This report is part of a series of analyses of publicly available North Carolina Department of Public Instruction student achievement data, known as the disaggregated data files. These FERPA-compliant files contain data on all the state’s public traditional and charter schools beginning in school year 2013-14. My intent in this series is to provide the public with discussions of the usefulness of the NCDPI data, and with descriptive statistics that I feel may be of interest. I will not perform modeling or hypothesis testing.

Quantitative Aspects of AIG Classification and Race in Chapel Hill-Carrboro Public Schools for Grades 4 and 5

In Grades 4 and 5, both Black and Hispanic students are more frequently in schools with lower percentages of AIG students. This is not a strong effect but it is more pronounced for Hispanics. Contrastingly, both White and Asian students are more frequently in schools with higher percentages of AIG students. This implies an opportunity equity deficit, although this data, taken alone, is not a strong determinant. As a reminder, the NCDPI publicly available data does not provide the number of AIG students by race (subgroup), so my results are inferences. That data does exist and should be analyzed to gain a better picture of AIG-related equity. It is evident that there are schools with markedly high and low percentages of AIG students over at least the past several years. The causes and consequences of this are surely of public interest but are not addressed in the NCDPI publicly available data.

The North Carolina legislature mandated supportive treatment of academically gifted students in its Article 9B https://ncleg.net/EnactedLegislation/Statutes/pdf/ByArticle/Chapter_115c/Article_9B.pdf The NCDPI disaggrgated data files do not break down numbers of AIG students by race. NCDPI describes its framework for implementing the legisture’s requirements at https://www.dpi.nc.gov/students-families/enhanced-opportunities/advanced-learning-and-gifted-education/academically-or-intellectually-gifted NCDPI provides summary data for LEAs, but not for schools, in a remarkably awkward format at https://files.nc.gov/dpi/documents/advancedlearning/aig/student-data/childcount/2018-19_lea_summary_.pdf

It has been reported in multiple studies that minority students are identified as AIG at lower rates than are White students. This report reviews data for individual Chapel Hill-Carrboro schools to compare AIG classification characteristics. I will look at grades 4 and 5. Grade 4 is the earliest that students are identified as AIG.

AIG Percent by Grade and School

Remember that two points do not constitute a trend.

Figures 1 show percentage of AIG students for grades 4 and 5 by school across the years from 2013-14 to 2018-19. The plot for 681LEA is the average for each grade.




Deviation from Average AIG Percent by Grade and School

Figures 2 presents the same data as in Figures 1, but in a more informative fashion. Here, the plots show the difference between the percentage AIG and the average for that grade and year. It is evident that the Margaret Seawall school was consistently above all the other schools in percentage AIG for both grades.




Percentage Subgroup Composition of Schools by Grade

Figure 3 shows the percentage composition of the schools aggregated for grades 3 through 5. The bars should of course all reach 100%, but to reduce clutter I have not shown the remaining portion, which is primarily Multiracial students. That there is a significant number of Multiracial students suggests that further exploration of this category is required within the customary single-race construction. A csv file containing percentages of students by race, and percentage AIG, by grade by school can be found here.

Figures 4 display the same data as in Figure 3, but only for one year, 2018-19. They explicitly compare grades 4 and 5 racial (subgroup) composition of the schools to the CH-C elementary school average for the grade (681LEA). For both grades, Northside (618300) had the lowest percentage AIG students and in both years the representation of races was reasonably comparable with the grade’s average (618LEA). Seawell (618310), with the highest percentage AIG students, in grade 4 had an over-representation of Asian students, typical portions of Black and Hispanic students, and a reduced proportion of White students. In grade 5, Seawell had a markedly elevated representation of White students, although this was comparable to the grade 5 average proportion of White students, this elevated proportion being absorbed by a decreased proportion of Asian students. The most notable differences between actual and average proportions are those for Asian students and Hispanic students.

Is There Any Indication of a Relationship Between Percentage AIG and Subgroup?

