Test's publisher says scoring mistake made

Error affects 10,000 exams taken in Vista

VISTA — The state’s school test publisher, already fixing a
massive scoring error that delayed the release of statewide test
results by two weeks, has discovered it has incorrectly scored about
10,000 exams in Vista.

Harcourt Educational Measurement acknowledged yesterday that it
had made a mistake in comparing scores of students in Vista’s year-
round schools with the national sample group. As a result, the
performance of individual students, schools and the district has been
underranked in reports the San Antonio-based publisher sent Vista
last week.

??? The same mistake has occurred on tests from year-round schools in
Long Beach and Oxnard.

Harcourt detected the mistake in the Vista scores and called the
district yesterday. The district clerical pool had spent days
folding reports and stuffing envelopes to be mailed to parents next

The latest error means Vista parents probably will not receive
their children’s corrected scores in the mail for about three weeks.

California Department of Education spokesman Doug Stone said
yesterday that the state has 126 districts with year-round schools,
and that he didn’t know if the latest error would affect the expected
release of scores on July 15.

“The CDE will not post the (test) results until we can assure the
public that the data are complete and accurate,” Stone said.

The department was scheduled to post county, district and school
scores on the Internet on June 30, but the state sent the data back
to Harcourt for corrections after the publisher admitted it had
miscalculated the average scores for limited-English-proficient and
English-speaking students.

Computer programmers mistakenly included the scores of students
who had already become fluent in English in the limited-English-
proficient category, a mistake Gov. Gray Davis called a “colossal

That error is likely to have falsely inflated the scores of
English learners.

It was not detected until districts across the state, including
Oceanside, had released results on their own that showed seemingly
huge improvements in English learners’ average scores. The scores
already had become fodder in the debate over Proposition 227, the
voter-approved ban on most bilingual education in California that
went into effect in the just-finished school year.

The mistake with English learners affected only group averages,
not individual scores. In the case of the year-round schools,
though, individual students’ scores have been reported incorrectly.

The problem stems from the complexities of standardized testing.
Everyone takes the same test at about the same time, but year-round
students typically have fewer school days before the test than their
counterparts on a traditional school calendar.

As a result, it’s expected the year-round students will answer
fewer questions correctly.

Ed Slawski, a senior research scientist for Harcourt, explained
that the company adjusts the national average downward for year-round
students to offset their disadvantage in number of school days. The
adjustment wasn’t made for Vista.

The Vista students’ raw scores of the number of correct answers
have been reported correctly. The problem is they were not properly
converted to national percentile ranks — how they did against test-
takers in a national sample group. Without the adjustment, the
rankings are unfairly low.

Vista assessment coordinator Katie Langford said the national
averages are based on test-takers who had attended at least 136 days
of school. Yet some Vista test-takers had only 105 days completed at
the time of the testing.

Year-round schools in Escondido and San Diego so far appear
unaffected. South Bay school district officials said they’re
continuing to analyze data.

“Why did it happen in a handful of districts? That’s puzzling,”
Slawski said. He said that he did not know whether it was a computer
error or an employee error but that Vista students should have been
compared with an adjusted average for students who had attended
roughly the same number of days of school.

The error affects only Vista’s elementary and middle schools
because its high schools run on a traditional schedule.

Harcourt is being paid $34 million to develop, score and report
results of spring testing of 4.2 million California second-through-
11th-graders. The state Board of Education will consider on Aug. 2
whether to withhold some of that money as a result of the publisher’s

Comments are closed.