Did More Arts Education Raise Test Scores? A Lottery of 10,000 Students Answers

Reading didn’t move. Math didn’t move. Science didn’t move. Three other things did

Introduction: The Claim That Art Makes You Smarter

Two things are often mixed together in the mind of a parent signing a child up for art or music. One is wanting the child to enjoy it. The other is the expectation that “this will help with schoolwork in the end.”

That second expectation is widespread. Art builds creativity, creativity carries into learning, and it comes back as grades. The logic shows up in academy advertising and in education policy debates alike.

And yet in actual schools that expectation has worked in reverse. In the US, arts education has been shrinking continuously since the 1980s. As test-based school accountability spread, schools funneled resources toward tested subjects. In a 2012 survey, about half of public school teachers said instructional time and resources for art and music had been reduced. For math the figure was 10%; for English, 12%.

The cuts were not evenly distributed. Schools with more low-income, Black, Hispanic, and immigrant students were cut harder. From the 1980s to the 2000s, arts education participation fell 25 percentage points for Black students and 19 points for Hispanic students. For White students, 1 point.

Both sides of the debate stood on thin evidence. Neither the claim that art raises scores nor the premise that art gets in the way of scores had been properly tested, because no study had verified it with random assignment in real schools.

A team from Texas A&M University and the University of Missouri ran that experiment in Houston.

The Core Question

The question is simple.

“If you substantially expand arts education opportunities in a school, what actually changes for students?”

What mattered here was which outcomes to look at. The team did not look only at scores. Alongside attendance, disciplinary records, and standardized test scores from administrative data, they measured school engagement, empathy, and college aspirations by survey.

How They Studied It: Forty-Two Schools Split by Lottery

Why a Lottery Was Possible

Houston had a program called the Arts Access Initiative (AAI) — a coalition of some 50 local cultural institutions and philanthropies supporting arts programming in schools with thin arts provision.

The problem was that more schools wanted in than there were places. In the first two years, 60 schools applied and only 21 could be served. That made research possible: AAI’s leadership, in consultation with the research team, agreed to assign by lottery.

Assignment was random but paired to increase precision. Schools were matched on level (elementary/secondary), student composition (share on free lunch; Hispanic, Black, and White shares), existing arts resources (presence of arts specialists, number of partnerships), and prior academic achievement, and within each pair one school was drawn into the treatment group and one into the control.

The final sample is 42 schools and 10,548 students in grades 3–8, in the seventh-largest school district in the US.

What Was Provided

Participating principals had to commit $1–10 per student from their school budget. Local philanthropies matched it 1:1.

The result was that treatment schools had on average **$14.67 per student per year** to spend. Counting only what the school itself put in, that is **0.08%** of Houston's per-student budget ($9,663).

With that money schools formed partnerships with local arts organizations. Treatment schools gained on average 5.03 more partnerships than control schools, and 9.86 more arts educational experiences over a single school year.

What Kind of Art

• Discipline: theater 54%, music 18%, visual art 16%, dance 12%
• Format: teaching-artist residencies 33%, in-school performances and workshops by professional artists 31%, field trips 27%, after-school programs 9%

The most-chosen format was the teaching-artist residency — an artist forms a relationship with the school for a semester or a year and teaches an entire grade once or twice a week. Especially popular in elementary schools without a full-time arts specialist.

Before the experiment began, the two groups’ arts partnership counts were statistically identical (2.76 treatment, 2.80 control).

Results: Three Things Moved, Four Did Not

In models controlling for prior achievement, three results were statistically significant.

What did not move:

• Reading: no effect
• Math: no effect
• Science: no effect
• Attendance: no effect
• In the full sample, school engagement, college aspirations, and cognitive empathy were also not significant

That is: expanding arts education did not raise reading, math, or science scores.

Why Only Writing

Writing was the only tested subject that moved. The team’s account here is remarkably candid.

There are many dubious claims about arts education producing transfer effects to other tested subjects, but there is little theoretical basis or rigorous evidence to support such expectations.

Even while explaining a favorable result in their own study, they draw a line against the transfer story generally.

So why did writing rise? The team’s explanation is that writing, unlike other tested subjects, genuinely overlaps with arts activity. Many of the programs included opportunities for self-expression and reflection, and some focused on creative writing or integrated writing.

