How Mississippi Escaped the Bottom Tier of U.S. Education

U.S. reading and math scores have been declining for over a decade. Why has Mississippi risen in the ranks, and does this matter?

Hey, everyone! My name is Jordan, and I have a Ph.D. in Economics from the University of Pennsylvania. I want to ask important questions and answer them meaningfully using the hard work economists have put into their research.

My goal is to have 500 subscribers by the end of the year. I enjoy teaching others about economics, so I put a lot of effort into condensing research and making graphs to make these posts fun (hopefully) and understandable. If you learn anything interesting from this, please like, share, and/or subscribe! Thank you to all my current subscribers for reading all my posts.

A few months ago, Mississippi’s education system drew a lot of attention in the media. For years, the state ranked at or near the bottom in U.S. education. In 2013, Mississippi posted the lowest fourth grade math scores and the second-lowest fourth grade reading scores on the NAEP (National Assessment of Educational Progress), a national test that randomly samples schools and then students in those schools. Eighth grade reading and math scores were similarly poor in 2013.

But then, something abruptly changed around 2013; Mississippi’s scores began to climb rapidly, far outpacing the national average and even other Southern states like Alabama and Louisiana. Of course, this is why Mississippi’s education situation became popular in the media and even became known as the “Mississippi Miracle.” The graph below highlights the sharp rise in math scores, especially in 2019. The reading scores look fairly similar to this graph as well. How did such a dramatic improvement occur within a decade, even as most states saw their scores decline? But more importantly: what do these scores even mean?

Most of the discussion has been around the increase in reading scores, which can be seen below. Many claim that changes in the way reading is being taught can account for this increase, like a focus on phonics. Specifically, Mississippi enacted the Literacy-Based Promotion Act in 2013, and there is some causal evidence to show this policy could have positively impacted reading scores.

Source

Before going into possible explanations, it’s important to understand what this trend actually reflects. Usually, when I see any article or post referring to a “score” or even a “ranking,” I am very skeptical, and you should be as well. Headlines often make claims without solid evidence. For the NAEP results in particular, I was initially skeptical for the following reasons:

1. What does the score or ranking even mean?

2. Who were the students chosen to take the test in each state?

3. Did the types of questions on the test change over time, or have they remained fairly the same?

Thankfully, the NAEP seems to have a thorough system designed to ensure its results are meaningful and not driven by factors like sampling only large schools or drastically changing test questions over time. There are still some more complicated issues with the way NAEP calculates scores that I won’t go into here, but to answer the questions I initially had:

1. Different topics in the framework contribute a certain percentage to an overall score on a scale of 0-500. For instance, “number properties and operations” make up 40% of the exam questions, while “geometry” makes up 15%.

2. The NCES divides schools by rural, suburban, and urban areas and decides to choose schools from each of these groups. They also do this with percentages for each race and ethnicity. Then, they make sure they are not just choosing the biggest schools within each of these groups; there is a chance to pull smaller schools, but that chance is still smaller than choosing a bigger school. Once the schools are chosen, the NCES then randomly chooses students in the school to take the test.

3. The framework for math and reading hasn’t really changed much over time — at least for 4th graders and 8th graders. In 2017, the NAEP tested a version where students had to take the test on the computer, but they made sure the results didn’t differ too much from those taking the test on paper.

So, this test seems fairly valid and committed to ensuring it is consistent over time, even if there are critiques to the way it calculates final scores for different students. However, whether this exam is actually representative of what a student should actually know and understand is to be debated; should “numbers and operations” actually makeup 40% of the exam?

Which factors are more likely to account for math score differences?

If you ever read my article on why people in the U.S. South are dying earlier than the rest of the country, then you saw what is called a variance decomposition. Basically, all this means is that we look at different topics or variables and determine which of these topics or variables can account for the differences in the math scores we see in the data. Here, I apply the same idea to NAEP fourth grade math scores using a wide set of survey questions provided by the NAEP. I only look at fourth grade math scores this time — if I went even further into all the details for all the scores, I would end up with an actual paper. Plus, a lot of focus has already been put into the reading scores.

