A Plus Colorado, an advocacy group that uses research to push for higher student achievement, has withdrawn a report that cast a dim light on the college prospects of most Colorado high school graduates. The report was based on flawed data from the Colorado Department of Higher Education that excluded the University of Colorado Boulder and Colorado State University, both well-regarded schools that are major destinations for students who stay in-state.

A Plus Colorado plans to reissue the report in a few months with correct data. Chalkbeat wrote a story based on the report, which we have removed because the new report may not support the premise of the original version.

That report had found that just 4 percent of Colorado high school graduates went on to enroll in one of the schools ranked among the nation’s top 150 universities and top 150 colleges as identified by U.S. News and World Report. The implication was that many high schools aren’t doing a great job at preparing their students for higher education.

The finding caught the eye of Superintendent Walt Cooper of Cheyenne Mountain School District in Colorado Springs because he knew his students had done much better than the report indicated. A Plus Colorado CEO Van Schoales credited Cooper for flagging the mistake.

In a statement posted to its website, A Plus Colorado apologized.

“A Plus Colorado deeply apologizes for the misinformation provided in our report and will reissue the report, A Seat at the Table: Colorado Students’ Access to Top Colleges, with the corrected matriculation data for all Colorado high schools and the 300 selective colleges and universities,” the statement said. “This is the first time that A Plus Colorado has had to reissue a report because of missing or inaccurate data. Accurate education data and analysis is the core value for A Plus Colorado.”

Beth Bean, the chief research and strategy officer for the Department of Higher Education, took responsibility for the mistake and said measures are being put in place to prevent this from happening again.

“We try to share data with external advocacy groups because the information they put out is good, and we don’t have the manpower,” she said. “Obviously we want it to be accurate.”

All sides of the education debate depend on data to make their case, and when there are mistakes in the numbers, there can be ripple effects. In 2016, Padres y Jovenes Unidos had to walk back a report that seemed to show a spike in out-of-school suspensions after years of declines. In that case, the school districts had reported bad information to the state.

Here’s what happened with A Seat at the Table, according to A Plus Colorado and Bean:

A Plus Colorado submitted a data request to the Colorado Department of Higher Education for matriculation data for a list of 300 colleges and universities for the classes of 2009 to 2015. Due to a data cross-walking error, the underlying data provided by the Colorado Department of Higher Education did not include data from the full list of these colleges and universities, including University of Colorado — Boulder and Colorado State University, leading to a significant discrepancy in the reported and actual matriculation data.

While A Plus Colorado validated the overall matriculation rate at each school, the data provided by the Colorado Department of Higher Education was aggregated across the requested list of colleges. As such, A Plus Colorado was unable to validate whether specific higher education institutions were missing from the data set. The Colorado Department of Higher Education has since improved their validation process for future data sharing and reporting.

Bean said validation processes are being developed for one-off data requests like those made by A Plus Colorado. Such measures have been in place for years for the reports the department generates for the state legislature, and Bean said this incident shouldn’t cause any doubt or question on the accuracy of reports on, for example, remedial education or concurrent enrollment.

“We stand behind having solid data,” she said. “We’ve been doing those for years, and the code has been validated. We’re now building that into our external data reports.”