Toward Diverse Student Representation and Higher Achievement in Higher Levels American Educational Meritocracy

Michael T.Nettles and Catherine M.Millett

University of Michigan

INTRODUCTION

At the beginning of the 20th century, merit became synonymous with standardized intellectual tests in American education and in key sectors of the workforce. From the First World War when the U.S. Army administered intelligence tests to new recruits, the practice of sorting, selecting, and placing people based upon their test scores was launched. Colleges and universities began the practice using standardized intellectual tests for selecting students a few years later and the practice has grown steadily ever since. Despite the value placed upon other human attributes, tests and assessments are the most powerful levers of opportunity to higher-status education and employment.

Here at the beginning of the 21st century, standardized intellectual tests retain their lofty status as the core indicators of educational merit in America. From elementary and secondary schools up through undergraduate, graduate, and first-professional education, an individual’s standardized test scores frequently determine the college or university he or she will attend and the curricula that he or she will experience. The higher one’s scores, the higher the quality of schools, curricula, colleges, and professional schools he or she is invited to attend. Higher-quality education in turn leads individuals to higher-status employment and ultimately to a higher quality of life. The process of tracking begins in the early years of school. Moving from the lower to higher status tracks that lead to success is very difficult.

While the central focus of America’s meritocracy during the 20th century has been the performance of individual students on standardized intellectual tests, the spotlight and the influence of tests, and more recently assessments, has



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 143
The Right Thing to do, The Smart Thing to do Enhancing Diversity in the Health Professions Toward Diverse Student Representation and Higher Achievement in Higher Levels American Educational Meritocracy Michael T.Nettles and Catherine M.Millett University of Michigan INTRODUCTION At the beginning of the 20th century, merit became synonymous with standardized intellectual tests in American education and in key sectors of the workforce. From the First World War when the U.S. Army administered intelligence tests to new recruits, the practice of sorting, selecting, and placing people based upon their test scores was launched. Colleges and universities began the practice using standardized intellectual tests for selecting students a few years later and the practice has grown steadily ever since. Despite the value placed upon other human attributes, tests and assessments are the most powerful levers of opportunity to higher-status education and employment. Here at the beginning of the 21st century, standardized intellectual tests retain their lofty status as the core indicators of educational merit in America. From elementary and secondary schools up through undergraduate, graduate, and first-professional education, an individual’s standardized test scores frequently determine the college or university he or she will attend and the curricula that he or she will experience. The higher one’s scores, the higher the quality of schools, curricula, colleges, and professional schools he or she is invited to attend. Higher-quality education in turn leads individuals to higher-status employment and ultimately to a higher quality of life. The process of tracking begins in the early years of school. Moving from the lower to higher status tracks that lead to success is very difficult. While the central focus of America’s meritocracy during the 20th century has been the performance of individual students on standardized intellectual tests, the spotlight and the influence of tests, and more recently assessments, has

OCR for page 143
The Right Thing to do, The Smart Thing to do Enhancing Diversity in the Health Professions become much wider in scope. In addition to individual achievement, the measurement of merit today also encompasses the test and assessment performance of population groups (race/ethnic, class, and sex) and educational institutions (schools, colleges, and universities) in addition to individuals. Aggregate test or assessment scores of subpopulation groups are published today with greater frequency and are more often widely disseminated than in the past. Standardized tests and assessments reveal the relative status within the merit hierarchy of race/ethnic groups and their educational institutions. Institutions are ranked, rated, and classified based to a large degree upon the average of their students’ scores on various types of tests and assessments. The consequence is that individual merit and societal expectations are more formally intertwined with race group membership and the predominant race of the schools and colleges and universities. Unless institutions make score adjustments or use special weighting and calibrations on selection criteria, African Americans and Hispanics will be underrepresented in the higher status of American meritocracy. Because of differences in the score performance of students of different race/ethnic groups, there are separate merit scholarship programs for people of various race/ethnic groups. The separation of the National Merit Scholars and the National Achievement Scholars based upon race/ethnicity is one example. Here the Preliminary Scholastic Assessment Test (PSAT) is the instrument used for awarding merit. The history of lower African-American and Hispanic performance and the expectation that students from these two groups will perform relatively low on the test, has caused the National Merit Scholarship Organization to establish racially distinct programs in order to ensure that African Americans and Hispanics are included in relatively representative numbers. Several leading testing and assessment organizations, colleges, and universities have launched efforts to complement or even supplant traditional indicators of merit with alternative criteria. Criteria such as grades, educational and career aspirations, amount of effort, perseverance, and heritage, are variously considered as being complementary to test scores or to perhaps be possible alternative criteria. But, uniformity and stability are important elements of standardized tests and assessments, which places them in a preeminent role as the most equitable criteria for determining educational merit. Grades earned in school or college, for example, are often only considered to be valuable after weighing the quality of the school/college and curricula in which they were earned. The practice of weighting schools on the basis of their quality serves as a means of establishing uniformity and standardization in grades and other school-based indicators such as class rank, honors, and awards. Because there are no common standards for awarding grades, conditioning is the norm. Tests and assessments administered by independent objective agencies appear to be void of bias because they are administered under similar time and physical conditions and are therefore more credible indicators of merit. There are rarely, if

