington university admissions rely more on standardized tests. Suddenly the OCR finds the gap worthy of action, so naturally it attacks the tests.
To the OCR, bias is all about numbers. The OCR assumes black students and white students across the nation are equally prepared for higher education, so it views any measuring tool whose results indicate otherwise as biased. Think of it as the Dr. Watson School of Deduction going from a flawed premise to an illogical conclusion. By Watson analysis, a gap in scores can mean only one thing: the tests are biased.
The crux of the guidelines, as far as how they would affect the use of standardized tests, is their use of "disparate impact" analysis. According to the guidelines: "Under a disparate impact analysis, the focus is on the 'effects' of the application of a facially neutral policy or practice, regardless of whether the adverse consequences for a particular race, national origin, or gender were intended." In terms of the standardized test, although the use of such an instrument as the SAT is "facially neutral," the fact that blacks score significantly lower on the SAT than whites (certainly not "intended") could mean that the practice has "adverse consequences" and is thus illegal. Institutions that require tests with such "effects" must be able to prove the test is "valid and reliable for the purpose for which it is being used" and also show that the test is "the least discriminatory practical alternative that can serve the education institution's educational purpose."
Such vague terms "reliable," "least discriminatory," "practical" beg for judicial interpretation, which is precisely the guidelines' threat. Let one rejected minority applicant decide the SAT or ACT isn't "reliable," "practical," etc., and the institution that required it will be hit with a lawsuit that, no matter how the court decides, will be expensive and fraught with bad publicity.
Should the OCR's guidelines be put into practice, many institutions would probably opt to give up or at least rely much less on standardized tests in admissions decisions. In that case, fewer students nationally would take the tests, and the education gap which will remain untreated will now go unobserved. But the headlines are dominated not by test-score gaps, but by minority enrollment. As long as black enrollment into higher education remains unaffected, no educrat will lose sleep over black students' poor learning.
The Strivers Score
To the institutions' rescue gallops the ETS, armed with its new Strivers scoring system. Strivers analysis aims to place test-takers' scores in the context of 14 different personal categories, including their gender, race, ethnicity, parents' education and income, schools' academics and mothers' employment status (for more details, see page 33). According to those factors, each test-taker is assigned an "expected" score, supposed to be what a test-taker with his particular mix of factors would normally score. Those who exceed their expected score by 200 points or more are considered Strivers. They are marked, according to Anthony Carnevale, vice president for public leadership at ETS, with "drive and can-do attitude," and they are the kind of students admissions officers try to find.
But most if not all institutions already have processes that take into account the intangibles of individual applicants' backgrounds. So in one sense the program is redundant. With race-preferential admissions in jeopardy, however, the Strivers program might meet a critical need of universities, should they wish to maintain their enrollment of protected minorities without appearing to discriminate. Thus, the scores the ETS would offer to quota-minded (or, to use the less illegal term, goal-minded) institutions would already be adjusted according to sex, race and ethnicity, saving the institutions from having to do the dirty work.
Race can be taken out of Strivers analysis, however; the ETS will be offering institutions a race-blind model of the Strivers program as well as one that factors in race and ethnicity. Factoring in race, of course, lowers the expected scores of black and Hispanic test-takers more than not doing so, which makes it more likely that they would be identified as Strivers.
The ETS has already demonstrated a proclivity for altering the SAT according to the political winds. Following years of falling scores by test-takers across the spectrum, which should have prompted concerns nationwide over the effectiveness of public schools, the ETS "recentered" the SAT scores to bump them up to previous levels. Now, as race-based discrimination in collegiate admissions hobbles on its last legal legs and universities scrounge for ways to fill all their protected categories, up steps the ETS to do the discriminating for them with this new way of reading SAT scores according to race, sex and ethnic background. It's a brilliant tactical move in a realpolitik sense, never mind that it scuttles any claims to objectivity the SAT had left.
That persistent problem
Neither of these schemes offer any solution to the real problem; namely, that black students lag their white peers in the classroom. It is the ugly crack in the proud facade of American education. Few educators have come forward with the putty to patch it, however. Most prefer to hang a sunny landscape over the crack to hide it.
The Education Dept. has taken the easy route. It neglects to take affirmative action in the real sense of the words early enough in our children's schooling to fight the education gap at its genesis. No, only after high-school graduation does the Education Dept. fight the problem, now in full bloom. It combats it by having universities discriminate against white and Asian students for being better prepared what it calls "affirmative action."
It will be helped now by the ETS. The Strivers program will cover the scoring gap nicely, with lots of sophistic hooey to justify doing so. One of the primary instruments for measuring the continuing failure by society and the education system to reach minority students will be rendered mute. There will be no harsh grackles to interrupt this sunny new morn.