Here, I’m trying to figure out which factors help explain why some states had higher fourth grade math scores than others in 2017/2019. I focus on these years because there are more background questions available, COVID wasn’t affecting schools yet, and they give us a clearer view of Mississippi’s steady improvement. Ideally, I would compare how changes in these factors relate to score changes from 2011 to 2019, but the same survey questions usually don’t exist for all of those years. So instead, I look at what seems to matter most for student performance in 2017/2019.

In the graph below, I start by looking at broad categories to see which ones explain the largest differences in the math scores across states. This does not mean any of these factors directly “cause” scores to rise or fall. Each one is influenced by many other conditions or things happening within a state. Economists spend months or even years studying whether a specific policy truly causes changes in test scores. What this analysis can tell us is which areas appear to matter more than others when it comes to understanding these differences.

For the nerds: I used PCA to create indices for each topic (with MICE used in imputation) then simply ran OLS on all the indices. This accounts for 65% of variance, with a similar adjusted R-squared.

When we combine the two Title I funding categories, the top factors are tied between how this Title I funding is used and how math is taught — the latter includes questions to teachers like “How much emphasis do you put on numbers and operations?” and “Is the math program structured by district assessment results?” The third most important topic is how teachers view the school and students, with questions like “Are you satisfied with being a teacher?” and “Is your teaching limited by disruptive students?”

What’s surprising is how little technology access seems to matter for test scores. This topic included questions like “Do you use the Internet for homework at home?” and “What is the availability of laptop computers for student use?” The results don’t show whether these things help or hurt achievement — only that they don’t explain much of the difference in math scores across states. Teacher training, such as teachers’ education levels, also appears to matter far less than many of the other factors.

Together, these topics explain about 65% of the differences in math scores across states, which is fairly strong. There are still many important factors we aren’t capturing, like state economic conditions, students’ home environments, specific outcomes from policies like 3rd grade retention, or even how much students care about the test. In a formal research study, economists would try to include data on these additional areas whenever possible. But the results here are still promising.

Next, I look at the individual survey questions to see which specific items best explain the differences in scores across states. The strongest factor by far is the percentage of schools using federal Title I funding to improve the schools themselves; Title I funding is federal money sent to schools with many low-income students. The graph doesn’t show whether more funding raises or lowers scores, but in the model I used, schools that spent a larger share of this money on general school purposes rather than directly on students tended to have lower scores.

The second most important factor is the emphasis teachers place on geometry, which includes topics like angles and calculating areas of shapes. As mentioned earlier, geometry makes up about 15% of the NAEP math test, so heavier focus on this material seems to matter. This is not to say learning geometry doesn’t help with learning other math topics.

For the nerds again: To cut down on variables, I used elastic net. Still ended up with about 65% in explained variation.

Next, I take a closer look at the top two factors: how they changed in Mississippi and what the economics research says about their links to student achievement. These are simply factors that shifted over time and may or may not be tied to specific state policies, though I also discuss the education reforms that could have influenced them. One limitation of my analysis is that my model slightly underpredicts Mississippi’s 2019 score, even though it comes very close to the score in 2017. The main reason the model predicts 2017 so well is Mississippi’s strong emphasis on geometry instruction.

Title I Funding

Title I is the largest federal K–12 funding program, providing extra funding to schools with many low-income students. It’s meant to supplement, not replace, state and local dollars. Schools can use the money in two main ways: 1. for general schoolwide needs or 2. for students who are at risk of falling behind. Because Mississippi has a high share of low-income families, it receives a significant amount of Title I funding relative to its size.

The graph below shows the percentage of Mississippi schools using Title I funds for schoolwide programs and for students. From 2011 to 2019, when the increase in math scores happened, these percentages barely changed. The only major shift came during COVID, when schools received extra emergency funding and adjusted how they used Title I dollars. Overall, it’s unlikely that changes in Title I funding explain Mississippi’s jump in math scores in particular.

This analysis doesn’t tell us that increased funding toward Title I doesn’t increase scores or help students; we need a deeper causal analysis to understand that. If a state sees an increase in funding, the state could reduce its own spending as a response. There is some evidence that an increase in Title I funding could positively impact student achievement, but the response from states and how schools use these funds are both important factors.

