Right after Christmas, the Bluegrass Institute released my short paper on Reading proficiency rates rising in some Appalachian schools. This paper showed some schools in the Eastern part of Kentucky – as of 2019, before COVID messed everything up – were producing astonishingly high third grade reading proficiency rates as high as 90% on the state’s KPREP assessments. These schools succeeded despite serving really high percentages of students that qualified for the free and reduced cost school lunch program, a statistic often used as a proxy for student poverty rates.

The Appalachian schools paper employed a really neat analysis tool created by the Education Consumers Foundation that shows each school’s performance on a scatter plot of reading proficiency rates versus school lunch eligibility rates.

Analysis in the report highlights schools like Goose Rock Elementary School in Clay County, where third grade reading proficiency soared from just 23% on the 2012 KPREP to an astonishing 89.7% rate in 2019. Even more astonishing, Goose Rock did this despite having a very high, 85% school lunch rate!

The Appalachian schools paper covers more. Teachers in Clay County’s public schools got special help from an Elgin Children’s Foundation program to make these remarkable results happen. Elgin came in and showed the teachers how to teach reading much more effectively. The results speak for themselves.

One important message from the Clay County experience is that poverty isn’t an excuse. Regardless of family income levels, if kids get the right instruction, they can perform.

So, that’s the exciting case for reading instruction. Proper teaching rather than poverty is what matters the most.

But, what about math?

The Education Consumers Foundation doesn’t have a math analysis tool. So, I created my own scatter plot using raw data in the Kentucky Department of Education’s School Report Card system. Data is available for all of the state’s schools in these spreadsheets, including files for KPREP results by grade and subject, the counts of students eligible for free and reduced cost school lunches and each school’s total membership (you probably call that enrollment).

I used the school lunch counts and membership data to compute an eligibility rate for school lunches for each school.

Next, I combined the calculated school lunch rates and the KPREP Grade 3 Math proficiency rate data for all students tested in each school to create this scatter plot. Math proficiency rates are plotted against the vertical axis and school lunch eligibility rates are plotted along the horizontal axis. Each dot represents a Kentucky elementary school that had all the data needed to create the plot. 

I also sketched in a linear regression line, colored red, which shows on average how schools performed in 2019 in relationship to their school lunch rates. If a school’s dot actually plots above the line, it performed better than average. A plot below that red line shows a school that’s under-performing even when poverty is considered.

I identified some of the schools connected to some of the dots to help point out some things.

Something really remarkable, a number of high poverty schools plot out in the upper right part of the graph with some really impressive math proficiency rates. Perhaps the standout of standouts is the Caywood Elementary School in the Harlan County School District. Despite a super-high poverty rate of 90.9%, Caywood posted an even higher number for its KPREP Grade 3 Math proficiency rate in 2019, an astonishingly high 93.3%!

Not too far behind Caywood is Goose Rock Elementary. It seems good performance in reading may spread over into better performance in other academic areas because Goose Rock posted an 82.8% proficiency rate in KPREP Grade 3 Math despite that 85% school lunch rate I mentioned earlier. Wow!

Now, drop down to the lower right part of the graph. Here we find very high poverty schools that performed very poorly on KPREP Grade 3 Math in 2019, well below the red average performance line. Jefferson County Public Schools’ (JCPS) Roosevelt-Perry school is one of the schools found in this corner of the graph.

I would not be surprised if teachers in Roosevelt-Perry were quick to offer an excuse that their school’s low performance is due to student poverty. However, schools like Caywood and Goose Rock discredit that excuse. Something else is going on here. Could it be teacher experience? Maybe it’s teacher preparation? We need answers, not excuses, if this school is going to improve. Why are Goose Rock and Caywood succeeding despite high student poverty while Roosevelt-Perry isn’t?

I have a concerning question: have Goose Rock and Caywood ever gotten much attention for their high performances? I don’t recall hearing anything about them before I started looking at this scores-versus-poverty data.

Focusing back on the scatter plot, not much separated from Roosevelt-Perry’s plot is the plot for the Trunnell Elementary School in Jefferson County. This school performed only a little bit better than Roosevelt-Perry in 2019 KPREP Grade 3 Math with a 15.7% proficiency rate and a poverty rate of 84.9%.

Consider another JCPS school, the Kenwood Elementary. It’s poverty rate of 85.1% is essentially identical to Trunnell’s, but Kenwood’s third grade math proficiency rate in 2019 was 59.8%. That’s not great, but it’s well above the red expected performance line and nearly four times higher than Trunnell Elementary’s third grade math proficiency rate.

There’s more. Trunnell and Kenwood are located only 2.1 miles apart by road, about a 4-minute drive. That’s a much shorter drive than many school bus rides in Jefferson County.

Why are results so different for two public schools from the same school district that are located close together and have virtually identical student poverty levels?