OCR for page 143
The Right Thing to do, The Smart Thing to do Enhancing Diversity in the Health Professions ever, adjustments made in test scores based upon group membership or educational and personal background. Rather, efforts to achieve racial/ethnic diversity are expended either by deciding the amount of weight to assign to test scores relative to complementary criteria, or by policies like the ones that are used by the National Merit Scholarship Program to segment their applicant pools based upon socio-demographic characteristics. Despite the popular anti-testing rhetoric, standardized tests and assessments have become so highly regarded as mediums of meritocracy that they are used to validate grades, curricula, and teaching; to certify the learning and accomplishments of students; to accredit and reward institutions; and to set priorities for public policies. Reversing the underrepresentation and progress of African Americans and Hispanics in higher levels of American education requires an increase in their participation and performance on test scores and grades. Only then will the members of theses two racial/ethnic groups and their educational institutions achieve higher status in American meritocracy. While all other non-cognitive measures are useful, they are only useful after being conditioned with other criteria. A person’s race or heritage, for example, may be used as a criterion for diversifying representation in certain strata of educational institutions such as college enrollments, but even race as a criterion is sometimes only an acceptable criterion after social class is taken into account. Social class appears to be an acceptable means of conditioning race and leads to boosting the opportunities of economically disadvantaged African Americans and Hispanics to attain better educational preparation and quality of life. At the same time, however, using social class as a criterion in conjunction with race reduces the prospect of achieving diversity at higher status colleges and universities. The African Americans and Hispanics who most often possess the backgrounds and credentials to succeed in the most prestigious colleges and universities are not likely to be the most economically disadvantaged. This paper examines the challenges that African Americans and Hispanics face as they pursue greater representation in the higher levels of America’s educational meritocracy. Included in these analyses are the trends in representation of African Americans and Hispanics relative to Whites and Asians, and their performance and the performance of their educational institutions on important indicators of educational merit. Additional criteria are introduced that are likely prospects for increasing the representation and achievement of underrepresented students. These additional criteria are correlates of test and assessment scores and other measures of academic performance. The focus is on the accumulation of human intellectual capital from the early years of preschool through high school and the college years and then the movement from college into first-professional education. The paper concludes with a brief description of four programs that intervene with African-American and Hispanic children during their middle and early high school years to increase the quality of their academic preparation.

OCR for page 143
The Right Thing to do, The Smart Thing to do Enhancing Diversity in the Health Professions Four central questions are addressed in the paper: What is the current status of population diversity in the United States and at various levels of education? As the primary mediums of meritocracy, how are tests and assessments performing in achieving equality of representation for various under-represented groups? What are the correlates of student performance on tests and assessments that are likely to lead to higher performance and consequently to greater access and achievement by underrepresented population groups? What actions are needed for expanding merit and eliminating underrepresentation based upon class and race/ethnicity over the long term and what criteria can be used by colleges and universities to identify additional talent in the short term? COMPOSITION OF THE U.S. POPULATION AND ENROLLMENTS IN HIGHER EDUCATION An important symbol in American meritocracy is the academic degree attainment of the adult population. The U.S. population consists of approximately 285,000,000 people, approximately 197,412,000 (69%) of whom are adults 18 years old and older. Figure 1 illustrates the distribution of African Americans, Asian Americans, Hispanics, Native Americans and Whites throughout various age groupings of the entire population from birth through 65 years and above. African Americans and Hispanics comprise a larger share of the school-age population than of the overall adult population, and Whites are represented better at higher age levels than in the younger ages. For example, African Americans and Hispanics constitute 13.9 percent and 18.1 percent, respectively, of children in the category of birth to 5 years, 12 percent and 10.6 percent in the range of 35 to 44, and only 8.1 percent and 5.3 percent in the 65 and above range. On the other hand, Whites are better represented at higher age levels from 62.7 in the earliest ages of 0 to 5 to 72.8 percent in the 35- to 44-year-old age range and 83.9 percent in the age range of 65 and above. Asian Americans retain their representation at around 2.8 percent at each age range (U.S. Census Bureau, 2000a). Table 1 shows that among adults who are 18 years and older in the United States the educational attainments of African Americans and Hispanics are not commensurate with their representation in the population, whereas Asian Americans far exceed their representation at higher levels of educational attainment.

OCR for page 143
The Right Thing to do, The Smart Thing to do Enhancing Diversity in the Health Professions FIGURE 1 U.S. Population by Race and Age in 1999 SOURCE: U.S. Census Bureau, Population Division, Population Estimates Program, 2000.

OCR for page 143
The Right Thing to do, The Smart Thing to do Enhancing Diversity in the Health Professions TABLE 1 Highest Level of Education Attained by Persons Age 18 and Over, by Age, Sex, and Race/Ethnicity (Numbers in Thousands): March 1998 Persons age 18 and over Total Population Bachelor’s Master’s Professional Doctorate Total Population 197,412 30,087 9,295 2,586 1,869   15.2% 4.7% 1.3% 0.9% Race/Ethnicity   African 22,552 2,075 662 111 71 American 11.4% 6.9% 7.1% 4.3% 3.8% Hispanic 19,833 1,348 306 132 79   10.0% 4.5% 3.3% 5.1% 4.2% Other 8,658 2,001 646 199 156   4.4% 6.7% 6.9% 7.7% 8.3% Whites 146,369 24,663 7,681 2,144 1,563   74.1% 82.0% 82.6% 82.9% 83.6% Gender   Males 95,008 14,861 4,656 1,759 1,354   48.1% 49.4% 50.1% 68.0% 72.4% Females 102,402 15,228 4,640 827 514   51.9% 50.6% 49.9% 32.0% 27.5% \1\Civilian non-institutional population—Data not applicable or not available. NOTE: Data are based on a sample survey of the noninstitutional population. Although cells with fewer than 75,000 people are subject to relatively wide sampling variation, they are included in the table to permit various types of aggregations. Because of rounding, details may not add to totals. SOURCE: U.S. Department of Education, National Center for Education Statistics, 2000. Table 1 shows that while African Americans comprise 11.4 percent of the adult U.S. population, they represent only 6.9 percent of the bachelor’s degrees, 7.1 percent of the master’s degrees, 4.3 percent of the first-professional degrees, and 3.8 percent of the doctorate degrees in the United States. Hispanics comprise 10 percent of the adult population, but 4.5 percent of the bachelor’s, 3.3 percent of the master’s, 5.1 percent of the first-professional degrees, and 4.2 percent of the doctorates. Asian and other population groups (primarily non-resident aliens) account for only 4.4 percent of the U.S. adult population, but 6.7 percent of the bachelor’s degree holders, 6.9 percent of the master’s, 7.7 percent of the first-professionals, and 8.3 percent of the doctorates. Similarly, Whites comprise 74.1 percent of the U.S. adult population, but 82 percent of the bachelor’s degree holders, 82.6 percent of the master’s, 82.9 percent of the first-professionals, and 83.6 percent of the doctorates (U.S. Department of Education, 2000).

OCR for page 143
The Right Thing to do, The Smart Thing to do Enhancing Diversity in the Health Professions Table 1 also reveals an interesting sex difference in the attainment of merit among adults. Women and men comprise roughly the same share of bachelor’s degrees (50.6 percent female, 49.4 percent male) and master’s degrees (49.9 percent female and 50.1 percent male), but men hold a rather large advantage in the attainment of first-professional and doctorate degrees. Men constitute 68 and 72.4 percent of the first professional and doctorate degrees, respectively, compared to women who hold 32 and 27.5 percent (U.S. Department of Education, 2000). Table 2 shows the race/ethnic distribution of the traditional college-age population of 18- to 24-year-olds, first-time full-time freshmen enrollment at four-year colleges and universities, bachelor’s degree recipients, and the new medical school entrants in 1997. The gap between representation in the population and among the three indicators of merit for African Americans and Hispanics is similar to the gap observed in Table 1 for the overall adult population. African Americans represented 14.3 percent of the 18- to 24-year-old population but only 11 percent of the first-time full-time freshmen at four-year colleges and universities, 7.8 percent of bachelor’s degree recipients, and 7 percent of new entrants into medical school. Hispanics were 14.3 percent of the 18-to 24-year-olds in the population, 8.3 percent of the first-time full-time freshmen, 6.3 percent of the bachelor’s degree recipients, and 6.1 percent of new medical school entrants. Conversely, Asian Americans represented 3.9 percent of the 18- to 24-year-olds in the U.S. population, but 5.9 percent of the first-time TABLE 2 Population (18–24), First-Time Full-Time Freshmen, Bachelor’s Degree Recipients, First-Time Medical School Enrollment, 1997 Race Population and Degree Total Other African American Asian American Hispanic White U.S. Population Age (18–24) 24,980,036 222,857 0.9% 3,584,530 14.3% 973,449 3.9% 3,561,018 14.3% 16,638,182 66.6% First Time, FT FR 4-yr Enroll 1,153,336 33,247 2.9% 126,360 11.0% 67,876 5.9% 95,561 8.3% 830,292 72.0% BA Degree 1,188,385 74,822 6.3% 92,170 7.8% 67,452 5.7% 75,012 6.3% 878,929 74.0% New Medical School Entrants 16,165 587 3.6% 1,134 7.0% 3,131 19.4% 984 6.1% 10,329 63.9%   SOURCE: U.S. Census Bureau, Population Division, Population Estimates Program, 2000, 1997 IPEDS Enrollment Data, 1996–1997 IPEDS Completion Data, Association of American Medical Colleges.

OCR for page 143
The Right Thing to do, The Smart Thing to do Enhancing Diversity in the Health Professions full-time freshmen, 5.7 percent of the bachelor’s degree recipients, and 19.4 percent of the new entrants into medical school. Similarly Whites made up 66.6 percent of the 18- to 24-year-old population, 72 percent of the first-time full-time freshmen, 74 percent of the bachelor’s degree recipients, and 63.9 percent of the new entrants into medical school. It is important to observe that both African Americans and Hispanics are represented among new entrants into medical school at roughly the same rate that they are represented among bachelor’s degree recipients. This is due mainly to the lower representation of students who are non-resident aliens among new medical school entrants than among bachelor’s degree recipients and not because African Americans and Hispanics enter medical school at the same rate after undergraduate school as Asian Americans and Whites. In the scheme of American meritocracy, attending college and attaining a bachelor’s degree contribute to higher socioeconomic status. But, attending and attaining a degree from relatively prestigious colleges and universities yields advantage in the labor market and in American society (Bowen & Bok, 1998). Table 3 reveals that African Americans and Hispanics are underrepresented among first-time freshmen overall relative to their representation in the 18- to 24-year-old population, but their underrepresentation is most severe among the most prestigious colleges and universities. African Americans are even overrepresented in the least prestigious colleges and universities. Table 3 shows that African Americans comprise 6.3% and 5.2%, respectively, of students enrolled at the most competitive and highly competitive colleges and universities, but 19% and 15.3%, respectively, of the less competitive and the non-competitive colleges and universities. Hispanics comprise 5.5% in the category of most competitive colleges and universities and 6 percent in the highly competitive category. Asians are overrepresented in the most competitive (13.4%), highly competitive (11.7%), and very competitive categories (7.1%). White students are represented among first-time, full-time freshmen comparable to their representation in the 18- to 24-year-old population in the most competitive colleges and universities (69.8%), the highly competitive category (74.2%), and the other four categories. Among the nation’s practicing physicians, Table 4 shows that African Americans comprise just 3.2% and Hispanics 2.4%, which are far below their representation in the U.S. adult population. Even in the fields of internal medicine and obstetrics and gynecology, where African Americans and Hispanics are best represented among all the practice specialties, African Americans comprise just 3.2% and 6.4%, respectively, and Hispanics just 2.3% and 3.3%, respectively. A larger number of Asian Americans (25,441) were practicing physicians than both African Americans and Hispanics overall.

OCR for page 143
The Right Thing to do, The Smart Thing to do Enhancing Diversity in the Health Professions TABLE 3 First-Time Full-Time Freshmen Attending Non-Specialized Four-Year Colleges and Universities (N = 1,395) by Selectivity: 1997 Selectivity Grand Total* Total Race Groups African American Asian American Hispanic White U.S. Population ages 18–24 24,980,036 24,757,179 3,584,530 14.3% 973,449 3.9% 3,561,018 14.3% 16,638,182 66.6% Total First-time Freshmen Enrollment 1,063,710 1,033,677 119,465 11.2% 65,338 6.1% 62,970 5.9% 785,904 73.9% Most Competitive (52) 50,932 48,332 3,196 6.3% 6,814 13.4% 2,780 5.5% 35,542 69.8% Highly Competitive (90) 120,152 116,656 6,259 5.2% 14,051 11.7% 7,196 6.0% 89,150 74.2% Very Competitive (249) 234,596 228,301 13,669 5.8% 16,593 7.1% 13,693 5.8% 184,346 78.6% Competitive (583) 413,401 402,298 53,145 12.9% 20,034 4.8% 23,055 5.6% 306,064 74.0% Less Competitive (292) 156,342 152,527 29,666 19.0% 5,869 3.8% 10,625 6.8% 106,367 68.0% Non-Competitive (129) 88,287 85,563 13,530 15.3% 1,977 2.2% 5,621 6.4% 64,435 73.0% NOTE: Grand total for the U.S. population includes Native Americans. Grand totals for freshmen enrollment include Native Americans and non-citizens. Row percents are calculated from the grand totals. SOURCE: U.S. Census Bureau, Population Division, Population Estimates Program, 2000, IPEDS 1997 Fall Enrollment, Barron’s Profiles of American Colleges, 1999.

OCR for page 143
The Right Thing to do, The Smart Thing to do Enhancing Diversity in the Health Professions TABLE 4 Primary Practice Specialty by Race/Ethnicity of U.S. Medical School Graduates Practice Specialty African American Asian American Hispanic Native American All Minorities All Physicians Colon & Rectal Surgery 16 29 22 1 68 1,033 Family Practice 2,410 1,979 1,992 278 6,659 64,611 Internal Medicine** 5,031 7,770 3,653 192 16,646 157,450 Medical Genetics 4 7 3 --- 14 250 Nuclear Medicine 23 43 42 --- 108 1,434 Obstetrics-Gynecology 2,526 1,515 1,303 82 5,426 39,257 Total 20,895 25,441 15,703 1,126 63,165 654,748 Internal Medicine is the most popular primary practice specialty overall as well as the most popular primary practice specialty for African Americans and Hispanics. SOURCE: AAMC: Minority Graduates of U.S. Medical Schools: Trends, 1950–1998. TABLE 5 Typical Undergraduate College Criteria for Admission Test Scores High School Class Rank High School Grade Point Averages in College Preparatory Courses High School Academic Program Student Essay Geography Alumni Relations Extraordinary Talent TABLE 6 Typical Medical School Criteria for Admission Test Scores (MCAT) College Curriculum/Major College Grade Point Averages Pre-Med Courses (Biology, Physics, English, Chemistry) Overall Quality (Selectivity) of College Attended Faculty Recommendations

OCR for page 143
The Right Thing to do, The Smart Thing to do Enhancing Diversity in the Health Professions Representation among new entrants into college, among students attending prestigious colleges and universities, among bachelor’s degree recipients, among new entrants into medical school, and among practicing physicians are important markers of merit in American education. The fact that African Americans and Hispanics are severely underrepresented is problematic for achieving representative diversity at upper levels of the educational status hierarchy. Understanding the relative interests and aspirations of African Americans, and their achievement and performance at prior levels of education, is important in addressing their challenges to equality and representation in higher levels of education and in the overall hierarchy of American meritocracy. PARTICIPATION AND PERFORMANCE OF UNDERREPRESENTED STUDENTS ON EDUCATIONAL TESTS AND ASSESSMENTS Table 5 presents the criteria for admissions that are commonly used by colleges and universities to select undergraduate students. Table 6 presents the same for medical school admissions. The weight assigned to each criterion depends upon the heterogeneity of the applicant pool at each college, university, or medical school. The more homogeneous the applicant pool on any given criterion, the less weight the criterion has on decisions that the institution makes concerning whom to admit. Some of the criteria that colleges and universities rely on for making undergraduate, graduate, and first-professional degree program admissions decisions are good indicators of what American society accepts as indicators of merit. This is especially so in the case of grades and test scores, which are the two criteria that are used at every level of education as the most important for admissions decisions. The remaining criteria, such as geographical location of the hometown and special talents, are most often used for the purpose of achieving diversity in the student body. The weight that colleges and universities assign to test scores and grades is more often a reflection of the type of information they need in order to discriminate among applicants within their applicant pool. Even though test scores and grades are typically required in admissions applications at all levels, their weight among the admissions criteria vary from one institution to the next. In fact, it is not surprising for less weight to be assigned to test scores in the most selective colleges than at less selective ones, because the weight is most often dependent upon the variance in the applicant pool rather than the value that colleges and universities believe the test has as a representation of student achievement. In colleges and universities that have large applicant pools and where the test scores of the applicant pool are homogenous, the weight assigned to scores is likely to be low and more weight may be placed on alternative criteria that provide more discrimination among the applicants. Conversely, tests may be the

OCR for page 143
The Right Thing to do, The Smart Thing to do Enhancing Diversity in the Health Professions income, fathers’ education, whether English is the primary language, a measure of the courses taken in high school and their performance in those courses, their intended college major field, attendance at a public or private high school, the availability of Advanced Placement (AP) courses, and the proportion of African-American and Hispanic enrollment. Tables 13 and 14 depict the relationship of these variables to students’ scores on the ACT and the SAT, respectively. Table 13 presents the regression findings for the ACT assessment. The model accounted for 50% of the variance in students’ ACT composite score achievement. The strongest predictor overall (based on the t statistic) was the measure of high school courses taken and grades achieved. This measure includes the number of years that a student took courses in the subject areas of math, English, social science, natural science, foreign language, and art/music (students were given an additional point for taking honors courses in a subject), multiplied by the average grade achieved in the subject. Socioeconomic measures were associated with ACT performance. Students whose first language is English scored .9 points higher than those whose first language is not English. Compared with students whose family incomes were $80,000 or more per year, students with family incomes of $30,000 or less scored, on average, 1 point lower on the ACT. Those with family incomes of $30,000 to $60,000 scored .5 points lower, and those with family incomes of $60,000 to $80,000 scored .3 points lower. Compared to Whites, on average, African-American students scored 1.9 points lower, Asian students scored .4 points lower, Hispanic students scored 1 point lower, Native American students scored .6 points lower, non-citizens scored 1.5 points lower, and other citizens scored .5 points lower. Males scored .7 points higher than females. High school experiences and characteristics were also significant predictors of ACT performance. For every point they had on the high school curriculum and achievement measure, students scored .15 points higher on the ACT. Students intending to major in a subject other than natural science scored lower, on average, on the ACT. This difference was strongest for students intending to major in other disciplines (they scored 1.2 points lower) and in health sciences (1 point lower). Students achieved .1 point higher on the ACT for each AP course offered at their high school. Students attending public schools achieved .7 points lower on the ACT than students attending private schools. Each 10 percent increment in African-American enrollment at the high school was associated with achieving .16 points lower on the ACT. Each 10 percent increment in Hispanic enrollment at a particular high school was associated with achieving .27 points lower on the ACT. As shown in Table 14, after controlling for all the other variables in the model, the combination of course work and student grade achievement was the most substantial predictor of students’ SAT scores. However, this variable was of somewhat less relative importance as a predictor in the SAT model than the ACT model. This is a reflection of differences in the nature of these tests. The

OCR for page 143
The Right Thing to do, The Smart Thing to do Enhancing Diversity in the Health Professions SAT is intended to measure students’ aptitude for college-level work while the ACT is intended to measure students’ mastery of high school-level curriculum. Other prominent predictors of SAT scores were fathers’ education, sex, and the number of AP courses offered in the school. The model accounted for 52 percent of the variance in students’ performance on the SAT. There was a positive relationship between measures of students’ socioeconomic status and SAT performance. Compared with students whose fathers had completed a bachelor’s degree and some graduate education, students whose fathers were high school graduates or less scored on average 64 points lower on the SAT. Students whose fathers had some college but were not bachelor’s degree recipients achieved, 45 points lower and those whose fathers were bachelor’s degree recipients achieved scores 25 points lower. Compared to students whose family incomes were $80,000 or more per year, those with family incomes of $30,000 or less scored, on average, 36 points lower on the SAT; those with family incomes of $30,000 to $60,000 scored 18 points lower; and those with family incomes of $60,000 to $80,000 scored 14 points lower. Compared with Whites, on average, African-American students scored 68 points lower, Asian students scored 7 points more, Hispanic students scored 33 points lower, Native American students scored 22 points lower, non-citizens scored 42 points lower, and other citizens scored 8 points more on the SAT. Students whose first language is English scored 13 points higher than those whose first language is not English. On average, males scored 56 points higher on the SAT than females. Students’ experiences and achievements in high school are also important predictors of SAT performance. For every point they had on the high school course measure, students achieved 5.5 more points on the SAT. Students intending a college major other than natural science scored lower, on average, on the SAT. This difference was strongest for students intending to major in education (they scored 57 points lower) and students intending another major (56 points lower). High school characteristics were also related to students’ SAT performance. Students achieved an average of 4 points more on the SAT for each AP course offered at their high school. Students attending public schools achieved 29 points lower on the SAT than students attending private schools. Each 10 percent increment in African-American enrollment at the high school was associated with achieving 6 points less on the SAT. Each 10 percent increment in Hispanic enrollment at the high school was associated with achieving 9 points lower on the SAT. A statistical method known as bootstrapping was conducted to learn more about the factors that distinguish high-performing test-takers from their lower-performing counterparts who are in the same general socioeconomic status. The bootstrapping technique is described in Appendix A. The analyses presented in this paper are for 1999 test-takers whose family incomes were between $30,000 and $60,000 and whose fathers’ education was at least a bachelor’s degree. The analyses are also confined to the following four subpopulations of test takers:

