About the author
Ruth Curran Neild is a research scientist at Johns Hopkins University and co-authored Unfulfilled Promise, a comprehensive 2006 study of dropping out in Philadelphia.
Email this article to:
Analysis: Dropout rates - It’s all in the coding
by Ruth Curran Neild
It was part of the “Texas Miracle:” Dropout rates in high schools in Houston and Dallas had plummeted.
But did they?
Robert Kimball, the assistant principal at Houston’s Sharpstown Senior High School, couldn’t reconcile the numbers with what he saw every day. Sharpstown had three times as many freshman as seniors during the 2001-2002 school year, but the school district reported that the high school’s dropout rate was zero.
Kimball called a local television station with his story. “60 Minutes” conducted an investigation.
It turned out that the school’s 0 percent dropout rate hinged on how the district classified – or “coded” – the students who left the school. Students who had actually dropped out were given codes indicating that they had transferred to another district or returned to their home country.
It was that simple: Change the code you give a student, and thereby change the dropout rate. That’s why the details matter so much when it comes to figuring out how many students actually finish high school and get a diploma.
Many layers in calculating rates
Calculating dropout rates starts at the school level. Let’s take an imaginary student in Philadelphia who has stopped attending school. She hasn’t shown up in more than a month and is 17 years old, which means that she can legally drop out of school in Pennsylvania.
Sometimes schools have information about what happened to the student who stopped attending. For example, if the student has enrolled in another school, the new school might request a copy of her grades or her vaccination record.
But many times, no information comes to the school. Dropouts are not known for announcing their plans to leave. They just stop coming to school.
Some schools have personnel who go out to the student’s home to find out the story; sometimes it’s just hearsay from other students.
The code that a school assigns to a student who leaves is related to how much the school investigates the situation. It is also affected by whether the district requires documentation of a transfer to another school district or is willing to accept the word of a neighbor that the family has moved out of the city.
In a case like this, the attendance clerk at the school might formally drop the student from the school rolls and code her as being “over the compulsory school age” – a code that counts her as a dropout. But if an acquaintance of the student reports – however informally – that she has moved out of the city, the attendance clerk might code that student as a “transfer” – a code that doesn’t count her as a dropout.
If the school had no good information, it is possible that no reason why the student left would be entered when the student disappears from the rolls. Under current Pennsylvania guidelines, a student whose status is coded as unknown will be counted as a dropout.
Recordkeeping not emphasized
Across the country, there has been little emphasis to schools and districts on the need to keep good records about why students are no longer coming to school. Federal and state accountability legislation has emphasized boosting test scores more than increasing graduation rates.
As the Texas situation demonstrates, there is some room for schools to hide their true dropout rate by deliberately miscoding dropouts as transfers. A recent study of the Pittsburgh dropout rate by RAND researchers John Engberg and Brian Gill created a furor on that city’s school board when the researchers noted that a portion of students coded as “transfers” had similar academic characteristics to dropouts and adjusted their dropout estimates accordingly.
The researchers who examined School District of Philadelphia data for the Unfulfilled Promise report on dropouts (see “Report offers detailed analysis of out-of-school youth”) did not note any obvious deceptive misuse of codes, although they noted that they could not verify the accuracy of how individual codes were assigned.
But now, with more national attention being devoted to high school dropouts, there are new standards of recordkeeping for districts and schools and new technologies that should help produce better estimates. Pennsylvania is starting to implement some of the best practices for determining dropout rates recommended by the National Governors Association. For example, among the group’s recommendations is increasing the state’s capacity to track student progress. Starting with this year’s kindergarten class, the Pennsylvania Department of Education (PDE) assigns each public school student a unique identification number so that districts won’t have to guess whether a student transferred or dropped out (assuming that they transferred to a public school in Pennsylvania).
However, the state still hasn’t clarified what standards of evidence schools should use in determining which students have transferred to another district, especially if they leave the state. In its current guidelines to districts, the state hedges on what it means to “know” that a student has transferred. There is still more work to be done.