There is one further piece of support. Texas’s writing test (STAAR) has two parts: multiple-choice items covering mechanical skills like spelling and grammar, and an actual written composition. Splitting the scores:

• Composition: 0.18 SD increase (p < .001)
• Grammar and spelling: 0.09 SD increase (p = .02)

The side that expresses your own thinking rose twice as much as the side that internalizes rules. Consistent with the reading that what children gained from arts activity was expressive power.

The Effects Were Larger for Elementary Students

Splitting results by school level makes the difference clear.

Elementary students showed a larger writing effect (0.20 SD), and also positive effects on school engagement, college aspirations, and cognitive empathy — none of which were significant in the full sample.

The team gives three reasons. First, AAI concentrated on elementary schools in its first two years, and 86% of the sample was elementary. Second, elementary schools often lack an arts specialist, making them more likely to receive a deep residency lasting a semester or more. Third, educational interventions generally have larger effects earlier.

Middle school students, by contrast, showed declines in school engagement and college aspirations. The team’s explanation: middle schools were not the focus of support, programming was scattered and one-off, and the share of students who actually experienced it was low. This may be the backfire that comes from raising expectations without delivering enough. The team adds, though, that this negative result is not statistically robust and may be chance.

Practical Takeaways

The Expectation That Art Raises Grades Should Be Set Down

This study substantially expanded arts education, randomly assigned 10,000 students, and found that reading, math, and science scores did not move at all. It is the answer from the most rigorous design in this field.

If the reason for enrolling a child in the arts is “it’ll help with schoolwork in the end,” that ground is weak.

But Art Did Not Eat Into Grades Either

The result has to be read in the other direction too. Adding arts time cost the tested subjects nothing. The team’s concluding sentence aims at exactly this point.

Despite the logic of curriculum narrowing, we find no evidence that increasing arts educational experiences is harmful to academic growth in tested subjects.

Cutting the arts for the sake of test scores has no evidentiary basis.

What Did Change Is Not Trivial

Discipline fell, the ability to put one’s own thinking into writing grew, and the desire to help someone treated unfairly increased. These don’t show up on a test sheet, but recent research reports that such indicators predict later life outcomes better than test scores do — high school graduation, college enrollment and persistence, rates of trouble in school.

And this came from roughly $15 per student per year.

Limitations

The team names several. And there are things you must weigh when reading the results.

• The team itself lowered confidence in the discipline result. Under an alternative statistical test (randomization-based inference) it did not reach the significance threshold (p = 0.14). The team states they are “somewhat less confident in the statistical significance of this result.”
• Emotional empathy was a single item. The item is “I want to help people who are treated unfairly.” A multi-item scale was intended, but its reliability fell short of standards and was not reported; only the one easily readable item made it into the results. The team writes that “as a single item it must be interpreted with caution.”
• These are schools whose principals volunteered. Only principals who committed to expanding arts education applied. Random assignment balances motivation between treatment and control, but whether the same results would appear in schools with uninterested principals is unknown.
• The condition of the control group may have inflated the effects. AAI recruited from the schools with the least arts provision, so the comparison schools were severely arts-deprived. There is no guarantee the same program produces the same effect in schools where arts education is already in place.
• Houston is a large city rich in arts resources. Cultural-institution partnerships with schools are active there. Smaller cities and rural areas may lack this infrastructure.
• We do not know what produced the effect. Theater, music, dance, and visual art were mixed, and formats varied. The study cannot distinguish which kind of arts experience was decisive.
• Only short-term outcomes were examined. Whether these effects accumulate, persist, or fade over time is unconfirmed.
• About 90% of the sample qualified for free lunch, leaving too little variation to examine differences by income.

Closing Thoughts

The answer this study gives is different from the one people hoped for, but more accurate.

Arts education did not raise test scores. It changed how children express their own thinking, and how they treat other people.

The sentence “children who do well in the arts do well academically” is not supported by this experiment. But the experiment also collapses the opposite sentence. Art is not a tool for academics — and it does not obstruct them either.

That the reason for letting a child draw and touch an instrument does not have to be grades — perhaps that is what this study confirmed.

Source: Bowen, D. H., & Kisida, B. (2023). Investigating the causal effects of arts education. Journal of Policy Analysis and Management, 42(3), 624–647. https://doi.org/10.1002/pam.22449