Originally published in The Akron Beacon Journal

Home Page


Monday, Dec. 11, 2000



Beacon Journal staff writer
(One of a series)

Imagine a new sandlot baseball league -- youngsters from every area of Akron grouped into eight teams.

At the end of the season, look at the standings: one team sits on top, one at the bottom.

Now say to the players and their coaches, ``Who wants to switch teams?''

It would be natural that many of the better -- or at least more dedicated -- players and coaches on the teams near the basement would move up the ladder.

What happens next season? Likely the same teams appear at the top and bottom of the standings. The only change would be a wider gap between the best and the worst.

But what if instead of a baseball league, the story was about Akron Public Schools?

Substitute schools for teams, students for players, teachers for coaches.

Finally, rank the schools according to how well their students perform on the state proficiency tests.

Wouldn't many of the better students and teachers switch teams -- or schools -- if given a chance?

They have done exactly that, according to the findings of a six-month Beacon Journal examination of factors that might explain the large difference in student performance among Akron's public schools.

The gap is real.

Like the mythical baseball league, the Akron district is divided into eight clusters, each composed of a high school and its feeder middle and elementary schools.

While the clusters have roughly the same enrollments, their performance -- as measured by state proficiency tests -- is anything but equal. The percentage of students passing all the tests in the top performing cluster, Firestone, was nearly four times that of the lowest cluster, Buchtel.

The Beacon's inquiry wasn't guesswork. The goal was to find statistically verifiable data that might provide possible explanations.

The study compared the test results to data gathered from the district and Ohio Department of Education measuring a variety of educational and economic factors that could affect student performance.

Most didn't appear to affect test scores. Per-pupil spending essentially was spread equally among the schools. Differences in teacher and student attendance rates were too small to be meaningful.

Nor was race found to be a significant factor. Although the percent of African-American students in each cluster ranges widely -- from less than 15 percent in Ellet to Buchtel's 97 percent -- the best performing cluster, Firestone, is nearly 43 percent black.

Finally, three factors were identified that appeared related to test performance:

  • Poverty, measured as a percent of lower-income students receiving free and reduced-cost lunches in each cluster.

  • Low teacher experience -- the percent of classroom instructors with less than six years experience.

  • Student flight -- the percent of students living in each cluster who left to attend schools in another cluster through the district's ``open enrollment'' program.

    The three accompanying graphs illustrate the relationships: As poverty, teacher inexperience and student flight increase, test scores decrease.

    To test whether the correlations were accidental, the Beacon Journal used a statistical technique called regression analysis, which compares two series of numbers to determine how well one series could predict -- or explain -- the differences between the numbers in the second series.

    The analysis found that poverty and low teacher experience could explain more than half the difference in the test results among 48 Akron elementary and middle schools. (Two schools, Stewart Primary and Miller South, were not included because they are magnet schools that draw from all parts of the area; high schools weren't included because of the tendency of older teen-agers not to apply for free and reduced-cost lunches out of embarrassment.)

    The probability of the relationship being coincidence is less than 1 in 1,000.

    A limitation of regression analysis is that, while it shows correlations, it doesn't prove cause-and-effect. Teacher inexperience could contribute to low test scores, but low teacher experience also could be an effect of low scores. Like the coaches in the story of the sandlot league, veteran teachers may want to go to better performing schools.

    Likewise, students may choose to leave their home schools in search of a better education. But it's also possible for student flight to contribute to poor test scores if the students leaving a cluster are more likely to do better on the exams than the classmates they leave behind.

    To investigate that possibility, another statistical methodology, called ``geocoding,'' was used to roughly gauge the performance potential of students leaving and staying in each school cluster by determining the household income of their neighborhoods. Numerous studies have shown a high correlation between the family income of students and how well they do on tests.

    To obtain the neighborhood household income data, the Beacon Journal first obtained from the school district the addresses of more than 25,000 students -- about 84 percent of all students enrolled in Akron's public schools. The only addresses missing were those of students whose parents signed forms asking that such information be withheld.

    A computer was used to match the addresses to U.S. Census block groups, for which estimates of median household income are available. In Akron, an average block group contains about 350 households.

    The average neighborhood incomes of students staying and leaving each cluster were compared. The result indicated that student flight was contributing to keeping test scores down in the district's two poorest-performing clusters: Students leaving Buchtel and Central-Hower to enroll elsewhere had higher neighborhood incomes than those who stayed.

    © Copyright 2000 The Akron Beacon Journal

    Home Page