The publicly available NCDPI data provides percentage AIG by school, but that number is not broken down by race (subgroup). That is in the NCDPI data, but it is not accessible except under FERPA-compliant conditions. This prevents a direct assesment of the extent to which minority students participate in AIG benefits. What I present in Figures 5 and 6 are a recasting of Figures 2 and 3. Figures 5 show a comparison of the 2018-19 overall, grade-specific AIG percentage, by school, with the percentage Asian, Black, Hispanic, and White students, and also for the combined White and Asian, and the combined Black and Hispanic, student counts. In each of the plots for a grade, the percentage AIG on the vertical axis is the same (what you see is what you get), while the horizontal axis shows the percentage of students by race (subgroup).



Figure 6 displays percentage AIG for 2018-19 grades 4 and 5 for all CH-C schools as bars, with the intent of making it somewhat easier to see how the two grades compare. The error bars indicate what percentage AIG would be if there was one additional, or one fewer, student classified as AIG, with the total number of students unchanged. Any differences in percentage AIG that are within the error bars should be considered insignificant. Figure 6 confirms that Seawall has claim to being highest in percentage AIG for grades 4 and 5. This also shows that Northside is distinctively low in grade 4, but that grade 5 has three low percentage AIG schools, Northside, Carrboro, and McDougle.

The Figure 6 plots for grades 4 and 5 show that both Black and Hispanic students are more frequently in schools with lower percentages of AIG students. This is not a strong effect but it is more pronouned for the Hispanic students. Contrastingly, the grouped White and Asian students are more frequently in schools with higher percentages of AIG students. This implies an opportunity equity deficit, although this data, taken alone, is not a strong determinant. As a reminder, this NCDPI data does not provide the number of AIG students by race (subgroup), so my results are inferences. That data does exist and should be analyzed to gain a better picture of AIG-related equity.

It is evident that there are schools with markedly high and low percentages of AIG students over at least the past several years. The causes and consequences of this are surely of public interest but are not addressed in the NCDPI publicly available data.

Detailed data is in Appendix A.

Appendices

Appendix A. Percentage of AIG students by grade by school by year.

A csv file containing percentages of students by race, and percentage AIG, by grade by school can be found here.

The following table shows percentage of AIG students by grade by school by year.

grade school_code 2013-14 2014-15 2015-16 2016-17 2017-18 2018-19
04 681300 25.9 26.7 21.3 19.0 21.3 19.0
04 681304 34.6 17.4 16.8 20.5 14.5 35.6
04 681310 44.9 34.3 50.0 38.7 51.0 57.5
04 681311 22.4 15.8 22.5 34.8 32.5 34.2
04 681312 31.1 14.9 25.3 32.0 35.3 35.8
04 681316 23.5 8.0 20.0 19.7 20.4 31.6
04 681320 47.9 33.3 30.4 37.3 39.3 40.6
04 681326 28.2 22.0 31.5 17.1 19.1 34.6
04 681330 19.1 14.7 16.9 21.1 28.4 36.8
04 681332 28.8 17.5 29.8 23.5 23.4 37.9
04 681334 26.6 25.0 23.3 21.4 27.9 29.2
04 681LEA 30.6 21.4 26.6 26.1 28.9 36.2
05 681300 29.7 28.7 29.2 27.3 25.9 19.5
05 681304 31.7 37.3 22.9 27.7 28.6 21.5
05 681310 45.0 45.0 46.7 51.2 53.0 58.2
05 681311 31.9 21.1 25.4 28.8 34.8 35.0
05 681312 30.2 30.7 34.9 29.6 32.7 47.6
05 681316 23.9 20.4 19.6 25.7 28.4 26.0
05 681320 44.1 41.4 33.3 36.5 36.4 41.3
05 681326 36.2 26.8 23.7 35.7 15.7 20.2
05 681330 25.9 22.6 28.7 25.8 34.6 34.0
05 681332 34.8 23.3 27.8 39.8 31.4 29.5
05 681334 38.5 23.3 37.7 25.7 23.4 31.2
05 681LEA 34.7 29.8 30.7 33.0 32.5 33.8

Appendix B. AIG Subgroup Correlations Grade 4 2018-19

I present here grade 4 2018-19 correlation tests for percentage AIG versus percentage subgroup for the eleven CH-C elementary schools. Spearman and Pearson tests are run and plots are presented (instances with zero students for a subgroup are removed). The correlations are weak at best, and the plots show why, being characterized by the isolated extreme percentage AIG instances of Seawall and Northside.

