One of the reasons that ???data quality??? can be so elusive is that bad data can just sit there, in a drawer in someone's desk, never getting discovered. It's an old axiom: data that never see the light of day tend not to be as reliable as data that are viewed and used. With more visibility comes more accountability for data quality.
Why this dry discourse on data? Well, it turns out that last week's release of America's Private Public Schools has brought to light some isolated examples of bad education data in circulation. Take South Dakota, where we found that 16 percent of the schools qualify as ???private public schools??? (four times the national average) because they serve virtually no poor students. Except???that conclusion was based on inaccurate information in the federal Common Core of Data, thanks to faulty reporting by the Mount Rushmore State. As Josh Verges of the Sioux Falls Argus-Leader, pointed out, ???the true number [of ???private public schools???] is less than one-tenth what the report says.??? Verges writes:
The apparent cause of the bad data is the school district's conversion in early 2008 to a new student information system, Infinite Campus, which is used by most of the schools in the state. Mary Stadick-Smith, spokeswoman for the Department of Education, said state officials thought when they retrieved the Sioux Falls data that it had been fully converted. It had not, so what the state reported to the NCES was inaccurate."Basically, the information they gave us was incorrect," NCES statistician Patrick Keaton said.
Researchers use the Common Core of Data all the time; big errors like South Dakota's are surely the exception, not the rule. Still, on the bright side, we hope that by using these data, and ensuring that they get viewed by lots of eyeballs, we are increasing the odds that errors will get caught and corrected. We also hope we're helping to pressure education officials to clean up their data. Because, after all, if states can't accurately report how many low-income students a school serves, why on Earth do we believe that they can get more complicated issues (like student achievement gains or teacher effectiveness measures) right?
--Mike Petrilli
