This is a guest post from Benjamin Feldman, research, evaluation and accountability officer for the Baltimore school system, in response to the graduation rate report released today by America's Promise:
T.S. Eliot called April the cruelest month, blaming lilacs as the source of misery. From the perspective of the Baltimore City Schools’ Division of Research, Evaluation, Assessment and Accountability, the real April misery is the annual reappearance of graduation rates derived from a tortured statistic called the “Cumulative Promotion Index,” or CPI. According to the CPI, in 2005 Baltimore had crawled from the dungeon of graduation rates, adding 7.7 percent points to achieve a 45.5% rate. Thank heavens for Indianapolis: they’re at 30.5%, even worse than Cleveland at 34.4%. The national average is reported as a pathetic 70.6%.
Since EdWeek first published the CPI, I have made something of a career talking about graduation rates: how they are calculated, what the politics are, who is to blame. I even gave an evening’s lecture at the School of Public Policy at UMBC two years ago, so keen was the interest in this perennial topic. Now that The Sun publishes a continuing blog on education issues, the time is ripe to offer the community a fuller insight into this most-important of all K-12 education outcome measures.
What’s a graduate?
For most readers of this blog, this question seems simplistic. We remember the late spring day when in rented cap and gown, we waited to hear our name called, shook the principal’s hand, and got our rolled up diploma. Equally memorable for me is a lunch-and-learn discussion I had 20 years ago with my first boss in Research and Evaluation who posed the question, “What’s a graduate?” The answer then was a student (through the year in which he or she might turn 22 years old) who had accrued a certain number of credits by earning minimum grades, including a particular distribution of credits. Some grades, like honors, needed special weights. In addition, the student had to pass a battery of Maryland Functional Tests in reading, mathematics, citizenship, and writing. (The writing test was the killer.) Moreover, the student had to discharge all local requirements such as turning in his or her library books or paying outstanding fees. The principal and/or the counselor had to sign off on the graduation. Lastly, the various City Schools databases of record had to agree with the above information, and completed rosters had to be manually confirmed by the school.
Since then, the more challenging Maryland High School Assessments have superseded the old Maryland Functional Tests. These can be passed straight out, satisfied with a combined score of 1602, or satisfied with passing grades or quality independent Bridge projects, senior theses or extended independent research projects. The required 22 credits now include three in mathematics, four in English/language arts, three in social studies, three in science (including biology and two others with laboratory experiences in earth, life or physical science) one in technology education, one in fine arts plus a combination of foreign languages, advanced technology, electives or career and technology education. And a half credit in gym and a half credit in health; plus 75 hours in service learning. On top of all of this, the library books still have to come back, the fees settled, the principal endorse, the databases confirm everything, and the student not be older than 22. The Maryland State Department of Education (MSDE) is our partner in all tracking these pieces.
What’s a graduation rate?
As noted above, the determination of the individual graduate is more complicated than the question initially implies. How then do school systems, states, and the nation get to a graduation rate in which we have consensus and confidence?
The most accurate method would be to implant a data chip into every child at matriculation to his or her first school. Then from huge CIA-type agencies we could continuously scan the environment to ascertain the whereabouts of these students. We would determine the first time the student puts a foot into ninth grade whether in Baltimore or in Guam and then test the graduation status (according to the above or comparable rules) four years later. Or we could test this at the time the student turns 22. Absent such a Big Brother solution, districts and states must struggle to estimate the performance of a cohort of students.
The reader will recall that a cohort is any group that shares some salient characteristic at the moment the group is defined and labeled. For the purposes of calculating a graduation rate, the characteristic of choice is the school year in which the student becomes a first time ninth grader. To make this real to Maryland readers, this year, high school Adequate Yearly Progress will be predicated on the status of 12th graders who were first-time ninth graders during 2004-05. If a ninth grader in 2004-05 was taking a second shot at freshman year, he or she is not part of this cohort. If a student was in the eighth grade in 2004-05 and skipped a grade, joining sophomores in 2005-06, he or she would be part of the cohort.
If only this were so simple! What happens with the student who shows up in our schools in the 10th grade from out of state? In City Schools, if the student or families do not give us the facts, we make the best determination we can from whatever information is available. A national student ID number would be great. It doesn’t exist. Moreover, MSDE is having a devil of a time establishing a unique state ID number. This is a discussion for another blog on another day.
