Numbers game

Colorado is about to release a torrent of test results. Here are four storylines worth watching.

Sheridan School District sixth grader Monica Dinh takes part in a practice session last year (Photo By Craig F. Walker / The Denver Post)

The state education department is scheduled Thursday to publicly release a mammoth amount of data detailing how Colorado students performed on last spring’s standardized tests.

We’ll get to dive into state, district and school results from English, math, science and social studies tests, the PSAT and SAT, and student academic growth, which tracks how much students learn each year compared to their academic peers.

The data — beloved or loathed depending on which educator you ask — is supposed to gauge how well students grasp the state’s academic standards that are designed to prepare them for either college or a career.

The state also uses the results, along with other factors such as graduation rates, to issue quality ratings for schools and districts. And in some instances, teachers are rated based on the data.

Here is background and some storylines to keep in mind in advance of the release:

First a reminder of where we stand:

Three years ago, the state made a monumental shift in its testing system. Colorado was one of about a dozen states to drop paper-and-pencil standardized tests in favor of a new multi-state computer-based test.

The PARCC tests would measure critical thinking, a major component of the state’s new academic standards, which devalued rote memorization.

Prior to the first release, school officials in Colorado and across the nation warned that test scores would likely be low considering the newness of the academic standards and tests.

Indeed, they were.

In 2015, only 43 percent of fourth graders met the state’s expectations on the English test. Math was worse: Only 37 percent of third graders were able to complete math equations at grade level.

In 2016, the state saw a slight uptick in scores, mirroring national trends.

However, state officials worried about how far behind students with learning disabilities were compared to their peers.

Here’s a look at the changes in test scores in English and math:

English

Math

 

With three years of data from PARCC, we can — finally — talk about trends. But what are we going to learn that we didn’t already know?

For the last two years, state and school district officials have warned about two things: First, don’t compare the results of PARCC to that of previous standardized tests. Second, they said we needed three years of data to pinpoint trends in student performance.

Why three years?

Derek Briggs, a professor at School of Education at the University of Colorado Boulder who also sits on the technical advisory board for PARCC, said one reason why we might need three years of data is because of exaggerated bumps sometimes found in the second year of a new standardized test.

“One explanation for this sort of trend was that it would take teachers/schools a year to figure out the emphasis on the new assessment, so in the first year, the alignment between teaching and instruction isn’t optimal, so student performance in the first year is depressed,” he said in an email. “Then in the second year, it snaps back up once instruction and assessment are better aligned.”

Briggs added that so far, no state that updated its test to align to the Common Core State Standards like Colorado did had a second year bump.

So, now we have three years of data: What can we say?

It’s difficult to make sweeping declarations about state trends — especially in a local control state where so many decisions about what students learn is made at the school and district level.

But Juan D’Brot, a senior associate at the Delaware-based Center for Assessment, said that at the three-year mark, school officials and parents alike can start to better understand what’s working or not at individual schools.

“It can serve as a gut check about a school’s general performance over time,” D’Brot said. “If you have three points that are moving upward or constantly moving downward, we can quickly create a story around that.”

It’s more difficult to draw conclusions if a school’s results are less consistent, he said.

And there are some state-level benefits.

“This trend data can help the state evaluate their own efforts to work with districts and schools,” he said. This is especially valuable when school leaders use a variety of data points including patterns of student growth.

The state is suppressing data in an effort to “protect student privacy.” How much will be redacted?

Colorado was once considered one of the most education data-friendly states. But beginning with the first release of PARCC data in 2015, the state began blacking out more school-level data than it had in the past.

The effects of the new so-called “suppression rules” were even more pronounced in the state’s 2016 release. The state shielded roughly 4,000 data points that year, frustrating education reform advocates who say this data helps parents make better decisions about schools.

Stay tuned to see what we won’t learn about school performance due to these rules after Thursday’s release.

After two years of delayed and drawn-out data releases, the state is giving us everything on time and all at once. But the promise of getting data back quicker is still elusive.

In 2015 and 2016, testing data dribbled out of the state education department over several months — state-level results first, then school level, then student growth data. This was a departure from a decades-long routine of releasing test score data in August.

On Thursday, the state will release almost everything all at once. (District and school performance data disaggregated by different student groups is expected within a month.) This is a major victory for the state and the makers of PARCC because one of the longest-running criticisms of the test was how long it took to get data back to schools.

