Yesterday we looked at some National Assessment of Educational Progress (NAEP) Grade 4 mathematics data for students eligible for the federal school lunch program (a poverty measure) for charter schools versus non-charter schools. The data was broken down by race and was obtained from the NAEP Data Explorer.

Now, let’s see how eighth grade students in poverty in charter schools and non-charters compare. Data for this grade is only available from 2005 on. Here is the white data.

Poor whites in charter schools scored notably below their counterparts in the 2005 NAEP by 8 NAEP Scale Score points. By 2011, things had reversed and poor whites in charter schools scored 4 points above their counterparts in traditional classrooms across the country.

It should be noted that there has been some year to year up and down in the scale score differences. This is probably due to the large sampling error in the NAEP charter school samples. However, a linear regression analysis of the trend over time confirms a positive trend that definitely favors charter school students.

This next table shows the blacks data.

This trend is even more pronounced, with poor blacks in charter schools moving from 8 points behind their traditional public school peers to scoring 6 points ahead of them. Unlike the situation with white students, the improvement was consistently positive.

Finally, here is the Hispanic table.

In this case, the notable lead Hispanics have held in charter schools has been cut by Hispanics in traditional schools, but even today poor Hispanics in charter schools outscore their traditional school peers by 9 points.

I want to reiterate some caveats I mentioned in Part 1 of this series. First, the National Center for Education Statistics has not seen fit to provide any information about exclusion of students with learning disabilities and English language learners broken down by charter and non-charter categories for NAEP assessments. If exclusion rates differ, that could bias the results shown in these tables.

Also, it is reported that a number of charter schools simply do not offer the federal school lunches, so even students in poverty in those schools will be carried as non-poverty students. That also might impact the data in the tables.

Still, even with these caveats, it seems likely that as of 2011 charter schools across the nation are really starting to perform notably better for students in poverty than do traditional public schools.

And, so far, Kentucky’s kids are missing out completely on this trend that helps boost overall public school performance in other states that now have charter schools.

Perhaps that helps explain why when we look at all white students in every state who are in the school lunch program, Kentucky only statistically significantly outscored poor whites in just two other states (Alabama and West Virginia) in the 2011 NAEP Grade 8 Mathematics Assessment – just 2 states!

Maybe if Kentucky had charter schools, the map above (also assembled with the NAEP Data Explorer) would look different.

By the way, I don’t plan to look at NAEP reading due to higher differences in exclusion rates across the states. In fact, Kentucky led the nation in both fourth and eighth grade NAEP reading in 2011 for its very high rate of exclusion of students with learning disabilities. In consequence, even the Kentucky Department of Education is now reporting that our NAEP reading scores are not comparable to those in other states.

Information about charter school performance from the National Assessment of Educational Progress (NAEP) is sketchy and inconclusive. An important part of the problem is that the NAEP student sample sizes for charter schools have been rather small, which creates a lot of measurement error in NAEP charter school scores. The large measurement error in turn makes it impossible to detect small to modest trends.

Still, given that the trends shown below may not yet have risen to the point where they can be declared statistically significant, there does seem to be a trend in the nationwide NAEP Grade 4 mathematics results for students of different races who are eligible for the federal free and reduced cost lunch program (a proxy for poverty). The trend favors charter schools.

Here is how the NAEP results for white students eligible for the school lunch program look according to the NAEP Data Explorer.

Notice that in 2003 (earliest charter data collected in NAEP), charter school poor whites scored a point behind poor whites in traditional schools across the nation. As of the latest data for 2011, that has changed, and poor whites in charter schools now outscore their public school counterparts by 4 points.

Also note that between 2003 and 2011 poor whites in charter schools improved their scores for fourth grade math by 11 points, while their public school counterparts only gained 6 points.

Now look at the data for blacks.

The score differences for blacks have shifted around a bit, which may be due to the inadequate sampling sizes in charter schools, but over time the trend does seem to favor charter school students in the lunch program. I ran a regression of the change in scores over time, and there is a positive slope to the best fit line. That supports a trend favoring blacks in charter schools.

Lunch eligible charter school blacks improved their score by 12 points while blacks in traditional public schools only improved by 8 points.

Finally, here are the results for Hispanics in the lunch program.

The trend in the difference scores is the most dramatic of all, with lunch eligible Hispanics in charter schools moving from two points behind to six points ahead of their traditional school counterparts.

Lunch eligible charter Hispanics increased their scores from 216 to 232 between 2003 and 2011, a 16-point rise. Their public school counterparts only improved by 8 points.

I need to point out some caveats to this data. There is no information on exclusion rates for learning disabled and English language learners broken out by charter and non-charter schools. Different levels of exclusion could bias the data.

Also, it is reported that a lot of poor students in charter schools don’t join the federal lunch program simply because the charter schools they attend do not offer it. Those kids are still poor, but they don’t show up that way in the data. That could bias the charter school information, as well.

These caveats further highlight the limits of the NAEP data, of course. Hopefully, NAEP’s collection of charter school performance will improve in the future.