The OCR guidelines and the Strivers program are designed to fend off ending race-based discrimination in collegiate admissions. For their creation, their designers will garner good press. And they'll still be playing fast and loose with black students' education.
But there'll be no interrupting the good feelings just to worry over that persistent education gap. Not as long as black college enrollment remains constant. -
Next pages:
The tables on the following pages have been chosen to illustrate the extent of the racial gap in scores on standardized tests and its practical effects.
Disparate Impact Analysis Excerpt from the OCR Guidelines
The following is excerpted from "Nondiscrimination in High-Stakes Testing: An Overview," the draft guidelines sent by the U.S. Education Department's Office for Civil Rights to college officials. The excerpt discusses disparate-impact analysis in determining whether a standardized test used by universities in making admissions decisions or for other assessment needs is discriminatory.
III. Disparate Impact Analysis
A disparate impact analysis may be applied to allegations involving discriminatory test use by educational institutions. Under this analysis, the use of any educational test which has a significant disparate impact on members of any particular race, national origin, or sex is discriminatory, and a violation of Title VI and/or Title IX, respectively, unless it is educationally necessary and there is no practicable alternative form of assessment which meets the educational institution's educational needs and would have less of a disparate impact on the basis of race, national origin, or sex.
In applying a disparate impact analysis, the following questions should be addressed:
A. Does the educational institution's use of an educational test result in the significantly disproportionate denial of an educational benefit or opportunity to members of a particular race, national origin, or sex?
B. If so, is the use of the test educationally necessary?
C. If so, do there exist practicable alternative forms of assessment which would substantially serve the school's stated purpose and are valid and reliable for that purpose, but which have less of a disparate impact on the basis of race, national origin, or sex?
Each question is discussed in more detail below. Where, based on evidence, there is a finding that the use of a test or assessment procedure caused or contributed to a disparate impact on members of a particular race, national origin, or sex (the first question), and the test or procedure does not meet the legal standard of educational necessity (the second question) or there is a practicable alternative form of assessment which would meet the educational institution!s educational needs and would have less of a disparate impact on the basis of race, national origin, or sex (the third question), there is a violation of Title VI or Tide EX under this disparate impact analysis.
A. Establishing Disparate Impact
Under a disparate impact analysis, a school's use of an educational test that causes or contributes to a disproportionate denial of an educational benefit or opportunity to members of a particular race, national origin, or sex is sufficient information to indicate a possible failure of compliance with Title VI or Title IX which should be investigated further. It is important to note that disparate impact by itself does not necessarily mean that discrimination has taken place. Disparate impact may lead to a finding of discrimination only when use of the test in question is not educationally necessary or when there is no practicable alternative form of assessment which would meet the educational institution's educational need and have less of a disparate impact on the basis of race, national origin, or sex.
B. Establishing Educational Necessity
Once it has been determined that a disparate impact exists, it must then be determined whether the use of the test or assessment procedure is educationally necessary.1 To meet the educational necessity standard, the test or assessment procedure must be valid and reliable for the purpose for which it is being used.
In evaluating the validity and reliability of a test or assessment procedure, generally accepted professional standards should be the foundation for such decision making. These standards include the Standards for Educational and Psychological Testing prepared by a joint committee of the American Psychological Association, the American Educational Research Association, and the National Council on Measurement in Education; the Code of Fair Testing Practices in Education prepared by the Joint Committee on Testing Practices; and the Uniform Guidelines on Employee Selection Proceduress.2 All decisions as to whether a test or procedure has met professionally accepted standards should be made in consultation with experts.
Notes
1. Where a test is being used as the sole or principal criterion for making educational decisions and where it was clearly not designed to be used as such, there is no basis upon which to conclude that the test is educationally necessary.
2. Although there are many principles in the Uniform Guidelines that apply to educational testing in general terms, the Uniform Guidelines do not address educational testing issues. There are critical, contextual differences between employment and educational testing that should not be overlooked when using the Uniform Guidelines as a resource in the educational setting. The Uniform Guidelines were adopted by and are currently used by the U.S. Equal Employment opportunity Commission, the U.S. Department of Labor, and the U.S. Department of Justice.
Variables Used in Striver Analysis
Personal and Family Characteristics
Family: Socioeconomic status of the student's family. This is a index of the eduation of parents, occupation and total family income. The index also incorporates a number of measures of living standards such as number of books in the household, kinds of electrical applicances, etc.
Mother: Whether the mother is currently employed. Researchers consider this to be an indicator of the education of the mother and father.
Gender and Race: The student's sex, race and ethnic background. The race and ethnicity category can be left out of the formula.
School Context
School Status: Whether the student attends a school where more than 50 percent of the students receive a subsidized lunch.