The experience level of classroom teachers in these two schools average 10.7 years in Trunnell while Kenwood’s is somewhat longer at 12.4 years. That fairly small experience difference doesn’t seem a likely explanation.

Looking at the web sites for Trunnell and Kenwood does reveal they use different approaches to teaching math. Kenwood doesn’t identify a specific program, but Trunnell says it’s using Bridges Math. According to the What Works Clearinghouse (WWC):

“The lack of studies meeting WWC evidence standards means that, at this time, the WWC is unable to draw any conclusions based on research about the effectiveness or ineffectiveness of Bridges in Mathematics.”

Maybe there are curriculum issues in Trunnell. At the very least, it looks like there aren’t any rigorous studies supporting the math program Trunnell is using.

Now, I have another question: if you were a parent in Jefferson County, other things being equal, where would you want to send your child, Trunnell or Kenwood?

Moving over to the left bottom side of the graph, we find some of the upper-scale schools in Kentucky that don’t match their higher student income situations with commensurate math proficiencies.

I was surprised to find the Model Lab Elementary school hanging out well below the red average performance versus poverty line. Model Lab is attached to the Eastern Kentucky University and in theory should benefit from that close association with the college and its education department.

Model Lab Elementary’s mundane, 39.5% math proficiency rate is highly disappointing under any condition, but this is especially so considering that Model Lab’s lunch eligibility rate of 16.1% is much, much lower than Kentucky’s statewide average. Caywood Elementary has over 5-1/2 times that poverty rate. However, Model Lab Elementary produced less than half of the third-grade math proficiency that Caywood generated in 2019.

Maybe Eastern Kentucky University should hire some Caywood teachers to find better ways to train student teachers how to teach math effectively.

In the upper-left part of the graph we find some really upper-scale schools like the Johnson Elementary School in the Fort Thomas Independent district and the Anchorage School from the very upper-scale Anchorage Independent School District. Both schools have only single-digit school lunch eligibility rates. Regardless, neither of these wealthy area schools posted a math proficiency rate as high as some of the high-poverty schools found in the upper right side of the chart. Caywood and Shopville Elementary School are just two examples.

Looking at the red regression line, Johnson’s 72.7 math proficiency rate is about what would be expected based on its very low poverty rate. Anchorage, in contrast, does score a bit above it’s expected rate.

On the other hand, Anchorage spent a total of $22,274 per pupil in the 2018-19 school year according to Kentucky School Report Card data. Caywood only spent $14,872.

At least in 2019, Anchorage didn’t get really great bang for the buck, but Caywood certainly did.

Some final thoughts:

• Poverty doesn’t excuse low performance in schools. Given properly prepared teachers, much better results can happen even for very poor student groups.

• In education, just spending more doesn’t necessarily get you more. Spending has to be effective and efficient.

• Schools that might seem to perform well actually might not be very efficient.

• Teachers need access to tools that really work. There’s no guarantee the colleges they attended and the follow-on professional development offered in their schools provide that.

• There’s hope. Better results can happen. They already are right here in a few Kentucky schools.

• There are good examples here why parents would want school choice options. Why keep a child in a very low performing school if another, much better one is close by? In fact, our Excel spreadsheet might just help parents in a quest to find a better school.

Technical Notes: Getting to the data

First, go to the “Research Data” web page.

From the home page of the Kentucky School Report Cards, click on the “Explore Statewide Education Data” link just below the box used to search for a specific school or district. In the new web page that opens, click on the “Research Data” link to open a page with a whole bunch of links to Excel spreadsheets that contain all sorts of information.

How to access school lunch student counts.

From the Research Data page, select the “Overview” tab. In the new section that opens, look for the “Economically Disadvantaged” link in the “Student Group” section. The Kentucky Department of Education uses the school lunch student counts as its economically disadvantaged figures.

How to access student membership

Again, in the “Overview” tab, select the “Student Membership – Race/Gender” link in the “End-of-Year Enrollment” section.

How to access KPREP Scores.

In the “Assessments/Accountability” tab, in the Assessments Proficiency section, select the link that says “by Grade.” Note that this is a very large spreadsheet and takes some time to download. It is also challenging to navigate due to its enormous size (over half a million lines of information!).

How to access the school spending data.

In the Financial Transparency section click on the “School-Level Spending Per Student” link.

 How to access the teacher experience level in schools.

In the Kentucky School Report Cards data sets, “Overview” tab, click on the “School Experience” link in the “Faculty, Staff and Community” section.

By the way, except for the school funding data and the teachers’ years of experience data, I downloaded my copies of the Excel files in October 2019. The current versions might be missing some scores found in those 2019 downloads. Long after the files were initially released in 2019, the US Department of Education required the Kentucky Department of Education to suppress some KPREP score information that already had been publicly available for the better part of two years. I guess the US Ed folks don’t realize that anything once released in the web lives on forever. In this case, the horse was long out of the barn before the US Ed folks even started searching for their bridle.