Friday, August 23, 2013

How to Prevent Creaming in Higher Education Performance Regimes?

One of the most prominent concerns raised about President Obama's proposed performance-based funding plan for higher education is that it could reduce access by encouraging creaming.  In other words, what's to stop colleges and universities from simply raising the bars for entry, tightening their admissions policies, in order to improve graduation rates and lower default rates?

Good question.

I'd like to make a few points and then open this up for discussion.  It's one of the big areas that needs bright minds thinking hard in search of solutions, and I hope you'll jump in with good ideas.  We're going to have to look far and wide for solutions, as we can expect that folks in education probably don't have all the answers.

1. The problem already exists.  The number of colleges raising their admissions requirements over time tells this story.  So let's not pretend like we're creating a new problem. The question is whether we're making it worse.

2. NCLB approached this challenge through the use of value-added modeling.  It didn't work there and it's probably not going to work here either, especially since it's hard to believe that we can possible account for all inputs that are external to college, in order to focus on gains made by the college itself.  Now, I know many people will disagree with me on this, including my former student Robert Kelchen, so be sure to read up on their work on the topic.

3. A weaker version of value-added modeling is risk-adjusted metrics, a regression based approach to accounting for intial student differences when looking at outcomes. I'm not sure this is going to fly either, and the Left doesn't like it since it seems to perpetuate the idea that we should "expect" students from disadvantaged families to do worse in college.  No one actually wants that, and so we try things NCLB-style,  demanding growth in graduation rates for subgroups of students. But that too doesn't prevent colleges from admitting fewer students from a given subgroup.

4. Prohibition of creaming via more metrics.  Let's say we stipulate terms regarding enrollment and admissions, in addition to outcomes.  These may have to be differentiated according to college type.  For example, in order to receive Title IV aid, a college must :
  • Enroll at least 100 students who are Pell-eligible (all colleges) and
  • Maintain a % Pell that meets or exceeds the state average among high school graduates (all public colleges and universities) 
  • Admit at least 50% of Pell applicants (all private colleges and universities)
  • OR: Use a lottery process for admission into at least 50% of the entering class
Yes, this means that very small colleges would have to be very diverse in order to participate in Title IV. It also means that all colleges and universities may have to adjust their admissions standards somewhat and change their recruiting practices. Is that a bad thing?  If they don't like it, they simply need to prove their mettle by using a lottery for admission.  Then we can really get a handle on their value-added!

Let's talk this one through further.

5. Prohibition and regulation.  Schools could be selected for an audit based on troubling trends in their admissions data.  If they were found guilty of creaming, they could be put on probation and monitored for a period of time.  If they failed that, they'd be kicked out of Title IV.

We must also ask, how much creaming might be tolerable? If the use of performance standards forced more colleges to help low-income students graduate, while reducing access for some other students, at what point would this become intolerable?

Ok, enough from me-- what are your great ideas?