What Doesn't Meet the Eye

Endnotes

¹For a discussion of racial achievement disparities and trends in disparity, see Ferguson (2001). Also see various publications of the National Center for Education Statistics at http://nces.ed.gov/. For other recent collections and overviews concerning racial achievement disparities and potential actions to reduce them, see Jencks and Phillips (1998), National Task Force on Minority High Achievement (1999), and Walker-James, Jurich, and Estes (2001).

²For a recent report on cities that are making progress, see Snipes, Doolittle, and Herlihy (2002).

³Source: Calculated by the author from data available at http://factfinder.census.gov, Census 2000 Summary File 1: Detailed Tables, Tables P12A, B, D, and H. The numbers pertain to children ages 19 and under and exclude children for whom more than one race/ethnic category was indicated.

⁴Since at least the time of Brown v. Board of Education (1954), the belief among many has been that nonwhite children would excel in school if only they could have access to the same high-quality classrooms that white children attended. Indeed, Kain and Persky (1969), and others over the years, have sometimes argued against "ghetto development" because, they assumed, educational and other opportunities for blacks and others were greater in the suburbs. Definitive evidence on whether black children in fact do better in the suburbs has been slow to accumulate because of data and methodological problems. See my discussion of this in Ferguson (2002) and references included there. An interesting recent paper on the effects of racial integration is Hanushek, Kain, and Rivkin (2001). They find that increasing the percentage of blacks in a school has the most adverse effects on high-achieving blacks.

⁵Professor John Bishop of Cornell University developed the survey instrument.

⁶A few schools surveyed sixth and 12th graders, but since only a small minority of the districts did so, sixth and 12th graders are not discussed in the present paper. In most instances, schools administered the questionnaire to all students in each surveyed grade who were present when the survey was given. Some districts surveyed only selected grades, such as seventh, ninth, and 11th.

⁷"Mixed race" means that the student listed more than one race or ethnicity. There were many mixtures, about 40 percent of which were white mixed-race with one other group. Only 17 percent were black mixed-race with white. Some students indicated more than two groups. Hispanics and Asians reported more mixing with other groups than blacks or whites did.

⁸Observations with missing data for any given variable are not included in tabulations of that variable. Generally, the number of missing observations for any given variable is small.

⁹By class "level," I mean to distinguish whether students are taking honors or AP courses from whether they are not taking any. Whites and Asians enroll in honors and AP classes more often than blacks and Hispanics. Honors and AP classes typically require more homework, and students who take such courses spend more time on homework than students who do not—no matter what their race/ethnic group. Racial differences in rates of enrollment in honors and AP courses are not perfectly understood but seem due mostly to differences in academic proficiency. Most schools are working to increase black and Hispanic enrollments in honors and AP courses. However, there also are efforts underway to improve the quality of regular courses, so as to stem the flight of primarily white students away from them and to provide a higher-quality education to students who take them.

¹⁰Because student surveys did not include official identification numbers, the author is not able to merge the data with standardized test scores or data from official transcripts.

¹¹The survey asked about specific letter grades and made no mention of whether grades for honors and AP courses might receive differential weighting in computing grade-point averages. There is no way of knowing whether honors and AP students might have inflated the letter grades they reported to account for differential weighting.

¹²This was an extra question only for MSAN schools and was not on the printed Ed-Excel survey. Responses were entered in an extra response space at the end of the survey form.

¹³Here, "about half or less" represents a composite of three options on the survey: "about half," "some," and "very little or none."

¹⁴The data also include the fathers' years of schooling, but there are fewer missing values for the mothers'. The parents' education variables used in multiple-regression estimates combine mothers' and fathers' education data, and replace missing values using standard methods.

¹⁵See, for example, various chapters in Jencks and Phillips (1998) and references included therein.

¹⁶This finding comes from a multiple-regression analysis with fixed effects to control for school and grade level. Explanatory variables were the family background measures listed in Table 2, except that the parental education variable in regressions included the father's education as well. Missing values of explanatory variables were replaced with missing-value dummy variables.

¹⁷Using the standard deviation for the whole data set, across all schools, grades, and racial groups.

¹⁸Since they would correlate strongly with the variables we do have, it is uncertain how much more (or less) achievement disparity we could predict if such variables were included. Similarly, there is no way to know how much the findings might change if student responses were free of any errors or biases.