Here is the data for 2018-19 grade 4.

school_code AIGpct Asian Black Hispanic White Black_and_Hispanic White_and_Asian
681300 19.0 10.0 22.0 22.0 40.0 44.0 50.0
681304 35.6 0.0 5.7 34.5 52.9 40.2 52.9
681310 57.5 30.1 14.2 11.5 38.1 25.7 68.2
681311 34.2 9.2 6.6 18.4 55.3 25.0 64.5
681312 35.8 11.6 5.3 18.9 55.8 24.2 67.4
681316 31.6 0.0 0.0 56.6 35.5 56.6 35.5
681320 40.6 36.2 11.6 0.0 43.5 11.6 79.7
681326 34.6 11.5 15.4 15.4 55.8 30.8 67.3
681330 36.8 7.9 22.4 15.8 47.4 38.2 55.3
681332 37.9 11.6 8.4 12.6 57.9 21.0 69.5
681334 29.2 27.7 12.3 0.0 49.2 12.3 76.9

In these plots, Exclude and Include refer to whether Seawall and Northside are part of the data being analyzed. The plots omit schools with no students for a subgroup. The blue lines are simple linear regression fits, with 95% confidence regions in grey. The more horizontal a regression line, the less relationship there is between percentage AIG and percentage subgroup.

The following table shows Spearman and Pearson correlation tests computed for the data. This table presents data for both the “Exclude” (omit Seawall and Northside) and “Include” plots. The Spearman data includes rho and the p-value. The Pearson data includes the correlation coefficient, cor, and its 95% confidence interval. The tests agree that there is a negative correlation for Hispanic students, although that appears to be primarily a consequence of the large proportion of Hispanic students in the Graham Dual Language Magnet. There is at least a weak positive correlation for the combined White and Asian students. There is no indication that there is any correlation for the Black students.

Exclude
Include
Spearman
Pearson
Spearman
Pearson
subgroup rho.e Sp.p.e cor.e from.e to.e rho.i Sp.p.i cor.i from.i to.i
Asian 0.2162 0.6414 0.0838 -0.7143 0.7872 0.5021 0.1684 0.4785 -0.2721 0.8671
Black 0.0000 1.0000 0.0223 -0.6933 0.7157 -0.0909 0.8114 -0.2169 -0.7448 0.4780
Hispanic -0.6071 0.1667 -0.7767 -0.9652 -0.0570 -0.7333 0.0311 -0.3425 -0.8201 0.4163
BlackHispanic -0.3333 0.3853 -0.2475 -0.7829 0.4986 -0.4091 0.2139 -0.3304 -0.7764 0.3361
White 0.1674 0.6669 0.1797 -0.5500 0.7539 0.1002 0.7694 -0.0803 -0.6489 0.5458
WhiteAsian 0.3500 0.3586 0.2906 -0.4629 0.8003 0.5182 0.1069 0.3664 -0.2992 0.7922

Since the Graham Dual Language Magnet has only Hispanic and White students, it may bias the correlation calculations. Here are those calculations with Graham omitted:

Exclude
Include
Spearman
Pearson
Spearman
Pearson
subgroup rho.e Sp.p.e cor.e from.e to.e rho.i Sp.p.i cor.i from.i to.i
Black 0.0000 1.0000 0.0223 -0.6933 0.7157 -0.0909 0.8114 -0.2169 -0.7448 0.4780
Hispanic -0.3714 0.4972 -0.2701 -0.8872 0.6934 -0.6667 0.0831 -0.4265 -0.8698 0.3976
BlackHispanic -0.2143 0.6191 0.0612 -0.6725 0.7342 -0.3333 0.3488 -0.3105 -0.7864 0.3966

It appears that removing Graham weakens the Hispanic correlation. In either case, there is indication that grade 4 Hispanic students experience limited participation in the AIG program.