Low College-Bound Population: Whether the student attends a high school where less than 50 percent of its previous year graduates entered a four-year college.
Setting: Whether the student attends a public or private high school, whether the school is in an urban or rural area, and the location of the school by census region.
Student's Academic Characteristics
Rigorous: Whether the student attended a school that offered rigorous academic courses.
Language: Whether English is the student's native language.
Age: Whether the student is two or more years older than his or her peers.
Core Course Grades: An average of the student's grades for the core courses in English, mathematics, science, social studies, computer science and foreign languages.
Source: Amy Sinatra, ABCNEWS.com, Aug. 31.
Table 1. SAT scores by race, verbal and math, from 1987 to 1999 Verbal scores 1987 1997 1999 Math scores 1987 1997 1999
Black 428 434 434 Black 411 423 422
White 524 526 527 White 514 526 528
Gap -96 -92 -93 Gap -103 -103 -106
Table 2. Mean scores of 17-year-olds by race, and the racial gap between blacks and whites in years, National Assessment of Educational Progress, 1994
Black
White Gap in Years
Reading
266 296 3.9
Mathematics
286 312 3.4
Science
257 306 5.4
Writing
267 291 3.3
Source: America in Black and White: One Nation, Indivisible, by Stephen Thernstrom and Abigail Thernstrom, Simon & Schuster, 1997, p. 355.
Table 3.Number and percent of black and white students with high SAT scores, 1981 and 1995 1981
1981 1995 1995
Verbal Black White Black White
700-800 70 8,239 184 8,978
650-699 221 16,216 465 19,272
600-649 596 33,231 1,115 36,700
Total >600 950 57,685 1,764 64,950
Percent >600 1.2 8.0 1.7 9.6
Math
Black White Black White
750-800
24 5,077 107 9,519
700-749
132 16,257 509 29,774
650-699
393 35,353 1,437 51,306
Total >650
549 56,687 2,053 90,599
Percent >650
0.7 7.9 2.0 13.4
No. examined
75,434 719,383 103,872 674,343
Source: America in Black and White: One Nation, Indivisible, by Stephen Thernstrom and Abigail Thernstrom, Simon & Schuster, 1997, p. 399.
Table 4. How graduation rates of black students vary with their SAT scores
and the median SAT score at their college, 1972 high school graduates
Colleges with median SATs of:
Percent graduating of students with SATs of 900
Percent graduating of students with SATs of 1,000
700 or less 38 26
701-850
56 39
851-1,000 77 51
over 1,000 77 51
Source: America in Black and White: One Nation, Indivisible, by Stephen Thernstrom and Abigail Thernstrom, Simon & Schuster, 1997, p. 410.
Table 5. SAT scores of black freshmen, 1992, the racial gap in
scores, and dropout rates in selected elite universities
Percent dropping out Percent dropping out
Institution
SAT racial gap Mean Black SAT Black White
Harvard
95 1,305 5 3
Princeton 150 1,172 9 5
Brown 150 1,160 13 6
Pennsylvania 150 1,135 28 10
Cornell 162 1,118 23 8
Stanford
171 1,164 17 6
Northwestern
180 1,075 21 10
Columbia 182 1,128 25 12
Duke
184 1,126 16 5
Dartmouth 218 1,112 16 4
Virginia
241 979 16 7
Rice 271 1,093 26 11
UC-Berkeley 288 947 42 16
Source: America in Black and White: One Nation, Indivisible, by Stephen Thernstrom and Abigail Thernstrom, Simon & Schuster, 1997, p. 408.
Table 6. Passage of teacher competency tests1 in select states2, 1982-83
Percent passing the test
Whites
Blacks
California
76 26
Florida
90 35
Arizona
73 24
Texas 62 10
1. Tests used to determine public school teachers' basic level of knowledge of reading, writing and mathematics.
2. D'Souza notes that "[s]imilar white-black gaps were found in Georgia, Virginia, Oklahoma, and elsewhere."
Source: The End of Racism, by Dinesh D'Souza, Free Press, 1995, p. 306.
--------------------------------------------------------------------------------
RETURN TO CLARION SEPTEMBER/OCTOBER 1999
©1999 by The Pope Center for Higher Education Policy
P.O. Box 14502, Research Triangle Park, NC 27709-4502
phone: 919-789-8694, fax: 919-789-8690
Clarion is published monthly and contains analysis and commentary on policy issues in higher education in North Carolina.
Material published here may be reprinted provided the The Pope Center for Higher Education Policy receives prior notice and appropriate credit is given. Submissions are welcome.
The Pope Center welcome comments and ideas for improvement on this web site. Please contact our webmaster with questions or suggestions.
|