Schools received their results in June, the earliest data has gotten back to the schools since the state switched to PARCC.

But the timeline still falls short of one of the promises of new tests and the demands of the State Board of Education, which going forward wants data back to schools within 30 days.

Is the state’s gradual move away from PARCC at the high school level working to curb the opt out movement?

In 2015, Colorado became one of the nation’s epicenters for the testing opt out movement. Thousands of high schoolers, backed by their parents, refused to take the PARCC exams, claiming they served no educational purpose.

In some cases, entire schools sat empty during the state’s testing window.

In response, lawmakers eliminated some high school tests and changed others. In 2016, more high school sophomores took the state’s tests than the year before. Policymakers hope additional changes at the ninth grade level, set to take effect next spring, will move even more families back to the state’s testing system.

Will the trend continue? We’ll find out on Thursday.

And finally, here’s a roundup of previous coverage you might find helpful:

union power

Gutting Wisconsin teachers unions hurt students, study finds

PHOTO: Creative Commons / Michael Vadon
Wisconsin Gov. Scott Walker in 2015.

The high-profile fight to limit union power was replete with drama — including a recall election and state legislators fleeing to neighboring states.

In the 2011 battle in Wisconsin, Republican Gov. Scott Walker ultimately came out the victor. The controversial law passed, Walker won the recall, and the Democratic-aligned unions have lost much of their power.

But new research points to other losers in the fight: students in the state’s already struggling schools.

The first study to assess how Wisconsin’s high-profile weakening of unions, particularly teachers unions, affected students finds that it led to a substantial decline in test scores.

The findings come as the U.S. Supreme Court is set to hear arguments for a case, known as Janus, that could dramatically scale back union power across the country — essentially taking aspects of the Wisconsin model national. And they give credence to concerns from unions and their defenders that weakening teachers bargaining power would ultimately make schools worse, not better.

A report from the left-leaning Center for American Progress released Wednesday highlights this research — and the fact that teacher pay and average experience declined in the wake of the law, known as Act 10 — to argue that weakening unions ultimately harm schools.

“Those concerned about the quality of public education — and of all public services — should understand that Wisconsin’s Act 10 and associated budget cuts have not had the positive impact on education that its proponents claimed it would,” the CAP report argues.

Still, the research, which has not been formally peer-reviewed, only assesses the short-term impact of Wisconsin’s law. It adds to a complicated set of research findings on unions that doesn’t render a clear verdict.

Short-term effect in Wisconsin is negative, especially for low-achieving schools

The new research looks at the effects of Wisconsin Act 10, which became law in 2011 and severely limited the scope of collective bargaining and allowed members to opt of unions.

The paper’s author, Jason Baron, took advantage of what was essentially a natural experiment set up by the law. Act 10 did not affect all school districts at once — a handful of school districts were allowed to maintain union rules until their existing contract expired up to two years later. That helped isolate the immediate impact of the law.

Baron found that weakening unions led to declines in test scores, particularly in math and science. The effects were fairly large, comparable to sharply increasing class sizes. And the harm was not evenly distributed: Schools that started out furthest behind were hurt the most, while higher achieving schools saw no impact.

Other research may help explain why.

The law led to big cuts in teacher compensation, particularly for veteran teachers and especially in health insurance and retirement benefits, according to one paper. There was also a spike in teacher retirement immediately following the law’s passage.

As compensation drops, it may become harder for district and teachers to recruit and keep teachers. An increase in retirement also reduces teacher experience, which has been linked to effectiveness.

Another study found that some Wisconsin districts moved from a single salary schedule to a performance-based pay system after Act 10’s passage. Those performance pay systems were more likely to be adopted by higher-achieving districts, potentially allowing them to lure effective teachers away from struggling schools.

“Following Act 10, high-performing schools filled vacancies from teacher retirements by poaching high-quality teachers from low-performing schools through attractive compensation schemes,” the paper concludes. So while those retirements might have hit all districts equally, high-performing districts were better able to make up the difference — at the expense of low-performing schools.

There is one study that complicates the narrative in Wisconsin. As retirements spiked, it found that academic achievement actually increased in the grades that teachers left. It’s not clear what explains this.

The larger question of how teachers unions affect learning remains up for debate

A number of other recent studies have examined the relationship between teachers unions and student outcomes outside of Wisconsin. The results aren’t consistent, but the trend has been more positive for unions of late. A caveat: Some of these studies have not been published in peer-reviewed academic journals.

