UPDATED: PARCC Sets Cut Scores for Test, But Won't Say What They Are
The PARCC testing consortium made a pivotal decision Wednesday: It established the cut scores for its test. But it chose not to disclose most of those scores, or to say what portion of students performed at each of the five achievement levels.
At a meeting in Washington, PARCC's governing board voted to approve the cut scores for the English/language arts and math tests in grades 3-8. But it disclosed only one of the scores that mark the thresholds between the levels of mastery. On a scale of 650 to 850, students will need to score a 750 to reach Level 4, which in grades 3-8 connotes a "strong command" of the standards, and in high school signifies college readiness.
PARCC's governing board did the same thing Aug. 14, when it voted to approve cut scores for the high school tests. All officials would say is that each test will be scored on a scale of 650 to 850, and that students would need to score 750 to reach Level 4. They would not disclose the threshold scores needed to reach Levels 2, 3 and 5. They didn't reveal, either, what portion of high school students scored at each of the levels on the exam.
PARCC spokesman David Connerty-Marin would say only that states are still finalizing their data, so it's too early to disclose how students performed on the test, which was given for the first time this past spring.
He didn't say why PARCC wouldn't release the actual cut scores, and testing experts were baffled by that as well. Several psychometricians we consulted said they had no idea why PARCC couldn't disclose the scores that students would need to meet to reach each performance level on the test.
UPDATE: PARCC decided Thursday afternoon to release the rest of the cut scores. Updated "mock" score reports on its website, which had earlier had placeholder numbers for cut scores, now show the actual cut scores the board approved, according to PARCC assessment chief Jeff Nellhaus. They are: 700 to score at Level 2, 725 to score at Level 3, and, as previously reported, 750 to score at Level 4. The cut point for Level 5 will vary somewhat by grade and subject, but will be around 803, Nellhaus said during a webinar for the Education Writers Association.
"Cut points on the reporting scale ought to be something they're willing to let go of," said one assessment expert who's very familiar with PARCC's work to establish cut scores. "I can't understand why they wouldn't do that."
That same expert said, however, that the consortium's decision not to release "consequence data"—how students scored on the test, given the newly established cut scores—was understandable. He said that some states might still be finalizing data—for example, making sure that if they're reporting scores by subgroup, that they're placing students correctly into those groups.
Eleven states and the District of Columbia administered the PARCC exam. States are keeping one another apprised of when they will release test results, and they anticipate doing so between mid-October and the end of November. The consortium has developed score reports that many or all of its states are likely to use.
By contrast, the 18 states that used the Smarter Balanced exam last spring are each reporting results their own way. California is the most recent to do so; it released its results yesterday.
When Smarter Balanced set its cut scores last November, it released the scale, the cut points, and the projected proportions of students it expected to perform at each of the four categories on its test. (The student-performance data were only projections, since Smarter Balanced established its cut scores based on data from the 2014 field test, not from the operational test in 2015.)
PARCC's decision to delay the disclosure of its cut scores and its student-performance data prompted some questions and skepticism in the assessment world. Douglas McRae, who oversaw the design of California's assessments in the 1990s as a top executive at CTB/McGraw-Hill, said he is uneasy about PARCC's "lack of transparency and data."
When California established the cut scores for its STAR program in the early 2000s, he said, the cut-score recommendations made by panels of experts, as well as those made by the department of education's staff, were blended into a memo that was sent to the state board of education, and became part of the public record for the regional meetings it held to gather public input on the cut scores. How students would perform at each test level based on those recommended cut scores was also part of that record, he said.
"It's very unusual for a test-maker to announce final cut scores and not release estimated proportions of kids at each level [of the test]," he said in an email to Education Week.