OCR for page 143
The Right Thing to do, The Smart Thing to do Enhancing Diversity in the Health Professions African-American females, African-American males, white females and white males. Bootstrapping was used to identify predictor variables that significantly increased or decreased the test score gap between high-performing students (those who achieved a combined math and verbal score at or above 1200 on the SAT) and low-performing students (those who achieved a combined score less than 1200). The analyses in Tables 15 through 18 reveal that four measures contributed significantly to the test score gap between high- and low-performing students in all four subgroups: taking calculus, taking physics, participation in an academically-oriented extracurricular activity, and taking and achieving high grades in a comprehensive academic curriculum. Only measures that were significant predictors are included in the tables. The tables can be interpreted as follows: measures with a positive coefficient were associated with a significant increase in the test score gap between the high- and low-performing students, while measures with a negative coefficient were associated with a significant decrease in the test score gap. Holding all other variables in the analyses constant, if high-performing students took calculus and low-performing students did not, the average gap in their test score performance increased by 36.41 points for African-American females, 35.38 points for African-American males, 21.00 points for white females, and 27.96 points for white males. Conversely, if high-performing students did not take calculus and low-performing students did, the average gap in their test performance decreased by 34.74 points for African-American females, 53.80 points for African-American males, and 33.01 points for white males. A similar pattern was associated with taking physics: if high-performing students took physics and low-performing students did not, there was a significant increase in the average test score gap between the two groups for both races and sexes. In general, participation in academically-oriented activities had a beneficial association for low-performing students; the gap between their test scores and those of their higher-performing peers was significantly decreased. Finally, differences in overall course taking and achievement accounted for significant changes in the test score gap between high- and low-performers. For every unit increase in this curriculum measure (values ranged from 1 to 100), that gap increased by approximately 3 points for African-American females and males, and 2 points for white females and males. Other significant predictors were specific to sub-groups in the analyses. Table 15 reveals that for African-American females, taking honors English, attending a public versus a private high school, and the percentage of white enrollment at the high school were associated with significant differences in the test score gap between high- and low-performing students. Table 16 shows that participating in athletic extracurricular activities, the census region in which

OCR for page 143
The Right Thing to do, The Smart Thing to do Enhancing Diversity in the Health Professions TABLE 15 Bootstrapping Results for African-American Females Taking the 1999 SAT Predictors of Differences in SAT V+M score   95% Confidence Int Estimate SE SE Median T Lower Upper (Intercept) 260.63 14.26 15.16 17.22 232.13 287.55 Calculus = 1 36.41 11.08 10.62 3.44 14.65 56.50 Physics = 1 55.77 10.34 10.13 5.52 36.04 77.69 Honors English = 1 27.26 11.09 10.68 2.56 5.38 48.61 Academic activity = 3 −33.30 13.40 13.89 −2.39 −59.36 −5.88 High school = 1 −35.41 14.84 16.00 −2.18 −62.28 −4.67 Curriculum measure 2.91 0.26 0.26 11.09 2.44 3.44 % White enrollment 67.12 11.97 11.98 5.59 42.72 89.13 NOTE: Only the significant predictors of differences in 1999 SAT V+M scores are presented. Calculus = 1 means the >1200 student took calculus and the <1200 student did not. Physics = 1 means the >1200 student took physics and the <1200 student did not. Honors English = 1 means the >1200 student took Honors English and the <1200 student did not. Academic activity = 3 means the >1200 student received an award in an academically oriented extracurricular activity and the <1200 student participated in this type of activity. High school = 1 means the >1200 student attended a private school and the <1200 student attended a public school. SOURCE: Nettles & Millett Analyses of customized data files from the College Board and Educational Testing Service, 1999. students lived, and the percentage of white enrollment made significant contributions to differences in the test scores of high- and low-performing African-American males. Tables 17 and 18 reveal that athletic activity participation, attending a public or private high school, and racial composition of the high school were not significant correlates of increasing or decreasing the test score gap for white female and male students. For these subgroups, significant predictors were restricted to measures of high school course work, extracurricular academic involvement, and grade achievement.

OCR for page 143
The Right Thing to do, The Smart Thing to do Enhancing Diversity in the Health Professions TABLE 16 Bootstrapping Results for African-American Males Taking the 1999 SAT Predictors of Differences in SAT V+M Scores   95% Confidence Int Estimate SE SE Median t Lower Upper (Intercept) 267.27 15.14 15.83 16.89 238.03 296.73 Calculus = −1 −53.80 16.67 18.60 −2.88 −86.12 −18.56 Calculus = 1 36.41 11.08 10.62 3.44 14.65 56.50 Physics = 1 51.45 10.37 10.36 4.97 31.20 71.64 Academic activity = −3 −63.91 20.20 22.93 −2.79 −104.07 −24.73 Athletic activity = 3 36.20 14.42 15.16 2.39 7.13 64.56 Athletic activity = 4 −44.99 21.24 19.53 −2.32 −85.84 −0.08 Census region −32.19 15.38 15.95 −2.02 −61.43 −0.87 Curriculum measure 2.70 0.27 0.27 10.14 2.15 3.24 % White enrollment 39.89 12.22 12.42 3.22 16.13 64.38 NOTE: Only the significant predictors of differences in 1999 SAT V+M scores are presented. Calculus = −1 means the >1200 student did not take calculus and the <1200 student did. Calculus = 1 means the >1200 student took calculus and the <1200 student did not. Physics = 1 means the >1200 student took physics and the <1200 student did not. Academic activity = −3 means the >1200 student participated in an academically oriented activity and the <1200 student received an award in this type of activity. Athletic activity = 3 means the >1200 student received an athletic award and the <1200 student participated in an athletic activity. Athletic activity = 4 means the >1200 student received an athletic award and the <1200 student did not participate in an athletic activity. SOURCE: Nettles & Millett Analyses of customized data files from the College Board and Educational Testing Service, 1999.

