PBA Home >
Institutional Research & Analysis >
UG Graduation & Retention Rates >
At AAU Public Universities
Freshmen Graduation Rates at AAU Public Universities
Comparisons Using IPEDS Public-Release Data
Lou McClelland, University of Colorado at Boulder, December 2005
Six-year graduation rates for first-time full-time freshmen entering in 1998 are reported in the IPEDS Data Feedback Report,
Figure 7. IPEDS-reported graduation rates for the 34 US AAU publics
range from under 60% at Minnesota, Arizona, Kansas, and Stony Brook to over 85% at Michigan, Berkeley, UCLA, and Virginia.
The 1998-cohort graduation rates are closely related to freshman academic qualifications as measured by SAT and ACT scores
for students entering in 2001, the first year IPEDS collected test scores. For example, SAT totals (calculated as the average
of the 25th and 75th percentile scores) at schools with rates under 60% are 1100-1150, while SAT totals for the schools with
rates over 85% are 1270 to 1320. Differences in 2001 entering-freshmen test scores account for over 60% of the variance over
schools in the 1998 cohort graduation rates.
The average SAT total can be used to predict a school’s graduation rate. The predicted rate is 1.3 percentage points higher
for each 10-point increase in SAT total. With an average entering-freshman SAT total averaging 1148, CU- Boulder’s predicted
six-year graduation rate is 68%, range 55 to 80%. Our actual rate of 66% is slightly below 68% but squarely in the predicted
Several factors may account for remaining differences in graduation rates. These include
- Differences between test scores of students entering in 1998 and in 2001
- Entering student high school performance and courses taken
- Entering student demographics, especially financial capabilities, first generation college status,
and minority status
- Availability of financial aid
- The proportion of students who are from out of state – at CU-Boulder graduation rates for in-state and
out-of-state freshmen entering in 1998 are 69% and 63%, respectively, and our proportion of out-of-state students is relatively high.
- Proportions of students who are part-time and who live on campus
- Proportions of students who considered the school their first choice
- Student support services, academic advising and expectations, and campus history and traditions.
Some of these factors can be measured or estimated with additional IPEDS data or with supplemental data collected by
the AAU data exchange. Further work will explore these.
The Data Feedback Report also lists six-year graduation rates for 1998 freshmen by race/ethnicity.
Graduation rates are highest for white freshmen. On average for the 34 AAU publics, rates are lower for freshmen who
are Hispanic/Latino (11-12 percentage points lower), African American (15 points), Asian American (2 points), and
American Indian (17 points). IPEDS does not collect test scores for individual racial/ethnic groups.
Graduation rates for freshmen of color are generally higher at schools with higher rates for whites. In addition,
even controlling for the white graduation rate, rates for each other racial/ethnic group are related to 25th percentile
SAT/ACT math and verbal scores. Differences across schools in the white graduation rate, math 25th percentile, and verbal
25th percentile account for 73-82% of the variance across schools in Hispanic/Latino, African American, and Asian American
graduation rates. Numbers of American Indian students are too low to allow similar analyses.
For CU-Boulder, the predicted and actual graduation rates are as follows
- Hispanic/Latino: Predicted 55%, actual 58%
- African American: Predicted 52%, actual 54%
- Asian American: Predicted 63%, actual 63%.
All predictions are plus/minus about 10 percentage points. In all cases
our actual graduation rates for students of color are well within the predicted ranges but are at or above
the point prediction based on our rates for whites and test scores. This reflects well on our programs for
students of color.
Reference: IPEDS (U.S. Department of Education Integrated Postsecondary Data System) data sources
Reference: Combining ACT and SAT scores
From the Graduation Rate Survey (GRS): Percentage of first-time full-time freshmen entering in 1998 and graduating from
the institution within six years; includes graduation in the sixth summer. The rates reflect graduation at the university
entered only – students transferring elsewhere then graduating are not counted.
From the Institutional Characteristics survey: ACT and SAT scores (25th and 75th percentiles) for freshmen entering in
fall 2001, and numbers submitting each type of score.
Data Feedback Report issued 2005: Graduation rates are
in Figure 7. Test scores are not shown.
- Schools could report both ACT and SAT scores, only SAT, or only ACT. The number of students with each type of test
was also reported.
- ACT scores were converted to SAT by means of the following formulas:
- SAT total = 40.46 * ACT composite + 135.8
- SAT math = 20.48 * ACT math + 22.1
- SAT verbal = 18.56 * ACT English + 135.0
- The formulas are derivations (by regression) from ACT to SAT concordances published as follows
- Composite and math: Tables 3 and 6, respectively, of Concordance Between SAT I and ACT Scores for Individual Students,
College Board Research Notes>,
RN-07, June 1999,
- Verbal/English, not available from the College Board publications:
Table 5 of Developing a Concordance Between the ACT
Assessment and the SAT I: Reasoning Test for The University of Texas at Austin, Gary Lavergne and Bruce Walker, July 2001,
- Reported ACT 25th and 75th percentile scores were converted to the SAT metric using the formulas listed above. Then for each school a single
score for total, verbal, and math, for the 25th and 75th percentiles, was calculated as a weighted average of the SAT-from-ACT and raw SAT scores,
with the weights a function of the number of students reported for each test type.
- The overall average for the total only was then calculated as the average of the 25th and 75th percentiles.
- Example, CU-Boulder
- Reported ACT math 22, 28 (25th and 75th percentiles)
- Reported SAT math 540, 640
- Reported N’s ACT 3316, SAT 3950. These translate to weights of SAT .54, ACT .46. Note that the 7266 reported total
N is far over the reported total N enrolled of around 5000 because many students submitted both SAT and ACT scores.
- SAT math from ACT, by formula: 473, 596
- Weighted average single math score: 509, 620
- Grad rates: GRS with Data Cutting Tool, collection year 2004
- Test scores: Institutional Characteristics with DCT, collection year 2001
- ACT/SAT: L:/all/how/ACTSAT.xls
- L:/ir/tracking/gradcomp, gradrt05.sas and dctfile_12-2-2005 (test scores)
- Outputs: L:/ir/tracking/grad/comp/GradRt05_051205.txt