By Charlie Brennan, Education News Colorado
Colorado’s two largest school districts are near opposite ends of the spectrum in a report comparing “expected” high school graduation rates and actual rates in the nation’s 50 largest school districts.
Jefferson County had the fifth-highest graduation rate of the 50 districts (77.7 percent) in 2008, 8.5 points higher than the 69.2 percent expected rate. The state’s largest district with nearly 86,000 students, Jeffco stretches from poorer neighborhoods along Denver’s western border to well-to-do foothills enclaves.
Denver Public Schools ranked 48th with a graduation rate of 43.5 percent, 9.3 points below its expected rate of 52.8 percent. The district, whose boundaries are the same as the city’s, has enrollment of more than 78,000 students, Colorado’s second largest.
Jeffco Superintendent Cindy Stevenson was cautious about her district’s results. “As educators, you should always be careful the way you apply formulas. You do it one way, you look good, and you can have another way … that you don’t look as good.”
DPS Superintendent Tom Boasberg said, “I think it’s certainly fair to say our graduation rate should be and needs to be significantly higher.” He cited DPS reforms undertaken since 2008 and fresher statistics to argue that the district is improving.
“Diplomas Count,” the annual high school graduation study from Education Week includes a ranking of the nation’s 50 largest school districts, reporting their 2008 graduation rates and their expected rates, the latter determined by a formula based on 10 factors. That analysis was done by the Editorial Projects in Education Research Center.
The study, released last week, raises interesting discussion points about demographics, the pace of reform in urban districts and the value of such projections. (Story on the full study and links.)
Do urban and suburban make the difference?
Projecting graduation rates
The EPE study uses 10 factors in projecting expected district graduation rates.
- Average high school size
- Student-teacher ration
- Location (urban or other)
- Percentage minority enrollment
- Racial segregation
- Poverty level
- Economic segregation
- Per-pupil expenditures
- Instructional spending
Christopher Swanson of EPE said while the factors “are all kind of on equal footing … We do know some of these factors have a more direct effect on graduation than others, and they are the socio-economic level, socio-economic segregation, poverty level, racial composition and racial segregation. … Size matters, but it doesn’t matter as much as poverty and racial isolation.”
Christopher Swanson, vice president of Editorial Projects in Education and author of the graduation rates report, noted that Denver and other districts that fell short of expectations “are all, almost exclusively, inner-city urban districts; they are all kind of under-performing to various degrees.”
He mentioned, Detroit, which ranked 50th among big districts in its graduation rate and fell 13.7 percent short of expectations. Philadelphia, ranked 47th, showed a 7.5 percent shortfall. One notch higher, at 46th, was Los Angeles, with a 6.0 percent shortfall.
“When you get up higher in these rankings, you tend to see these big-county, wider districts; they may be diverse, and they tend to be more affluent than the inner city districts,” said Swanson.
But urban/suburban differences doesn’t seem to tell the whole story.
Chicago exceeded its expected graduation rate by 23.1 percent – the highest positive differential of any district on the EPE list. New York also did well, surpassing expectations by 7.7 percent.
Russell Rumberger, a University of California at Santa Barbara professor of education who studies graduation rates, said there are reasons that help explain New York and Chicago’s performance.
“New York has had Gates (Foundation) money to revive the whole system and add schools, and in Chicago there are similar things happening,” Rumberger said. “That shows that some of these interventions are actually paying off in raising grad rates.”
Paul Teske, dean of the School of Public Policy at the University of Colorado Denver, has also looked at the study. He agreed that New York and Chicago have seen the impact of reform efforts underway for well more than a decade.
“It’s disturbing to see Denver” falling short, he said. “If you compare Denver to other like-sized cities, such as Milwaukee, Baltimore, Houston, Albuquerque – Denver’s negative is bigger than all but Detroit, and we know Detroit is pretty much a disaster in a lot of ways. It’s not good news for Denver.”
Analyzing the study
Peter Fritz, principal consultant in the office of Dropout Prevention and Student Engagement for the Colorado Department of Education, also reviewed the report.
While he didn’t dispute the findings, he said, “It can be an art form to figure out which factors you should rely on and how much weight each factor should be given. But it can be a pretty powerful tool to tell whether districts are meeting expectations or are doing the most with what they have.”
