Saturday, October 10, 2009

Crossing the Finish Line at HBCUs

A. Overall Review 
I write this note to call my readers' attention to one of the most important books about higher education that I have encountered in the last twenty years. I am referring to a recent publication called Crossing the Finish Line by William G. Bowen, Matthew M. Chingos, and Michael S. McPherson (Princeton University Press, 2009).

I sincerely believe that this book should be read by every president, provost, and chancellor at every college and university in the country. However, I direct my strongest recommendations to the senior administrators at every HBCU that has a two year or a four year undergraduate degree program.

Their book summarizes the most important results of the authors' exhaustive examinations of the factors that are most closely related to the graduation rates of students at 68 public colleges, or to be more specific, at 21 "flagship universities" and at all of the public universities in Maryland, North Carolina, Ohio, and Virginia. Their inclusion of all public institutions in these four states enabled them to to include 11 of the best known public HBCUs in their analysis.

The authors' findings are derived from a sophisticated series of regression analyses of the massive databases that they constructed from the individual records of hundreds of thousands of students in the 68 public institutions in their sample. Although the civil liberties "angel" in me is appalled by their access to the private records of so many students, the educator in me screams: "Shut up! Look at what they learned from the data! Consider how much we can improve our programs by using their findings!"

In my opinion, this important book can be understood by any intelligent reader for two reasons:
  • First, the authors are very good writers; and

  • Second, the book only contains the authors' logic, their most important findings, and some superbly constructed bar charts and line graphs that illustrate their findings. They banish the underlying technical discussions of data collection, sampling, variable construction, handling of missing variables, etc, etc, etc to a 150 page appendix published on a Website ...
Given the massive size of its underlying databases, the rigor of its analysis, and the clarity of its presentations, this book is the fine product of master craftsmen who are showing the rest of us how the grand scale, social science research game is supposed to be played. Although the data underlying their findings only includes student records from public institutions, it seems likely that their findings will be highly relevant to private colleges and universities. The next three sections of this blog present my comments about three of their findings; the last two sections address related issues not covered by their research.

B. Transfer Students From Two-Year Colleges Are Good Investments
The authors found that transfer students from two year colleges are more likely to graduate than freshmen having similar high school GPAs and SAT/ACT scores (p. 142). They offer a plausible explanation of this pattern in their data:
"We believe that these superior graduation rates among transfers (after adjusting for differences in observable credentials and background characteristics) reflect strong selection effects. That is, students who come to four-year institutions from two-year colleges have already successfully managed the transition from high school to one kind of college experience. We strongly suspect that their subsequent success at four-year institutions, compared with the outcomes of first-time freshmen, reflects differences in aspirations, maturity, social capital, and coping skills (including a demonstrated ability to 'stay the course'). The two-year colleges, in short, are a 'sorting mechanism' that works to the benefit of the four-year institutions to which their students transfer." (p. 143)
The data shows that the transfer students in the study's massive sample did better than similar groups of entering freshmen. Therefore the authors' large scale research provides significant support for the recent efforts by a number of four-year HBCUs to establish special relationships with two-year colleges that facilitate the transfer of the two-year students to the four-year HBCUs upon graduation, e.g., Alabama A&M, Fort Valley State, South Carolina State, Cheney, North Carolina A&T, six HBCUs in South Carolina, and Prairie View.

C. Grade Point Averages (GPAs) Are Usually Better Predictors than SAT/ACT Test Scores

Paraphrasing the authors' findings for emphasis: If the test scores for an entering class of freshmen were increased by one standard deviation, the six-year graduation rate for this class would increase by less than 2 percent at all of the universities in the sample, e.g. from 70 percent to less than 72 percent. But if the high school grade point averages of an entering class increased by one standard deviation, the six-year graduation rate for that class would increase by 10 percent at the non-elite universities, e.g., from 70 percent to 80 percent, and by 6 percent at the elite/selective universities, e.g., from 84 percent to 90 percent.