If the world were static, a graduation rate would have graduates in the numerator and original ninth graders in the denominator. As the world is not static, MSDE has used a four-year “leaver rate” to estimate the cohort:
Grad rate=N graduates/9th dropouts+10th dropouts+11th dropouts+12th dropouts+12th grads
The denominator provides a cohort estimate, spanning four years. There is no credit given for students needing more than eight semesters and no control for migration across district boundaries if the coding is not absolutely clean. Old data from closed schools returns and changes the rate after the system has closed its books. Moreover, MSDE has in the past looked at each year’s demographic files as an entity unto itself. Thus, a student who leaves, comes back the next year, and leaves again, would be counted in the denominator twice. It is possible for a student to be in the denominator as a dropout AND as a graduate. MSDE works closely with us to understand this phenomenon caused by a highly mobile population. To get a more accurate view, we have to work with all four years as a single event, not as four separate events. And even if the state unique ID were to work perfectly, this might help Baltimore some, but it won’t do a thing to help Prince George's County, which has movement between itself and the District of Columbia and itself and Fairfax, Va. What everyone seems to agree is that each student should have a single disposition: as a graduate or as a dropout. Present data systems do not get us to such clarity. Moreover, the highly fragile or mobile family might not thoughtfully withdraw the student from one school district and carefully identify the student for a new school district. These cases become “magic children,” as it were, multiplying their presence across many databases, and skewing rates… always to the disadvantage of urban districts.
The Cumulative Promotion Index (CPI)
The CPI does not attempt to estimate a cohort. It attempts to estimate a process. It assumes a completely static world where students go nowhere. It assumes that school districts neither grow nor shrink. It does not attempt to look across four years, but rather takes a pair of years as the proxy for the entire high school experience. The new CPI is this:
Not only does this completely counterintuitive statistic paint the bleakest possible picture for the nation’s school districts, it also contradicts the calculation of every single state department of education in the country.
What’s coming: a four-year on-time graduation rate
The four-year on-time graduation rate is the consensus model selected by the National Governor’s Association:
Grad rate=year 4 diplomas/1st time 9th grade cohort + transfers in – transfers out
Transfers in and out are “on grade level,” that is, they become part of the original cohort regardless of the district of origin. Retained and lost students remain in the cohort. Early graduates are credited. Summer graduates are NOT credited, but are reported in year "4+1," and thus appear to be lost to the cohort and of no benefit to the next year’s statistic. Some states plan to allow credit for English language learners and special education students with specific plans calling for more than eight semesters to graduate.
Sara Neufeld questions why a graduation rate has to be nailed to a four-year metric. It is certainly true that students and families would like to finish high school in eight semesters, and it is equally true that a high school education is be delivered most economically to taxpayers if students pass through with conveyor belt speed. In reality, it doesn’t work like this. Neufeld speculates that as many as 20% of City Schools students need a fifth year. Our first look at last year’s numbers indicated that at least 10% of graduates had needed more than four years, and of these, 2-3% needed as many as six years.
MSDE is sensitive to this concern, and the state plans a variety of metrics to communicate a more thoughful multi-dimensional reality to stakeholders. MSDE may report four-year graduates and also graduates who took five years or more. MSDE works with us to recognize the important work we do to bring back students and give them another push toward a diploma or a GED. Under a punitive model like the CPI or an uninformed leaver rate, every time a district makes an effort to rehabilitate a lost student, it risks creating an inflated denominator and blasting the graduation rate. Should school districts be driven by the rate or should they do what is in the best interests of individuals?
We are gratified that the new America’s Promise Alliance report credits City Schools with a dramatic gain (looking back four years as it does, before our really exciting data movements in recent years), and we are glad to be out of the cellar. We agree that having a high school diploma is life-determining: without it a person’s financial horizon is bleak in the extreme. Readers of this blog know that the diploma is insufficient in the 21st century competitive global environment. The United States needs college graduates, students with professional preparation, students who are fluent in multiple languages, masters of information technologies and mathematics. We need critical readers, critical thinkers, and lucid writers.
If educators and policymakers want to have a meaningful discussion of graduation as the mission-critical outcome of K-12 education, we need to have a much better command of whom and where the cohort is. If our society believes that a competent graduate is the most important product we make as a nation, we ought to put the magnitude of resources into teaching and learning that we put into professional sports and popular entertainment.