Analysis reveals scores don't jibe

Second school's test scores also in question

Wade Spees
The Post and Courier

Discrepancies in Sanders-Clyde Elementary's recent test scores raise new questions about the school's record of achievement.

In 2007, Sanders-Clyde Elementary students' scores on the state Palmetto Achievement Challenge Test were among the best in Charleston County.

That same year, Sanders-Clyde students also took a lesser-known test and had some of the weakest scores in the county.

The Post and Courier uncovered the gap after requesting test scores from the school district.

There is less than a one percent likelihood that chance alone explains the wide gap, according to a statistical analysis requested by the newspaper.

The discrepancy in Sanders-Clyde's test scores raises new questions about the school's record of achievement. Further analysis revealed testing irregularities at another school, E.B. Ellington Elementary in Ravenel.

Like Sanders-Clyde, that school had a gap between its students' performance on the PACT and the other test, Measures of Academic Progress, or MAP.

In Ellington Elementary's case, the State Law Enforcement Division charged a fifth-grade teacher, Felicia Perrineau, with breaking state testing laws, and the state Department of Education invalidated the school's fifth-grade test scores from 2007.

Now, state law enforcement officials are looking at downtown Sanders-Clyde after a striking drop in test scores this spring.

PACT vs. MAP

Until a couple of weeks ago, Sanders-Clyde was one of the district's success stories. Former Principal MiShawna Moore transformed one of the lowest-scoring schools in the district to a place where students scored higher than district and state averages.

But a high number of student eraser marks on the 2007 PACT tests and improbable one-year academic gains triggered district officials to keep a close watch for the first time on the school's testing in May.

PACT test results released a few weeks ago revealed drops in every subject and in every grade.

The PACT test is one of two major exams that Charleston County elementary and middle school students take, and those scores are used to gauge schools' progress or failure. Schools with strong PACT scores can receive money and awards. Schools with low PACT scores can be taken over by the state.

There's only one version of the PACT test for every grade, and the test involves No. 2 pencils and paper. The tests that span several days are locked up daily.

Although strict rules govern PACT testing procedures, state officials concede that honesty is what stops school leaders from manipulating results.

The second big test students take is MAP. These test results mostly are used to guide teachers' instruction throughout the year.

MAP tests are given on the computer, and students' scores are generated electronically. The questions change for each student, depending on whether he or she answers correctly. That means a class of 30 students could be working on 30 different test questions.

"You can't cheat with MAP," said Janet Rose, the executive director of assessment and accountability for Charleston County schools. The only exception would be if a teacher stood over a student and coached him or her on individual answers, she said.

While the PACT and MAP tests are different, students' scores should roughly match. That hasn't happened at Sanders-Clyde.

Sanders-Clyde gap

The Post and Courier requested Sanders-Clyde's PACT and MAP scores for the past four years. After noting the magnitude of the test-score gaps, the newspaper asked the school district to statistically analyze the gaps among county schools for 2007 results.

Sanders-Clyde students have scored far higher on the PACT exam than the MAP test for the past four years. The gap between proficient and advanced students' performance on PACT and MAP in 2007 was the biggest in the county.

Sanders-Clyde was the only county school where the likelihood is less than 1 percent that the scoring gap occurred by chance alone in English and math, according to a statistical analysis of the 2007 scores.

Rose characterized the scoring discrepancies as "unusual" and "much greater" in comparison to other schools.

The gap between the scores widened every year, with paper-and-pencil PACT scores far outpacing the growth that showed up on the computerized MAP test.

For instance, the percentage of Sanders-Clyde students scoring proficient and advanced was on average 10 points higher on PACT compared to MAP. That was three years ago.

Two years ago, the percentage of students scoring proficient and advanced was an average 33 points higher on PACT than MAP. In 2007 that number increased to 54 points, a bigger gap than any other county school.

The score gap lessened significantly this past year after the district stationed monitors in Sanders-Clyde. For the first time, MAP scores were higher than PACT scores for proficient and advanced students, by an average of 6 percentage points. That put the gap in the school's scores more in line with other county schools.

Two other Charleston County schools, Dunston Elementary in North Charleston and E.B. Ellington Elementary, also had much higher PACT scores compared to MAP, according to the statistical analysis.

The probability that the gap in proficient and advanced students' scores in English at E.B. Ellington happened because of chance alone is less than 1 percent, the analysis showed. The gap for the same students' scores in math wasn't statistically significant.

Gaps elsewhere

It's illegal for a teacher to share PACT test questions with students before the exam. At Ellington Elementary in 2007, a teacher obtained a copy of the previous year's PACT test from someone in another school district, and the teacher gave that test to her students to study and prepare for PACT, Superintendent Nancy McGinley said.

PACT tests aren't exactly the same from year to year, but the state reuses some questions. The teacher said she didn't know it was a copy of the previous year's exam and thought it was a practice test, McGinley said.

District officials turned the information over to the state, but weren't informed that Perrineau, the Ellington teacher, had been charged.

Perrineau entered a pre-trial intervention program, and her charges were dismissed.

The state Board of Education could revoke or suspend Perrineau's teaching certificate.

That has not happened, and state Education Department Spokesman Jim Foster could not say whether Perrineau was under further investigation. The school district did not rehire Perrineau, who along with Moore could not be reached for comment.

At Dunston Elementary, the gap wasn't significant among its highest scoring students, but it was among those who scored on a basic level or above. The probability that the gap in math and English happened because of chance alone is less than 5 percent, the analysis showed.

Rose said she expected a larger score difference at Dunston because only the third graders at the K-3 school take both tests.

That's a much smaller number of kids compared to other schools where multiple grades are tested, and there's a greater likelihood for bigger score differences with a small number of students, she said.

In the case of Sanders-Clyde, officials weren't suspicious of the school's gains until they received PACT results in the fall of 2007, and at that point they checked with the state and decided to closely watch the school's testing the following spring, McGinley said.

McGinley has looked at MAP scores to highlight schools' mid-year progress, but those scores weren't studied especially closely, she said. The results mostly were used as an instructional tool to help teachers better target students' instructional needs, she said.

It's 20/20 hindsight to go back and say Sanders-Clyde should have been flagged as a major issue, she said, given the discrepancy in its PACT and MAP scores, but officials weren't looking at MAP results to see whether they aligned with PACT.

Given what's happened, the school district should watch this more carefully in the future and share any significant score differences with the state, McGinley said.

"This is something we have to figure out how to pay attention to going forward," she said.

Reach Diette Courrégé at 937-5546 or dcourrege@postandcourier.com.