The authors also found that it didn't matter what high schools the students attended. For students from the same high schools, their grades were far better predictors of graduation rates than their test scores.
"The parallel, and even more revealing, finding is that high school GPA is very positively and very consistently associated with six-year graduation rates whatever the level of the high school that the student attended. From the perspective of our interest in predicting the probability that a student will earn a bachelor's degree, the conclusion is straight-forward, with modest qualifications: 'a grade is a grade is a grade.' Students with very good high school grades who attended not-very-strong high schools nonetheless graduated in large numbers from whatever university they attended. On the other hand, students with relatively weak academic records in high school have much lower graduation rates than their higher-achieving high school classmates -- again, whatever the academic level of the high school they attended." (p. 122; italics in authors' text)
The authors interpret their findings as follows:
"Our interpretation of this entire set of findings is a simple one: High school grades are such a powerful predictor of graduation rates in part because they reveal mastery of course content. But the 'in part' formulation is critically important. In our view, high school grades reveal much more than mastery of content. They reveal qualities of motivation and perseverance -- as well as the presence of good study habits and time management skills -- that tell us a great deal about the chances that a student will complete a college program." (p. 124; italics in author's text)
D. Elite Black Students Should Reach for the Stars
Chapter 5 -- " High Schools and 'Undermatching' " -- presents some of the book's most disturbing findings. Whereas conservatives have often lamented that affirmative action would tend to place Black students in colleges that were above their capabilities, i.e., to "overmatch" them, the authors' data demonstrates the opposite phenomenon. The best Black students in their massive sample tended to "undermatch", i.e, to enroll in colleges that were below their capabilities. In the authors' own words:
"Within this highly qualified group of seniors, undermatches appear to have been more common among black students (especially black women) than among white students -- in part because a number of black students undermatched to HBCUs." (p 103)
The authors used the extensive data they obtained from North Carolina to determine that most of the highly qualified Black students never applied to the flagships at Chapel Hill and North Carolina State. Perhaps their most surprising finding was that highly qualified students who undermatched, i.e., chose to attend less challenging colleges, did not do as well as highly qualified students who enrolled in the flagship institutions. The authors pick up this theme again in Chapter 11:
"In other groupings by high school GPA, the pattern is essentially the same -- black male students who went to more selective institutions graduated at higher, not lower, rates than black students in the same GPA interval who went to less selective institutions. Moreover, contrary to what the overmatch or mismatch hypothesis would lead us to expect, the relative graduation rate advantage associated with going to a more selective university was even more pronounced for black men at the lower end of the high school grade distribution than it was for students with better high school records. ... There is certainly no evidence that black men were 'harmed' by going to the more selective universities that chose to admit them. In fact, the evidence available strongly suggests that students in general, including black students, are generally advised to enroll at the most challenging university that will accept them." (p. 109, 110; italics in authors' text)

E. But What About Non-Elite Black Students?
A few months ago, a high level manager of the United Negro College Fund (UNCF) alerted me to a research effort currently in progress that is finding that the HBCUs in North Carolina are doing more than twice as well as other state schools with regards to average Black students, i.e., the non-elite students who do not qualify for entry into the state's flagship universities. If this finding holds up at the end of the study, it will provide an important complement/correction to the book's assertion that non-elite Black students would be better off at elite institutions -- if they were accepted.

Hopefully, the research cited by the UNCF manager will also provide some data-driven insights as to why these particular HBCUs have been so effective in educating average Black students. And could their innovations be adopted by the non-HBCUs in North Carolina and elsewhere? Such findings would support our claims that HBCUs are not only historically important institutions; they deserve America's continued support for years to come because of the significant contributions that they continue to make to American higher education.  

(Note: North Carolina has three kinds of colleges: flagships, HBCUs, and other colleges that are neither flagships nor HBCUs. Bowen and associates lumped HBCUs together with the "others"; and perhaps the research cited by the UNCF manager combined flagships with "others". So it's possible that average Black students might be better off at flagships than at "other" colleges, but might not do as well at flagships or "other" colleges as they would at HBCUs.)

F. Institutional Research

When I summarized the book's findings for a colleague, he expressed skepticism because he felt that the advantages of GPAs over test scores as well as some of its other findings would have been discovered by various colleges and universities through periodic examinations of their own data -- if these findings were valid. Exploiting the findings from their self-studies would have given the innovators a competitive advantage over other institutions.

These kinds of self-studies are usually conducted by "Institutional Research" units (or units having similar titles). But when I used the Gateway's comprehensive, Google-based search tool (found on the Gateway's Home Page that enables users to search all HBCU Websites simultaneously, I only found 32 Websites that listed Institutional Research units. Most had small staffs; more specifically, they either listed no staff or only listed one or two staff members; and few of the listed staffers had advanced degrees. Therefore I think it is safe to infer that, at this time, the vast majority of HBCUs do not have the kind of staff that would enable them to apply sophisticated, statistical, computer-based analytical techniques to their student records to obtain the kind of findings reported by Dr. Bowen and his associates. (Note:Alabama A& M University is a notable exception --; Xavier University is another --; and FAMU is a third --

The authors' findings are based on student records from a broad sample of colleges and universities. Although their sample included 11 HBCUs in four states, the validity of these findings cannot automatically be assumed for HBCUs in other states. Indeed, they should also be confirmed by independent analysis of the student records for HBCUs in the four states included in the book's research. These considerations lead me to issue the following challenge to the few HBCUs that have IR units with sufficiently skilled staff:

  • Please analyze your student records so as to address the most important questions considered by Dr. Bowen and his associates (if you haven't already done so); then share your findings with the entire HBCU community and with interested colleagues at other colleges and universities by publishing them on your Websites.
Related notes: