New data that is meant to help the public understand how Colorado students who are learning English as a second language are performing on state tests is in many cases baffling and inconclusive.

The state released those numbers last week as part of a large publication of results from the 2016 and 2017 PARCC tests broken down by different students groups. Chalkbeat has been publishing findings from the data since. We’ve looked at the state’s achievement gap based on race and ethnicity, socioeconomic status, and special needs.

Today, in our last installment, we try to understand the gaps between the state’s English language learners and their native-English speaking peers.

It’s not easy. In fact, it’s nearly impossible for some school districts based on the data that is publically available.

That’s because of the way the state classifies English language learners and non-English language learners.

The state breaks down students into four categories of English proficiency.

The first three categories are students who are enrolled in some sort of English language development program. They’re classified as either being “non-English proficient,” “limited English proficient,” or “fluent English proficient.” (Students that reach proficiency can remain in development programs for two years of monitoring.)

The fourth category is a little more confusing. It includes students who are native English speakers, students who are learning English as a second language but are not enrolled in any development courses, and students who have exited out of a program designed to help them learn English.

Grouping students this way can mask how well students who were enrolled but tested out of language development programs are performing. This makes it difficult to check against claims made by some districts, such as Denver Public Schools, about the efficacy of their language learner programs.

It also limits us in comparing how well English learners are performing compared to native English speakers, and how well English learners not in those development programs are scoring on tests compared to their peers in those programs.

Another reason why it’s difficult to draw any sound conclusions is because of the state’s data suppression rules, which were designed to protect student privacy. When the state applies those rules to results in the seventh grade, we have no idea how students with no or limited English skills in the state’s 10 largest districts fare.

Take a look at the results:

Fewer data points are missing in the fourth grade results:

This data alone makes it difficult to gauge why some districts’ more advanced English language learners are outperforming other groups of students, and why they’re not in others.

We asked the state to help us make sense of the data, using Boulder as an example. Other achievement gaps, like for race and poverty, are particularly acute in Boulder. The only way they could provide us with insight was by collapsing some different groups of English learners into larger categories to generate larger numbers.

“We continue to work to try to find the right balance between protecting student data privacy and providing meaningful data for the public,” said Joyce Zurkowski, the education department’s executive director of assessment in an email. “Moving forward, we may want to consider collapsing (different groups) to increase the number of students. We could also combine across tests, creating an elementary and middle school group.”

— Chalkbeat reporter Yesenia Robles contributed.