At his press conference last week, the schools chancellor Joel Klein tried to explain how public school students had really improved this year in English Language Arts.

He used split bar graphs that presented the ELA test results for all of the students tested and, at the same time, showed how much better we do when 55,000 students with Limited English Proficiency are not counted in the mix. He might have added that the rate of inflation isn’t so bad if you don’t include the cost of food and fuel.

This two-track approach allowed the chancellor to paint a better-looking picture of the system’s performance in 2007 only based on the scores of students whom he called “English proficient” — a misnomer, since 44% of this exclusive group fail to reach the test’s threshold of proficiency.

The graphics were designed to compensate for a change in this year’s test population, in which far fewer LEP students, aka English Language Learners, were exempted from testing. Being required to test 31,000 more of them in English lowered overall citywide performance.

To compare apples with apples, the chancellor presented his alternate version of the data — extracting the ELL students from the 2007 totals. This yielded better-looking results and seems fair, until we see what he did in 2005.

That year, Mayor Bloomberg asked voters to judge how well he had improved the schools. He had taken control of them during his first term. His May 2005 press release hailed a nearly 10-point increase in the ELA scores by fourth-graders. Mr. Klein cited “tremendous gains” by minority students as evidence that their leadership was paying off for those who previously had been left behind.

Their joint news release announced that scores for Hispanic and Black fourthgraders shot up by 11.5% and 9.7% compared to 6.5% and 6.4% for White and Asian pupils.

But the Department of Education didn’t disclose enough information for the public and press to examine these facts. It turned out that two changes were made that removed thousands of low-performing students from the test population. This significantly inflated the electionyear results. Nowhere was attention called to the shift.

Sadly, minority students did not make huge gains. It was illogical to take credit for this feat and deceptive to suggest the gap was closing.

In capitalizing upon the city’s 9.9-point overall “improvement,” the mayor failed to report that 4.2 points were due to taking 5,182 low scorers out of the equation. Instead, a positively skewed impression was created. Much of 2005’s publicized record-setting increases were a phantom — a case of addition by subtraction.

In third grade, 3,012 Level 1 students were held back as the chancellor’s new promotion policy was implemented. They were retained because of scores that marked them as having “serious deficiencies.”

Some 2,170 more ELL test exemptions were granted in 2005 than 2004. I filed a Freedom of Information request and waited nine months for the department’s division of accountability to provide the details. This allowed the holdovers and exemptions to be analyzed by ethnicity to determine what impact their exclusion had in 2005.

Among the retained third-graders, 1,706 were Black and 1,099 Hispanic — making up 93% of the students denied promotion. All were Level 1. Had they advanced to fourth grade, they would have lowered their sub-group’s scores. Of the language-exempt students, 1,674, or 77%, were Spanish-speaking. I assumed that due to limited fluency in English, virtually all would have failed to reach Level 3, the ELA test’s proficiency cutoff point.

Recalculating the 2005 results with these children returned to their respective ethnic cohorts appreciably diminishes the celebrated strides made by Hispanic and Black students. Instead of the 11.5 point and 9.7 point increases ascribed to them, their one-year gains would have been a more modest 6.2 points and 6.1 points. In contrast, there were few White, 82, and Asian, 90, student holdovers. Including them reduces their gains less than a point to 5.9 points and 5.5 points.

While the ethnicity of 496 non-Hispanic exempt students could not be determined from the data, my reconstruction of the test population holds that the differences among the four groups would have been minimal.

By the way, last week’s press release made no mention of the achievement gap. But one graph, not referenced in the release, shows the remaining chasm — fewer than 50% of minority students are proficient in English, while 75% of White and Asian children combined have reached this plateau. This is in stark contrast to the 2005 press release, which blared misleading news about closing the gap.

Mr. Smith, a retired Board of Education administrative analyst, is writing a book about New York’s testing program called “Failing the Test.”