Fritz did find the study’s failure to include mobility as a factor as significant. Mobility, as defined by the CDE, includes a student changing grades, moving from one school to another or from one district to another during a school year.
In 2007-2008, Fritz said, CDE calculated the Colorado average for mobility to be 25.5 percent. In that same year, Jefferson County showed a mobility rate of 19.3 percent, while Denver’s was 30.5 percent. Fritz suggested Denver would have fared better had its high mobility been factored in to the projected graduation rate.
But Rumberger defended the omission of mobility.
“I’ve done some research, and at least in some instances, mobility is, in fact, created by the districts themselves,” Rumberger said, through forced relocations of students to resolve discipline issues, or for other reasons.
“It’s difficult to tell, sometimes, the degree to which a mobility is district-caused and the degree to which it is not,” Rumberger added. “And for a model like this, you want to control only for those things the district has no control over.”
Teske had his own questions concerning the study’s methodology, suggesting, for instance, that urban and suburban districts have such contrasting characteristics that they should perhaps be analyzed separately.
He also said individual factors in such a study need to be viewed carefully. The EPE study used poverty as a factor, as indicated by the percentage of students eligible for free and reduced lunch.
There is poverty, and then there is poverty, Teske said. “There really is a lot of difference between kids from a household that’s under $10,000 income, really poor, with one parent, no parent, or non-working parents, versus a kid who’s maybe got a single parent who is making $30,000 as a bus driver, or a fast-food worker, with a reasonably stable household, or apartment they live in year-round.”
Superintendents evaluate their districts
Boasberg said, “I do think it’s fair to say our graduation rate should be higher, and that’s exactly the purpose of the reforms under the Denver Plan and the changes in schools like West and Montbello.”
In addition to the dramatic turnaround plan approved for Montbello High School area last November, the district is now considering a proposal to phase out underperforming West High School and introduce two new grade 6-12 academies at its central Denver location.
“The need to increase rigor and increase the graduation rate is the driving factor behind the reforms,” Boasberg said. “We clearly recognize the need for very significant improvement in our high schools, and so that’s the driving reason for the reforms.”
Boasberg also pointed to statistics more current than those on which the EPE report is based.
“We would note the quite significant growth in the number of our graduates – more than 15 percent in the two years since the time the study examined,” said Boasberg. “In 2010, we had a 12-percent increase in the number of graduates, and we expect a further increase this year.”
He also noted that last year DPS registered a 5.4 percent jump in its on-time graduation rate over the previous year – 51.8 percent versus 46.4 percent in 2009.
Boasberg also said that the report does not include the more recent graduation rates of turnaround schools such as “Bruce Randolph, MLK, and this year, Manual; all three schools, following reforms, have graduations rates according to CDE of over 85 percent. It’s an incredible example of how the reforms are driving significantly higher graduation.”
Stevenson, despite how well her district fared in the study, said comparisons of graduation rates, particularly across the nation, are tricky.
“I think it’s nice, but I’m not ready to say happy times are here and all our problems are solved,” she said. “I have mixed feelings.”
Still, Stevenson said there are reasons for Jefferson County’s performance.
One, she said, is the district’s use of predictive analytics. There are key performance points in a student’s life at which negative results are warnings that a student is in danger of not graduating later.
Three that Stevenson highlighted are inability to read near grade level in third grade, unsatisfactory performance on math CSAPs in fifth and sixth grades, and two failing grades in ninth grade.
“Those are variables that you can use to predict – and then change those variables before they get” to higher grades.
“Juniors and seniors isn’t where you start. I believe you start in kindergarten,” Stevenson said.
What’s the value of projections?
Swanson of EPE said, “We haven’t engaged in any systematic analysis or investigation to explain why some districts over-or-under-perform relative to expectations. One of our hopes is that folks who are closer to the ground will take the expectations index – and other information – as a starting point for understanding and explaining what’s happening.”
“None of these measures is perfect, I guess,” said Teske. “I don’t want to suggest it isn’t a credible exercise. Even with the urban-suburban mix, the fact that Denver is more negative on the expected (graduation rate) than Baltimore is kind of surprising and worrisome.”
Fritz of CDE said, “I think it’s a good way to open a dialogue.”
And Rumberger noted, “It’s not perfect, but no statistical modeling is perfect.”