The easiest way for colleges to improve their graduation rates is to serve fewer disadvantaged students, according to two new studies released this week.
The new research found that the characteristics of incoming students largely predict their likelihood of completing college. The papers arrive at a time when many politicians have been endorsing ideas  for rewarding institutions that perform well on graduation rates. That concept, however, has been questioned by many at colleges that educate non-wealthy, less-prepared students.
One paper, a policy bulletin  the Advisory Committee on Student Financial Assistance released Wednesday, found that the percentage of a college’s first-time students who are Pell Grant recipients heavily influences its graduation rates. Other strong factors are the relative wealth of institutions and the average ACT scores of their incoming students.
“The more a four-year college serves Pell recipients, the lower its six-year graduate rate and the more difficult it is to improve that rate,” found the committee, which advises Congress on federal financial aid.
For example, at public institutions where Pell recipients account for at least half of total enrollment, and which have less than $1,000 in endowment funds per student, the average six-year graduation rate is 28 percent, according to the analysis, which drew its data from the U.S. Department of Education’s Integrated Postsecondary Education Data Systems (IPEDS). 
The average graduation rate rises to 67 percent, however, at public institutions with $34,000 or more in endowment funds per student and where Pell recipients make up less than 30 percent of enrollment.
Private colleges displayed a similarly inverse relationship between graduation rates and the relative wealth of students and institutions.
The second study, a research paper  by Tim Gramling, president of Colorado Technical University’s Colorado Springs campus, found that several characteristics of incoming students at for-profit institutions largely determine graduation rates.
As president of a for-profit institution, Gramling is not a disinterested party. His study was peer-reviewed and published by SAGE Open, an open-source journal of research articles.
The study’s sample was an unidentified for-profit university that enrolled 2,500 students. It found that the grade-point average, full-time enrollment status, race, number of transfer credits and expected family contribution of those students accounted for 87 percent of the variation in the university’s graduation rate.
When G.P.A. was eliminated from the sample, the remaining four characteristics predicted the graduation rate with 74 percent accuracy.
“Graduation from college can be predicted with high accuracy based solely on student characteristics known before they begin their studies,” Gramling wrote.
The paper cites previous research on the “intergenerational legacy” of college-going, which has found that students are more likely to graduate from college if their parents did. And based on the results of his study, Gramling said “student graduation may have little to do with institutional factors.”
Both studies warned about possible unintended consequences of basing federal financial aid reforms on overly simplistic completion data.
“Using raw output measures, such as rates of graduation or student academic progress, in the award or allocation of Title IV student aid,” the advisory committee paper said, “will harm low-income students and the colleges which serve them.”
That potential problem could be particularly worrisome if new sources of performance-based aid were to be added, according to the committee. In that scenario colleges should only be required to compete against peers that serve similar mixes of students. And more research is needed to determine which colleges are actually peers.
“Output measures must be adjusted to adequately reflect differences in inputs,” according to the paper. Important factors include “college mission, student characteristics, resources and factors beyond colleges’ control.”