¹⁹To form the SES categories, I began by using all of the SES measures in the data, but not race, to predict GPA. This multiple regression produced regression coefficients to use as weights in composite SES measures. The equation used a dummy variable for each value of each SES variable, in order to allow for nonlinearity in estimated effects. The equation also included school-grade-level fixed effects and gender. Missing values for explanatory variables were handled using dummy variables. The adjusted R-square for the equation was 0.23. Using the results, some students' SES characteristics (ignoring race/ethnicity) put them in the bottom 10 percent of predicted GPAs. I labeled this group the "lowest SES" group. Others' characteristics predict that they would be in the 40 percent of the distribution from the 10th to the 50th percentile (labeled "lower-middle SES") or in the 40 percent of the distribution from the 50th to the 90th percentile (labeled "upper-middle SES"). Finally, some would be in the top 10 percent, and this group is labeled "highest SES."

²⁰Multiple regressions for each achievement measure were estimated separately for each race/ethnic group, and the coefficients from these regressions were used to predict race/ethnic-specific achievement levels for each SES profile.

²¹See, for example, McWhorter (2000).

²²For each subject, the Ed-Excel survey question asked, "When teachers assign homework or after-school work, how much of it do you usually do?" Students could indicate "homework is never assigned," "none of it," "some of it," "most of it," "all of it," and "more than required." For each subject, I created an index with three values. If the student's answer was "homework is never assigned," "none of it," or "some of it," the index had a value of 3. If the answer was "most of it," the value was 4, and if the answer was "all of it" or "more than required," the value was 5. Combining these across the four subjects creates an index with values ranging from 12 to 20. For Table 5, values of 15 or lower are labeled "some, or not much of it," values of 16 to 19 are labeled "most of it," and values of 20 are labeled "all of it."

²³Analysis of whether time on homework and homework-completion rates correlate with the GPA that students report for the end of the previous term shows that both homework measures help in predicting it. However, recall that there are few race/ethnic differences in time on homework among classmates. Therefore, homework-completion rates, but not time on homework, help in predicting race/ethnic GPA gaps among students who take the same courses. Controlling for nothing but school and grade-level fixed effects, homework-completion rates predict about one-sixth of the GPA gap between whites and blacks, about one-fifth of the gap between whites and mixed-race students, and one-fourth of the gap between whites and Hispanic students, for those not enrolled in honors or AP classes. The analogous fractions are lower among students enrolled in honors or AP classes. About two-thirds to three-quarters of race/ethnic differences in homework completion are predicted by the MSAN Ed-Excel SES variables discussed above.

²⁴Another item in the table that shows race/ethnic differences is, "I don't want to embarrass my family." Here, the response ranges from 15 percent for whites to 33 percent for Asians, with blacks and Hispanics more like Asians than whites.

²⁵For other evidence of racial differences in motivational processes see Boykin and Bailey (2000); Steele and Aronson (1998); and Jussim, Eccles and Madon (1996).

²⁶Indeed, the racial differences in some districts are large enough to deserve special attention, but to elaborate on this point is beyond the scope of this paper.

²⁷Among MSAN districts, interesting research in this regard is being done at Oak Park-River Forest High School in Illinois, where a team of teachers and researchers is carefully studying the school experiences and academic records of that high school's students.

²⁸For example, Supovitz and Poglinco (2001) conducted case studies of principals identified by their peers as outstanding instructional leaders. Among other things that they had in common, these instructional leaders cultivated a community of instructional practice, creating safe and collaborative environments for teachers to engage with one another and also with a wide range of outside actors to deepen the work. Similarly, Spillane (2002) describes a "situated" approach to teacher learning that engages teachers in constructing knowledge, playing roles as leaders, being active learners, and participating in activities that stress the social aspects of learning. Topics are integrated around areas of potential reform, with both internal and external actors providing guidance, and activities utilizing a curriculum involving several types of devices, including teachers' own practice. The approach is social rather than individualistic and it promotes teachers' identities as learners in school and classroom contexts where relationships matter a great deal.

²⁹For Erikson's tasks and stages of identity development, see Erikson (1963). For Tuckman's model of group process, see Baron, Kerr, and Miller (1992, p. 14). On innovation diffusion, see Rogers (1983). For a conceptually structured case example of social work with gangs, see Fox (1985). For a discussion of developing customer relations in marketing, see Chapter 19 of Kotler (1986). For examples where I have applied the same model in other contexts, see Ferguson and Snipes (1996, available from the author) and Ferguson (1999, pp. 589-604).

³⁰The one hour includes a 10-minute stage-specific video; 10 minutes for teachers to respond in writing to a short list of prompts; 15 minutes for a panel of five students to respond to some prompts and take questions from the audience; then 25 minutes for open discussion among the faculty, leading to smaller group activities afterward.

³¹Selected schools from a majority of MSAN districts as well as schools from an equal number of other districts are participating in the professional development, the research component, or both.

Previous | Table of Contents | Next