OCR for page 143
The Right Thing to do, The Smart Thing to do Enhancing Diversity in the Health Professions TABLE 17 Bootstrapping Results for White Females Taking the 1999 SAT   SE Median   95% Confidence Int Predictors of Differences in SAT V+M Score Estimate SE t Lower Upper (Intercept) 214.05 14.24 14.29 14.99 186.69 241.94 Calculus = −1 −35.32 15.14 18.39 −1.91 −63.98 −5.17 Calculus = 1 21.00 9.39 9.60 2.19 2.65 39.07 Physics = 1 29.15 9.69 9.48 3.08 10.60 49.14 Honors English = 1 30.56 9.52 9.79 3.11 11.60 49.17 Academic activity = 4 30.19 15.79 15.29 1.98 0.03 61.34 Curriculum measure 2.07 0.25 0.25 8.41 1.60 2.57 NOTE: Only the significant predictors of differences in 1999 SAT V+M scores are presented. Calculus = −1 means the >1200 student did not take calculus and the <1200 student did. Calculus = 1 means the >1200 student took calculus and the <1200 student did not. Physics = 1 means the >1200 student took physics and the <1200 student did not. Honors English = 1 means the >1200 student took Honors English and the <1200 student did not. Academic activity = 4 means the >1200 student received an award in an academically oriented activity and the <1200 student did not participate in this type of activity. SOURCE: Nettles & Millett Analyses of customized data files from the College Board and Educational Testing Service, 1999. TABLE 18 Bootstrapping Results for White Males Taking the 1999 SAT   95% Confidence Int Predictors of Differences in SAT V+M Score Estimate SE Median t Lower Upper (Intercept) 221.55 14.05 14.58 15.16 193.56 248.77 Calculus = −1 −33.01 15.38 18.54 −1.80 −63.88 −1.88 Calculus = 1 35.38 10.29 10.79 3.28 15.21 55.52 Physics = 1 47.10 10.43 10.03 4.68 27.48 67.22 Academic activity = 4 39.91 15.94 15.28 2.60 9.05 72.00 Curriculum measure 1.64 0.24 0.24 6.79 1.17 2.12 NOTE: Only the significant predictors of differences in 1999 SAT V+M scores are presented. Calculus = −1 means the >1200 student did not take calculus and the <1200 student did. Calculus = 1 means the >1200 student took calculus and the <1200 student did not. Physics = 1 means the >1200 student took physics and the <1200 student did not. SOURCE: Nettles & Millett Analyses of customized data files from the College Board and Educational Testing Service, 1999.

OCR for page 143
The Right Thing to do, The Smart Thing to do Enhancing Diversity in the Health Professions TYPES OF INITIATIVES NEEDED FOR EXPANDING MERIT AND ELIMINATING UNDERREPRESENTATION School curricula have the highest probability of possibly supplanting or complementing test scores as indicators of merit. Students who experience the most rigorous curricula in school and earn relatively high grades are promising prospects for success in college and graduate and professional school. Among the best organizations in the nation are A Better Chance, Bank Street College’s ILEAD program, the Center for Talented Youth (CTY), and Prep for Prep. Although each one of these programs are unique, the goal of each is to identify students early, provide them with rigorous academic instruction, encourage them to adopt intellectual habits, move them to high-caliber independent and public schools, and guide them into the highest academic tracks in the schools that they attend. In each program—with the exception of the ILEAD—one objective is to identify promising young students at around fifth, sixth, or seventh grade and begin to compensate for the low level of academic preparation they receive in their local public schools. In Prep for Prep, students attend mathematics, history, English, and science classes taught by master teachers throughout the school year on Saturdays from 8 a.m. to 5 p.m. and every Wednesday from 4 p.m. to 6 p.m. The students also devote an intensive three weeks to further instruction during the summers. Standardized tests are used to identify and invite the students to attend and are used throughout the program to measure their progress. The goal of the program is to prepare and then to help students gain admission into independent boarding schools by ninth grade or into the AP and honors tracks of their public high schools. A Better Chance also identifies students as early as the fifth and sixth grades. Rather than providing instructions and a curriculum of its own, A Better Chance matches students with independent day schools, boarding schools, or a dozen or so outstanding public schools. The Center for Talented Youth at Johns Hopkins University is probably the oldest, most established, and largest of the programs. It has an array of academic-intensive courses for students, mainly during summers, as well as on-line instruction, workshops, and mentoring during the school year. ILEAD works with four Catholic schools in the Bronx, New York, helping to enrich their curriculum and providing professional development for their teachers. Each of these four aims to prepare students for admissions to the nation’s most selective colleges and universities. Each one recognizes that success in the competitive admissions process and persistence through completion requires the highest quality of academic preparation and competitive test scores. It is only through academic preparation that the academic and socioeconomic playing field can be leveled for underrepresented African-American and Hispanic youth, and that the colleges and universities at the height of America’s meritocracy can become more diverse.

OCR for page 143
The Right Thing to do, The Smart Thing to do Enhancing Diversity in the Health Professions CONCLUSION Standardized intellectual tests and assessments are important barometers of underrepresented student achievement in America’s educational meritocracy. The inter-correlation of individual, group, and institutional test and assessment scores reveals uniformity among these instruments and shows how the nation’s social hierarchy is structured. The nation’s meritocracy is constituted by all three types of tests and assessments. Underrepresented groups must focus upon all three in order to advance in the nation’s educational meritocracy. Regardless of the weight that colleges and universities assign to tests and assessments in the college admissions process, they remain the most challenging and most important focus for African Americans and Hispanics. Unless African Americans and Hispanics are able to close the achievement gap revealed by these assessments, they will not be able to expand their access to the most selective colleges and universities and first-professional schools, and they will remain underrepresented in the highest-status professions. These tests and assessments can be important in the long term for developing public policies that are aimed toward improving educational outcomes, and in the short-term for extending the benefits of merit to a broader representation of the population. APPENDIX A How Bootstrapping Was Implemented for These Data We used this iterative re-sampling method in order to better understand the inherent difference between high potential students (SAT V+M ≥ 1200) and everyone else. In general, the bootstrap method calls for selecting from the observed sample data a random sample of size n with replacement. Then the bootstrap method calculates the estimated parameter using the same analysis method for each sample drawn. The mean or median of the bootstrap re-sample estimates can be used as the estimated parameter value. Confidence intervals about the parameter value can be produced by taking quantiles (2.5th and 97.5th for a 95% CI) of the bootstrap re-sample estimates.1 For example, in a simple case, say you have 20 observations and you want a bootstrap estimate of a confidence interval about the true mean of that parameter. This could happen in a case where you do not know the underlying distribution of the parameter. You re-sample with replacement 20 data points from the original 20 observations. There, of course, will be repeat observations in each bootstrap re-sample. You could then follow the algorithm above to get a confidence for the true mean. 1   Source: Neter et. al, Applied Linear Statistical Models, 1996.

