Skip to main content

Welcome to a Grasp lesson

Create a free account to keep studying and create your own study path.

Lesson 14

Interpreting Research Evidence on TEFA’s Expected Impacts in Texas

Recommended resources

Hello. The previous lesson showed how a study can estimate the effect of a voucher offer without proving that attending a particular private school caused every later difference. This lesson asks a second question: even when a finding is credible for the people studied, how much should it change your expectations for Texas Education Freedom Accounts (TEFA)?

We will consider achievement, graduation and college enrollment, family satisfaction, and effects on students who remain in public schools. The aim is neither to dismiss research from other places nor to treat it as a forecast. It is to identify what each finding measures and what would have to be similar for it to travel to Texas.


Achievement: a result depends on the program and the measure

A useful research review by Patrick J. Wolf and Ron Zimmer brings together studies of private-school choice. Start with its account of achievement: it contrasts early, often city-based voucher studies with later evaluations of broader programs.

Vouchers and Similar Private School Choice Programs

Read Wolf and Zimmer's discussion of achievement to see why neither a uniformly positive nor a uniformly negative account fits the studies they review.

In “Key finding #1: The student achievement results of participants are mixed,” read pages 7–9, beginning where the authors introduce the achievement studies and ending with the Indiana results. Focus on the contrast between lottery-based studies and later statewide studies, and on whether the measured outcome is reading or math.

The distinction between achievement and attainment will matter throughout this lesson. Achievement studies usually measure performance on tests after a relatively short period. Attainment studies ask how far a student proceeds through schooling—for example, whether they graduate or enroll in college. A program might affect these outcomes differently; one cannot substitute for the other.

In the review, early lottery-based achievement findings range from negative to neutral to positive. The later examples resist a simple slogan, too. The authors describe a rigorous evaluation of the District of Columbia Opportunity Scholarship Program (OSP) that found no test-score effect after three years. They also describe substantial initial test-score losses in a Louisiana voucher study, with negative results persisting, though diminishing, over four years. In the Indiana study they discuss, math losses persisted after four years while reading results were null. These are findings about particular programs and comparisons, not estimates for TEFA.

Why might studies differ? The review notes changes in the students served, the scale and rules of programs, research designs, and tests. A nationally normed test asks how students perform relative to a broader testing population; a state accountability test measures performance against that state’s curricular standards. If private schools teach a different curriculum, the latter may capture some mismatch as well as learning. That possibility is worth considering, but it does not make an unfavorable score disappear. Nor can test choice alone establish what caused differences between studies.

For Texas, this evidence establishes a range of plausible outcomes and a reason to measure them. It does not establish that an ESA with TEFA’s eligibility, spending options, participating providers, and funding limits will reproduce any one voucher program’s test results.



Longer-term outcomes and family satisfaction

The previous lesson examined a New York voucher-offer study that found no overall effect on college enrollment, alongside a positive finding for a specified subgroup. A separate, federally commissioned evaluation of the District of Columbia OSP offers another instructive combination of outcomes. This June 2010 final report concerns applicants assigned through lotteries in 2004 and 2005; its dates matter when comparing it with later reports about the same program.

Evaluation of the Impact of the DC Opportunity Scholarship Program: Final Report | IES

Read the Institute of Education Sciences' summary for an example in which achievement, graduation, and satisfaction do not tell the same story.

Under “Summary,” first read from the account of lottery applicants and outcomes. Then read the following paragraph, beginning with the cumulative loss of students from DC Public Schools and ending with other student departures. Notice whose reports supplied the graduation and satisfaction findings.

The 2010 report finds no conclusive overall achievement effect. Yet, according to parent reports, 82% of students offered a scholarship graduated from high school, compared with 70% of applicants not offered one. That is a 12-percentage-point difference in a longer-term outcome—not a 12% increase, and not an administratively verified graduation measure in this summary. Parents whose children were offered or used a scholarship also reported higher satisfaction and school safety, while students’ own satisfaction and safety ratings were similar across offer groups.

These findings need not contradict one another. Test scores, graduation, and satisfaction ask different questions. The parent-reported basis of graduation and the difference between parent and student satisfaction are also part of the evidence, not details to omit when citing the result.

Return to the review for its broader discussion of attainment and satisfaction. It includes both stronger comparisons and observational findings that require more caution.

Vouchers and Similar Private School Choice Programs

Continue the review to compare college-related findings with the evidence on how families experience their chosen schools.

On pages 9–10, read the “Attainment” and “Parent and student satisfaction and perceptions of safety” subsections, from attainment through satisfaction. Pay particular attention to findings for an overall sample versus a subgroup, and to where the authors caution that an observational comparison may reflect selection bias.

The review reports null overall college-enrollment findings for studies of the District of Columbia OSP and Louisiana’s scholarship program, while the Louisiana study found a positive effect among students who began the program in high school. It also describes a positive Indiana college-enrollment association from an observational study. Because that comparison did not assign offers by lottery, differences between participating and comparison students are harder to rule out as an explanation. “Some positive attainment findings” is defensible; “vouchers cause higher college enrollment” is too broad.

