Effect of School Poverty on Racial Gaps in Test Scores: The Case of the Minnesota Basic Standards Tests, The

Journal of Negro Education, The, Winter 2004 by Myers, Samuel L Jr, Kim, Hyeoneui, Mandala, Cheryl

Surprisingly, the effect of school poverty in the first column sample is to increase test scores. The impact is small, however. A ten-percentage point increase in school poverty would be required to raise Black students' math test scores by 1.4 percentage points. Still, the effect is positive and not negative, as adherents to the school poverty thesis would expect. The effect on White students is also positive in 1996, although it is smaller than the effect on Black students.

The next two columns report the estimated coefficients on the school poverty variable when school poverty is interacted with all individual level variables in the model. There are two effects of school poverty in equations with interaction terms. There is the direct effect-or the impact of school poverty holding constant any indirect impacts. And there is the gross effect, which is the sum of the direct effects and the indirect effects.6 Again, we observe a positive impact of school poverty in test scores in 1996, although these effects are statistically insignificant for Black students.

Before turning to the HLM results in Table 1, observe that the anticipated negative effects of school poverty on test scores does emerge in later years. Columns one and two, reflecting the OLS estimations without any interaction terms, reveal inverse relationships between White students' math and reading test scores in 1998 and 1999 when no controls are made for individual poverty. Inverse relationships are found for Black students in 1999 when no controls are made for individual poverty.

The substantive explanation for the change in the results between 1996 and 1998-1999 is that not all schools were required to submit MBST scores in 1996. One large school district, St. Paul Public Schools, administered a different examination that year. St. Paul is the second largest city in Minnesota and although its concentration of African American children is lower than that of Minneapolis, the largest Minnesota school district, there are large pockets of poverty. The positive relationship between test scores and school poverty of White students in 1996 accordingly reflects the dominant influence of Minneapolis Public Schools on the overall state results. Including St. Paul and other school districts that failed to report data in 1996 reverses the odd conclusion that school poverty increases test scores.

The first six columns of Table 1 also reveal that the measured effects of school poverty vary with respect to controls for individual poverty and controls for interactions between school poverty and individual poverty. For example, the positive impact of school poverty on mathematics scores of White students in 1996 becomes negative once one controls for interaction effects. Whereas the school poverty effect for Whites in 1996 is .0165 without taking account of interactions between school poverty and individual-level variables, it is -.0484 with controls for interaction effects. And, even in later years when negative coefficients are estimated for the school poverty variable when no account is taken for interaction effects, when interaction effects are accounted for, school poverty registers a smaller effect. For example, the school poverty coefficient on the reading test in 1999 with no controls for interaction terms and no controls for individual poverty among black students is -.0697. Accounting for interactions, the gross effect is estimated to be -.0482. When individual poverty is taken into account, the impact of school poverty reverts to a positive effect in the OLS regressions.