OCR for page 143
The Right Thing to do, The Smart Thing to do Enhancing Diversity in the Health Professions For our data, we would like to investigate differences between high potential students and non-high potential students conditioning on certain uncontrollable factors. The variables that we decided to use were father’s level of education, parental income range, race and sex. After splitting the file on these uncontrollable variables, create HiPo (SAT V+M ≥ 1200) and ~HiPo (SAT V+M < 1200) data sets for each covariate class (or cell, if you think of these as crosstabs). There would be 4 (number of levels of father’s education) × 4 (number of levels of parental income) × 4 (race groups) × 2 (sexes) =128 total separate bootstraps to do. However, this would take too much time so we decided to do a subset of the combinations of father’s education by parental income. However, within each of those subsets, we would do all combinations of race by sex. Randomly select a person from the the HiPo and ~HiPo data sets for a specific covariate class derived after splitting the large data set for uncontrollable variables. Compute the difference between the HiPo and ~HiPo student on SAT V+M and the predictor variables. Repeat (1)—(2) 1,000 times to create bootstrap data set. Regress the differences in the SAT V+M difference, using the 1,000 observations from (3). Store the estimated coefficients. Repeat (1)—(5) 1,000 times. The result of these 6 steps will be 1,000 estimates for the coefficients and standard errors of the predictors. Take the median of those 1,000 estimates for the coefficient and the standard errors as the final parameter estimate for that coefficient and that standard error. Use the 2.5th and 97.5th quartile of the 1,000 estimates of the coefficients to derive a 95% confidence interval for the parameter estimate. Also, using the 2.5th and 97.5th quartile of the 1,000 estimates of the standard errors will yield a 95% confidence interval for the standard error parameter estimate. If a confidence interval does not include 0, call the variable significant. Assumptions made: After controlling for the unchangeable variables the students in HiPo are “the same,” and the students in ~HiPo are “the same” except for the uncontrollable factors. The HiPo and ~HiPo data sets are representative of the population, after eliminating cases with missing values on any variable. Results We used this methodology to derive our estimates for significant factors in predicting differences in SAT scores between high potential and non-high potential students. The predictors that we initially included in the model (after

OCR for page 143
The Right Thing to do, The Smart Thing to do Enhancing Diversity in the Health Professions splitting the file on the uncontrollable variables) were took calculus (Y or N); took physics (Y or N); took honors English (Y or N); academic activity status; athletic activity status; census region (4); went to public/private school; high school curriculum measure; student teacher ratio; percentage of white students in school. REFERENCES Association of American Medical Colleges (1998). April/August 1997 MCAT Performance by Sex, Racial/Ethnic Group, Age, Language Status, Undergraduate Major and Testing History [On-line]. Available: http://www.aamc.org/students/mcat/scores/examineedata/sum2000.pdf. Association of American Medical Colleges (1998). [MCAT test takers]. Unpublished raw data. Association of American Medical Colleges. (2000). Minority Graduates of U.S. Medical Schools: Trends, 1950–1998. Washington, DC: Association of American Medical Colleges. Barron’s profiles of American colleges. (1999). Woodbury, NY: Barron’s Educational Series, Inc. Bowen, W.G., & Bok, D.C. (1998). The shape of the river: Long term consequences of considering race in college and university admissions. Princeton, NJ: Princeton University Press. Donahue, P.L., Voekl, K.E., Campbell, J.R., and Mazzeo, J. (1999). The NAEP 1998 Reading report card for the Nation and the States. NCES 1999–500. Washington, DC: National Center for Education Statistics. Educational Testing Service (1997). [1996 NAEP Mathematics results]. Unpublished raw data. Educational Testing Service, Graduate Examinations Board Programs. [Graduate Records Examination results, 1984–1995]. Unpublished raw data. Law School Admission Council (2000). [Applicant Counts by Ethnic and Gender Group—1984–85 to Fall 2000]. Unpublished raw data. Nettles, M.T. & Millett, C.M. (2001). [Customized data files from ACT, Inc., 1999]. Unpublished raw data. Nettles, M.T. & Millett, C.M. (2001). [Customized data files from the College Board and Educational Testing Service, 1999]. Unpublished raw data. Neter, J., Wassweman, W., & Kutner, M.H. (1983). Applied linear regression models. Homewood, ILL: R.D.Irwin. Reese, C.M., Miller, K.E., Mazzeo, J. & Dossey, J.A. (1997). The NAEP 1996 Mathematics report card for the Nation and the States. Washington, DC: National Center for Education Statistics. U.S. Census Bureau, Population Division, Population Estimates Program (2000). Population Estimates for States by Age, Race, Sex, and Hispanic Origin: July 1, 1999 [Online] . Available: http://www.census.gov/population/estimates/state/sasrh/sasrh99.txt. U.S. Census Bureau, Population Division, Population Estimates Program (2000). Population Estimates for States by Age, Race, Sex, and Hispanic Origin: July 1, 1997 [Online] . Available: http://www.census.gov/population/estimates/state/sasrh/sasrh97.txt. U.S. Department of Education. National Center for Education Statistics. (2000). Digest of education statistics, 1999. (NCES 2000–031). Washington, DC: U.S Government Printing Office

OCR for page 143
The Right Thing to do, The Smart Thing to do Enhancing Diversity in the Health Professions U.S. Department of Education. National Center for Education Statistics. (1997). Integrated postsecondary education data system. Washington, DC: U.S. Government Printing Office.