For satisfaction, many studies in the review report favorable parent responses, but a family’s satisfaction is not a substitute for a test score or graduation record. It is a meaningful outcome in its own right: families can value a school’s environment, fit, or sense of safety. At the same time, a survey of families who chose to participate may reflect both their experiences and their prior preferences. The review also notes a later District of Columbia finding of no parent-satisfaction effect after three years, alongside improved student satisfaction. Keep the report, cohort, follow-up period, and respondent attached to each claim.

The Florida chart below illustrates another limit of a college-related result. It compares Florida Tax Credit (FTC) scholarship participants with non-participants, grouping participants by when they entered the program.

Percentages enrolling in a Florida public college within two years of expected high school graduation, comparing FTC participants and non-participants for elementary-or-middle-school and high-school entry groups. Asterisks mark differences reported as statistically significant at the 5% level.

For “any college” in the chart, the comparison is 45% versus 39% in the earlier-entry group and 48% versus 42% in the high-school-entry group. Both are six-percentage-point differences. But the outcome is enrollment in a Florida public college within a specified window—not enrollment at every college, completion of a degree, or a lifelong effect. The chart alone does not show how non-participants were selected for comparison or establish that FTC participation caused the differences. Statistical significance, indicated by an asterisk, does not answer that causal question. Moreover, FTC is a tax-credit scholarship, not a Texas ESA. Its chart is useful for understanding the scope of a claim, not for projecting a Texas college-enrollment rate.



Spillovers: outcomes for students who stay

So far, the question has been what happens to participating families. A public-school spillover concerns students who remain in public schools after a choice program begins or expands. Researchers might ask whether competitive pressure changes their achievement. That is different from asking what happens to a district’s revenue, even though a program could affect both.

Vouchers and Similar Private School Choice Programs

Read the review's account of public-school spillovers, keeping student test outcomes separate from fiscal effects.

On pages 11–13, in “Key finding #3: The systemwide effects of private school choice programs are generally positive or neutral,” read the “Systemic effects” and “Fiscal effects” subsections. Start at the discussion of competition and finances, ending at the final paragraph on policies that soften or delay district losses. Note the exceptions to the generally modest, positive test-score findings and the conditions that change fiscal results.

The review describes most estimates of competitive effects on remaining public-school students’ test scores as neutral or modestly positive. It also notes contrary findings in some settings and periods. This does not imply that every affected public school gains, or that a district’s finances improve. A change in average scores can reflect school responses and changes in who remains enrolled; a credible study has to address its comparison and the possibility that private schools are not randomly located.

Fiscal effects require a separate calculation—the subject of an earlier lesson. They depend in part on how many participants would otherwise have attended public schools, how enrollment affects district funding, and what costs a district can avoid when students leave. Funding an existing private-school student is different for a state budget from funding a public-school switcher. A modest positive spillover estimate for test scores cannot settle a disputed Texas budget forecast.

The 2010 District of Columbia report makes a narrower point about scale: it estimates cumulative transfers to OSP at about 3% of DC Public Schools enrollment over 2004–2009, against a larger background of students changing schools or leaving the district. That figure measures movement associated with that program. It is not an estimate of a test-score spillover or a transferable percentage for Texas.



What can early Texas evidence establish?

To transfer a finding responsibly, compare the program, the population, the outcome, and the time observed. These are practical checks, not requirements that two states be identical.

Question Why it changes the Texas inference
What was funded? A tuition scholarship and an ESA that can pay approved providers for several kinds of expenses offer different options.
Who could and did participate? A lottery among low-income applicants, a program drawing public-school switchers, and one also serving existing private-school or homeschool students can produce different participant and fiscal effects.
What outcome was measured? Test scores, parent satisfaction, graduation, and public-college enrollment are distinct findings; none stands in for all the others.
How long was follow-up? Early participation and satisfaction can be observed sooner than graduation, college enrollment, or sustained changes in public schools.

At TEFA’s outset, administrative counts can show awards, spending, participating providers, and take-up among defined groups if those data are collected and reported. They cannot, by themselves, show that TEFA raised achievement or improved public-school outcomes. An early family survey can report satisfaction among its respondents; it needs an appropriate comparison and careful follow-up before it can support a causal claim. Graduation and college outcomes require years of observation.

The most useful expectation is therefore conditional: older studies demonstrate that benefits and harms are both possible, that different outcomes can diverge, and that effects depend on design and context. Evaluating Texas will require dated, program-specific rules to define what TEFA actually offers, appropriate comparison groups, and enough follow-up for each claimed outcome.



Conclusion

Selected school-choice research offers evidence to test claims against, not a single predicted TEFA effect. Achievement findings are mixed; some attainment and satisfaction findings are favorable but vary in causal strength and in whose experience they measure. Studies of remaining public-school students often find neutral or modestly positive test-score spillovers, which do not resolve district or state budget effects.

In the final lesson, you will bring these distinctions together in a short verdict on a disputed Texas claim: what the dated rules require, what studies have observed, what fiscal forecasts assume, and what remains a value judgment or an open question.