  • On recent efforts to weaken unions: Research in Tennessee found that it led to a drop in teacher pay, but had no effect on student test scores. But a study of four states, including Wisconsin, that recently weakened unions found evidence of reduced teacher quality as a result.
  • On what happens when charter schools unionize: Two studies in California came to differing conclusions. One found that when charters unionize, student test scores go up, but the other showed no impact.
  • On the initial rise of collective bargaining: Another paper finds that students who went to schools where districts negotiated with unions earned less money and were more likely to be unemployed as adults. But this study looks at a fairly old data set — examining those who attended schools between 1965 and 1992.

Meanwhile, it’s not clear if any of this research is likely to influence the Supreme Court, as it considers the Janus case that could make life more difficult for unions. Last month, Chief Justice John Roberts called empirical studies on political gerrymandering “sociological gobbledygook.”

study up

Trump education nominee pleads ignorance about high-profile voucher studies showing negative results

At his confirmation hearing, Mick Zais, the nominee to be second-in-command at the Department of Education, said that he was not aware of high-profile studies showing that school vouchers can hurt student achievement.

It was a remarkable acknowledgement by Zais, who said he supports vouchers and would report to Education Secretary Betsy DeVos, whose signature issue has been expanding publicly funded private school choice programs.

The issue was raised by Minnesota Sen. Al Franken, who asked whether Zais, who was previously the South Carolina schools chief, was “aware of the research on the impact of vouchers on student achievement.”

He replied: “To the best of my knowledge, whenever we give parents an opportunity to choose a school that’s a good fit for their child the result is improved outcomes.”

Franken responded, “No, that’s not true. The academic outcomes for students who used vouchers to attend private school are actually quite abysmal.”

Franken proceeded to mention recent studies from Louisiana, Indiana, Ohio, and Washington, DC that showed declines in test scores after students move to private schools with a voucher.

Zais responded: “Senator, I was unaware of those studies that you cited.”

Franken then asked if Zais’s initial response expressing confidence in school choice was anecdotal, and Zais said that it was.

What’s surprising about Zais’s response is that these studies were not just published in dusty academic journals, but received substantial media attention, including in the New York Times and Washington Post (and Chalkbeat). They’ve also sparked significant debate, including among voucher supporters, who have argued against judging voucher programs based on short-term test scores.

Meanwhile, it’s worth noting that the research confusion was a bipartisan affair at Wednesday’s confirmation hearing.

Although Franken, who referred to a New York Times article on voucher research in his question, was broadly accurate in his description of the recent studies, he said that a DC voucher study showed “significantly lower math and reading scores”; in fact, the results were only statistically significant in math, not reading.

Franken also did not mention evidence that the initial negative effects abated in later years in Indiana and for some students in Louisiana, or discuss recent research linking Florida’s voucher-style tax credit program to higher student graduation rates.

In a separate exchange, Washington Sen. Patty Murray grilled Jim Blew — the administration’s nominee for assistant secretary for planning, evaluation, and policy development — on the performance of Michigan’s charter schools. Murray said that DeVos was “one of the architects of Detroit’s charter school system,” describing the results as “disastrous for children.”

Blew disputed this: “The characterization of the charter school sector in Detroit as being a disaster seems unfair. The most reliable studies are saying, indeed, the charter school students outperform the district students.”

Murray responded: “Actually, Michigan’s achievement rates have plummeted for all kids. In addition, charter schools in Michigan are performing worse than traditional public schools.”

(Murray may be referring to an Education Trust analysis showing that Michigan ranking on NAEP exams have fallen relative to other states. The study can’t show why, or whether school choice policies are the culprit, as some have claimed.)

Blew answered: “The most reliable studies do show that the charter school students in Detroit outperform their peers in the district schools.”

Murray: “I would like to see that because that’s not the data that we have.”

Blew: “I will be happy to get if for you; it’s done by the Stanford CREDO operation.”

Murray: “I’m not aware of that organization.”

CREDO, a Stanford-based research institution, has conducted among the most widely publicized — and sometimes disputed — studies of charter schools. The group’s research on Detroit does show that the city’s charter students were outperforming similar students in district schools, though the city’s students are among the lowest-performing in the country on national tests.