The de Blasio administration can cite a rising graduation rate and faster-than-average state test score gains as evidence that its education efforts are paying off — but not scores from a national exam considered the most reliable yardstick for student achievement.

New York City’s scores on the National Assessment of Education Progress, known as NAEP, were flat from 2015 to 2017, according to new data released Tuesday. The only significant change since 2013, the year de Blasio was elected, was a 7-point decrease in the proportion of fourth graders considered proficient in math.

New York City is hardly alone in posting stagnant scores. The vast majority of states and districts — including New York state — also saw similar proportions of fourth- and eighth-graders meet the proficiency bar in reading and math in 2017 and 2015.

NAEP is important because it has long been the only way to compare student performance across states, a distinction that has persisted even as many states adopted similar standards for what students should learn and begun testing students on whether they meet those standards.

Students in New York state performed slightly worse than the national average last year. And students in New York City also posted lower-than-average scores than students in the 26 other districts that participate in a city-level comparison. (Students in both district-run and charter schools take the test.)

The exam, which is administered to a sample of students across each state and district, also serves as a check on state test scores. In New York, those have risen incrementally in each of the last two years.

In 2013, then-State Education Commissioner John King said the NAEP results underscored that New York’s students weren’t where they should be. “The reforms we’re implementing will help get them there,” he said, referring primarily to the Common Core learning standards the state adopted in 2010.

Four years after his comment, it appears that those changes haven’t had the dramatic impact King had hoped. But it’s also important to note that it’s difficult to draw conclusions about the benefits of specific policies based on the results. NCES, the federal agency that administers the tests, warns against it — as do many researchers who focus on education.

“You should never think of NAEP, or even the state assessments, as a referendum on a particular package of policies or reforms,” Aaron Pallas, a researcher at Columbia University’s Teachers College, told Chalkbeat in 2015, the last time NAEP reading and math results were released for the city and state. (They were flat then, too.)

That doesn’t mean that local and state education officials don’t tout the scores when they rise — or face tough questions when they fall. This week, New York City’s new chancellor, Richard Carranza, could find himself facing questions about the scores in Houston, the district he just left. There, scores were down by notable — though not statistically significant — margins in all four test categories.

This year’s scores come amid a broad shift to computer-based testing that some education officials have cautioned would tamp down scores, particularly for groups of students with little online testing experience and in states that have not made the switch for their own exams. NCES downplayed the concerns, saying that it had taken extensive steps to account for the dynamic when analyzing scores.

Among the states with little experience with online testing: New York, where this year fewer than 300 of nearly 700 districts are having some schools administer state tests online, starting this week.