Emphasis on Teaching Geometry

We saw above that one of the more important factors was the percentage of teachers placing a heavy emphasis on geometry. This is incredibly specific, but it could be part of a broader curriculum change. In fact, it was a part of a curriculum change. In 2010, Mississippi adopted Common Core standards, but so did other states like Alabama and Louisiana. In 2016, Mississippi did heavily change its curriculum, and the standards mention an emphasis on geometry: “understanding that geometric figures can be analyzed and classified based on their properties, such as having parallel sides, perpendicular sides, particular angle measures, and symmetry.”

Earlier, we saw that one of the key factors was the share of teachers placing strong emphasis on geometry. That may sound incredibly specific, but it could be part of a larger shift in curriculum. In fact, Mississippi did revise its standards. The state adopted Common Core in 2010, just like neighboring states such as Alabama and Louisiana. But in 2016, Mississippi made significant changes to its own curriculum, including explicit guidance to focus on geometry: “understanding that geometric figures can be analyzed and classified based on their properties, such as having parallel sides, perpendicular sides, particular angle measures, and symmetry.”

I’m not on a mission to go into the specific differences in curriculum across states (partially because I personally hate digging through bureaucratic language), so I graphed the emphasis placed on different subjects in math in three states. The charts below show that Mississippi began putting noticeably more weight on geometry than its neighbors. The first graph shows Mississippi, followed by Alabama and Louisiana. The national trend is closely in line with Lousiana’s trend.

But is there anything uniquely important about teaching shapes, lines, and other geometry concepts? And should states even worry about this, beyond aligning with NAEP standards? There is some evidence from the psychology literature that spatial reasoning skills are potentially predictive of achievement in STEM topics, where spatial reasoning is formally being able to visualize and manipulate objects in your mind. This isn’t exactly “geometry,” but the two could be related. Additionally, spatial reasoning could also be effective in helping children understanding numerical reasoning in the long run, which could also be effective for testing well on other math topics.

Should we actually care about NAEP scores?

Mississippi may have boosted its NAEP math scores by changing its curriculum, but that raises a bigger question: do NAEP scores actually matter? Economists often use NAEP results to study policy questions like: how did No Child Left Behind impact NAEP test scores? Do school spending cuts matter for NAEP test scores? and do school finance reforms impact NAEP test scores? In these studies, NAEP scores are the outcome, not the thing being explained. That means the research tells us how policies influence scores, but not whether higher NAEP scores themselves lead to outcomes like lifetime income or college success.

One study examines how NAEP scores relate to long-term outcomes by tracking entire birth cohorts, which can be thought of as children born in 2000 or 2007. It finds that higher eighth grade NAEP math scores are linked to higher adult income, even after accounting for factors like age, parents’ education, and race. The authors caution, though, that unmeasured socioeconomic differences could still be driving part of this relationship. Still, this is the only study I found that connects higher NAEP scores with potentially better outcomes later in life.

Even if NAEP isn’t a perfect measure of “cognitive skills” (think of these skills as the ability to mentally solve problems), there’s a large economics literature on how cognitive skills and school quality affect economic growth. Broadly, better schooling raises human capital, which is kind of an abstract measure of schooling, skills, and job experience that raises productivity in the workforce. One famous paper in this literature found that countries that improved cognitive skills (as measured by international test scores) experienced relatively faster increases in economic growth. Of course, all this assumes test scores accurately measure cognitive skills.

What are the main takeaways here?

So we know Mississippi’s fourth grade reading and math scores rose sharply, and eighth grade scores improved a little as well. The whole point of this post was to consider which factors might play a strong role here while not being able to show which policies were most effective for this; these are the careful statements economists must make. What we can say is that a stronger emphasis on geometry seems to matter, which helps point us toward the types of policies most closely tied to that shift.

This raises a few important questions: How much of Mississippi’s improvement simply reflects the fact that part of the NAEP test emphasizes geometry? Is geometry actually meaningful for students’ future success? And why might stronger geometry skills spill over into other subjects? These are all questions I can’t fully answer with current research, but it’s the level of skepticism and curiosity economists typically hold. And whenever you see “rankings” or “scores” in the media, it’s worth approaching them with the same skepticism.