Sep 15, 2014

How bad data affects applicants or why you can’t trust scores reported by test-optional colleges



Along with data collected by the federal government via IPEDS (Integrated Postsecondary Education Data), information compiled by the Common Data Set (CDS) pretty much forms the basis for metrics used to power the college ranking industry.  

These same numbers and statistics populate any number of guides, websites, and search engines sold to college-bound students and their families who have come to rely on them for making decisions related to “fit” and likelihood of admission.

But sometimes numbers lie.  In absence of reliable systems to prevent fraud, individual data points are subject to interpretation and can be manipulated by college administrators anxious to improve their standing among peers.  

And this is true as much for data collected by the feds as it is for the CDS.  The difference is that the folks overseeing the CDS use the information to make money—lots of money.

Yet for years, the brain trust behind the CDS—the College Board, Petersons, and US News—has steadfastly resisted calls for reform in the way they do business.

Robert Morse, the US News rankings guru, insists that if colleges and universities are willing to lie to the federal government, they’ll lie to any organization attempting to rank or otherwise describe an institution based on information they freely provide.

And in the face of a series of scandals involving deliberate misreporting, Brian Kelly, also of US News, insists to Boston Magazine that “Ninety-nine point nine percent of the schools are treating this seriously and reporting with integrity.” 

But this is a point debated by college administrators surveyed by Inside Higher Ed, 90 percent of whom believe “other” institutions falsify data to make themselves look better in the eyes of the public.

Kelly also told Boston Magazine, “It’s not up to us to solve problems. We’re just putting data out there.”

It’s a little like a green grocer who refuses to take responsibility for selling bad fruit.  The grocer packages the fruit, displays it, and sells it at significant profit.  Yet when anyone complains about the quality of the fruit, the grocer suggests they take it up with the growers.

And in this case the fruit can be questionable if not outright rotten.

Take for example the question of how test-optional colleges report standardized test scores.  This has been a recent issue for the CDS “listserve,” through which “crowd sourced” technical assistance is provided to college-based staff responsible for completing CDS forms.

Question C9 on the CDS asks colleges to report a range of standardized test scores—ACT and SAT.  For test-optional colleges, the question arises as to which scores should be reported, particularly in cases where students submit scores but request they not be used for making an admissions decision.

When asked whether all scores or a subset of scores should be reported, helpful colleagues (in absence of a more formal system of technical assistance) responded in a number of different ways.

The first response came from a relatively new test-optional college, “…we only report on test scores used in the admission decision.”

The next response came from a college that is not test-optional but insisted “…the directions do not say submit scores only used in the admissions decisions.  It says use all scores submitted.”

Another college asked if colleges were submitting ALL scores received for each student or only the highest of each test, as the CDS does not suggest or recommend superscoring scores submitted.

After more debate about policies in use at different schools, Robert Morse, whose official title is chief data strategist at US News, finally intervened with his interpretation which was that schools should be submitting scores for all enrolled students who submitted scores with no differentiation made as to whether scores are used in admission or not.

This begged the question of what constitutes “submitted.”  Does it mean that the scores were simply provided on the application or those officially coming from the testing agency?  And if a student asks that scores not be considered for admissions, are the scores appearing on the form actually “submitted”?

The entire debate finally came down to the fact that the CDS has not chosen, for whatever reason, to make definitions crystal clear and/or to communicate exactly what scores are to be included—those used for admissions decisions or all scores submitted.

And more importantly, the CDS has failed to provide an explicit directive to ensure that schools do not exclude the “submitted” scores of certain groups of applicants—legacy, international, athlete, AND test-optional.

To the outside world, this debate may seem like counting the number of angels who can dance on the head of a pin.  But for applicants and those who advise them, the standardized test score ranges reported by a college can have an effect on determining “fit” for a particular college.  

These scores also have a very real impact on a college’s US News rank and may explain why some schools only want to report scores students ask to have used for admissions, as they would likely be higher.  And of course, colleges want to superscore their reports providing numbers that only reflect on the highest scores submitted by individual applicants.

But this is just one of many data points subject to interpretation.   It’s only when the “interpretation” strays too far from the acceptable that scandals are revealed and administrators lose their jobs.

With all the money made by the College Board, Wintergreen Orchard House, US News, and Petersons, surely some could be set aside for the development of clear guidelines and the provision of annual training or more consistent technical assistance for college-based staff.  

At a minimum, college administrators should be asked to “certify” the accuracy of the data provided.  Sanctions for falsified data should be clear.  And while more invasive (and not likely to happen in this lifetime), random data audits or spot checks should be conducted every year.

Because at the end of the day, organizations profiting from the data need to take some responsibility for its accuracy.

This is Part 2 of a two-part series. Part 1 may be found here.

No comments:

Post a Comment