Relationship between School Performance and Poverty, 2018

Atlanta Public Schools has the most diverse portfolio of schools in the state. We have schools in some of the wealthiest neighborhoods in Georgia and schools in some of the poorest. The graph below shows that these schools have very different math achievement levels:

Use the filters on the left to select different grades, subjects, or performance metrics.

The challenge index measures the percentage of students who are either directly certified1 for free lunch or English language learners. Only about 3% of APS students are English Language Learners, so for most schools the challenge index can be thought of approximately as a measure of poverty.

Comparing schools with similar challenge indices is helpful for understanding school performance. A school that is above the trend line has higher test scores than other schools with a similar level of poverty. But of course this view isn’t conclusive- two schools with similar challenge indices might differ in other ways that is not captured by the challenge index.

Click on an individual school to see that school’s data over time. In the graph below, we’ve selected Drew Elementary and Hope-Hill. We can see that Drew Elementary has performed consistently above the trend line and that student population has become more affluent over the past four years. The time-trend for Hope-Hill shows the school has high performance relative to peers in 2018 after performing at a similar level to their peers the previous three years.

To see the distribution of school performance in 2017, see last year’s post.

Notice the strong relationship between the challenge index and average test scores. The challenge index explains 87% of the variation in elementary school average math scores, and 93% in school average English Language Arts (ELA) scores. The difference in variance explained by poverty between math and ELA is consistent with academic studies that show successful interventions tend to have a larger impact in math than in ELA2. In other words, if ELA is better predicted by family socio-economic status, then there is less room for other factors to predict ELA performance, such as school effectiveness or interventions. This makes some intuitive sense- students are more likely to learn vocabulary at home than algebra.

Although the challenge index explains most of the school-level variation in test scores, it only explains 20-25% of student-level variation. This means there are high-performing students at low-performing schools and vice-versa.

The y-axis for these graphs uses average Milestones NCE score, which is similar to average test score, but converted to a common scale to average across grades and subjects. We display average test scores on this graph because an average is a more accurate summary than reporting the percentage of students above a cut score. However, percentages are also useful for understanding performance, so we allow the user to change the y-axis with the performance metric filter on the left.

Notice how the distribution changes for different metrics. The graph below shows the percentage of students scoring Distinguished on the Milestones. (See here for an explanation of Milestones achievement levels.) This metric emphasizes differences in achievement between our low poverty schools on the left, while our high-poverty schools show similar performance.

Finally, we also have this data for End of Course tests, which shows comparisons for high schools and eighth grade advanced math students.

To view other measures of school performance, see our school profiles or our school comparison page. Note that data releases occur as GADOE produces the data. So although we have 2018 Milestones results, we won’t have 2018 student growth percentiles or CCRPI until this fall. If you have questions about this data, feel free to contact us using the form on the About page. If you would like to know about similar posts, please subscribe to the blog.


  1. A student is directly certified for free lunch if their household receives state anti-poverty aid (SNAP or TANF) or if the student is homeless, an unaccompanied youth, or a migrant. This measure is used instead of the more general “free and reduced lunch” because many APS schools use the CEP option for whole-school free and reduced lunch. Poverty rates are lower when using direct certified instead of free lunch, but the rank-ordering of schools is mostly the same.
  2. For example, this Roland Fryer study.