NFES 2009-805July 2009

Metadata: Solving the Case of the Inaccurate Dropout Count, Chapter 2

Continued from Exhibit 1.5.

The school superintendent, Dr. Sanders, and Lincoln High School's principal, Mr. Howell, met on Thursday after school to study the dropout data. Despite Mr. Howell's assurance that more of his students were staying enrolled in school, the numbers hadn't changed:

Still enrolled: 4 percent 35 percent 61 percent

Dr. Sanders quickly realized that they were in over their heads. "I can read the numbers in the report as well as anyone, but I have no way of knowing whether they are right. What report did you send to central office for us to generate this dropout rate?" Mr. Howell was embarrassed to admit he didn't know the answer off the top of his head. "To be honest with you, I have no idea... My administrative assistant, Ms. Johnson, fills out the reports." Dr. Sanders became irritated—not at Mr. Howell, but at the way the district was handling its data reporting. "You know, Mr. Howell, these data are very important. We use them to generate our attendance counts, dropout and graduation rates, and state funding estimates. And yet, you and I don't know how the numbers got on that form. We've got to change the way we handle data around here. In the meantime, let's get our data guru, Mr. Olsen, on the job."

A few minutes later, Mr. Olsen was in the superintendent's office, explaining that the data would have been entered on the year-end enrollment reports submitted by each school in the district. Dr. Sanders and Mr. Howell nodded their heads, as it all sounded vaguely familiar. Dr. Sanders asked the obvious question, "Is there any chance that we calculated the rates incorrectly, Mr. Olsen?" "Well, yes, that's always possible," Mr. Olsen responded. "But it's not likely. We verify our processes quite thoroughly. Without knowing more, I think that the numbers are either correct or the error is a result of something else." "Well," Mr. Howell interjected, "I don't think the numbers are correct, so we need to figure out what the 'something else' might be." "That's right," Dr. Sanders added. "We need to determine what is causing the error, and how we can be sure that it isn't happening in other aspects of our data reporting."

Continued in Exhibit 3.4.