Less than 60 percent of students at our four-year colleges complete their studies and graduate. That depressing statistic has drawn many critics, and now it has occasioned a book, Crossing the Finish Line, by three well-connected members of the academic establishment–William Bowen, Matthew Chingos, and Michael McPherson (hereafter, BCM). The authors obtained some data on individual student performance unavailable to most researchers, uncovered some interesting facts and made some worthwhile observations. The question is, how much light has this effort produced.
BCM base their conclusions largely on a sample of data from a few dozen public universities. Since they are writing about public education, it would have been useful if at least one of the authors had some experience, either as a student or professor, in such an institution at some point in their eight decades of accumulative academic lifetime. Unfortunately, none studied or taught at a public university, meaning that they are not likely fully conversant with the cultural differences between, say, the Ivy League vs. the California State university system. Interestingly, with one exception, even all the announced authors’ book receptions are being held at such private establishment enclaves as Harvard or the Brookings Institution.
At the book’s beginning, BCM talk about the importance of higher college graduation rates to America’s continued economic and educational leadership. We learn that increasing graduation rates is critical in achieving the objective of much larger rates of higher education attainment. BCM give a lot of attention to what they term the “undermatching” of high quality students with lower quality institutions. According to them, in North Carolina alone in 1999 there were about 2,500 undermatched high school seniors. These undermatched students went to lower quality schools than their capabilities suggested were feasible (or to no school at all), and tended to have graduation rates that were perhaps 15 points lower than similarly qualified students going to top-flight public schools (adjusting for other characteristics such as socioeconomic status, the gap was typically closer to 10 points). Assuming the North Carolina numbers are representative of the nation, what these results suggest is that ending undermatching could raise the national graduation rate by something less than one percentage point (e.g.., from about 55 to perhaps 56 percent), a relatively trivial movement.
Moreover, if, in fact, all the undermatched had applied to better schools and most matriculated there, selective admissions schools like the University of North Carolina at Chapel Hill almost certainly would have turned away other deserving students who currently attend given rigid limits on entering class size. Therefore, the net graduation rate effect of undermatching is probably even less, well below one percentage point. But even this overstates the undermatching “problem.” BCM reveal that the undermatched students almost certainly ranked far higher (20 or more percentage points) in their graduating classes at the second tier schools they attended than they would have ranked if they attended the more selective flagships. Given the prospect of likely graduating ranked fairly highly, say in the 33th percentile at the University or North Carolina at Charlotte, or below average (say, in the 53th percentile) at UNC Chapel Hill, it is not obvious to me that the “right” choice is going to Chapel Hill. Rather than identifying a problem that seriously lowers America’s aggregate graduation rate, BCM may have largely merely quantified the extent to which good students avoid top reputation schools for altogether rational reasons.
BCM put extraordinary emphasis on the lower graduation rates (and other indicators of academic performance) of students from lower income homes and from minority groups. They note often that the graduation rate of one of those groups, say Hispanics, is 15 percentage points or so below that of whites or the national average. They lament that only perhaps one-third of that gap can be explained by such factors as high school grades or scores on the SAT exam—there is a 10 point unexplained gap that, if filled, would bring greater equality of opportunity and higher national graduation rates. Non-academic preparation/performance factors peculiar to these groups lower their success in graduating from American universities.
Yet there are serious problems of both emphasis and analysis. Numerous public universities have overall graduation rates of 40 percent or lower, compared with over 80 percent at many good schools. It seems to me that explaining the 40 or 50 point gap between schools perceived to be at the top and those well below average in terms of graduating students is a far more important question than the unexplained 10 point gap regarding minorities and poor, if the authors are genuinely interested in overall graduation rates, America’s relative international decline in higher education attainment, etc.
The interschool gap, adjusting for differences in admissions selectivity, is the subject of a superb short study for the American Enterprise Institute (AEI) by Rick Hess, Mark Schneider, Kevin Carey and Andrew Kelly. Both Sul Ross State University (Texas) and Shippensburg University of Pennsylvania are master’s level “less competitive” (using Barron’s classification) state schools, but Sul Ross has a 19 percent graduation rate and Shippensburg a 65 percent rate. Analysis of these differences might help explain why the overall U.S. graduation rate is around 55 percent rather than, say, 70 or 75 percent. From my analysis of the AEI study, if schools in the bottom one-third of average graduation rates within each selectivity category closed just one-half of the gap in their rate relative to the top one-third of schools in that same category, the national graduation rate would have increased by about 12 percentage points, compared with perhaps a 3 or 4 point rise in that rate if we erased a 10 percentage point graduation gap associated with low SES/race/gender issues. BCM’s obsession with race/class/gender obscures broader and quantitatively more important issues keeping Americans from finishing college.
The authors largely (except for black men) ignore lower graduation rates of males relative to females. They almost completely ignore adult learners. They ignore the near 10 percent of the college population going to for-profit schools -institutions that BCM did not even deign to mention, despite the fact they are growing rapidly in relative importance and disproportionately serve the low socioeconomic/minority population so important to them.
Moreover, I even question the conclusions about the unexplained performance gap with respect to minorities -they could be right, but I am unconvinced based on the presented evidence. They are reached on the basis of regression analysis assessing the impact of performance measures (grades, test scores). However, a 3.5 grade point average at Phillips Exeter Academy, or New Trier High School in suburban Chicago probably connotes a superior level of academic performance to a 3.5 average at most public high schools in, say, Chicago or New York City. The authors do not directly control for differences in family situations. Do children from single family homes fare worse than those from traditional two parent families? Do children coming from families heavily dependent on public assistance do worse than those from families dependent on work income? Almost certainly, the answer to these questions is yes, but this is not directly considered by BCM. In other words, the data, while more detailed than usually available, are not so good that I am convinced that environmental influences on academic performance e are fully accounted for, meaning the unexplained performance gap truly directly related to minority status could well be meaningfully smaller than indicated.
The authors note that private schools have higher graduation rates, and much higher four-year graduation rates, than public ones, even controlling for other characteristics. They speculate a bit about this, but do nothing to really analyze the question. Indeed, if there is an inherent advantage to private schools in terms of increasing educational attainment via higher graduation rates, one might legitimately ask: should governments reduce, not increase, their involvement in subsidizing colleges and universities? BCM do not raise that question.
Then there is the question of the sample of schools used. The authors looked at 21 mostly prestigious flagship universities, including such top schools as Cal Berkeley and UCLA, and the universities of Virginia, Michigan, North Carolina, Illinois , Wisconsin, Texas (Austin)-” public ivies.” They have 28 non-flagship institutions in the sample (one of which I teach at), along with some historically black institutions. The sample is not only non-random, but skewed towards the more selective, more elite, of the publics -there are more sampled students from the flagships than other state schools that I suspect enroll far more students nationally. In California, Berkeley and UCLA are represented, but none of the nation’s largest university system, the California State University schools. And geographically, it is a bit skewed as well – double digit numbers of schools are included from both Ohio and North Carolina but only two from much more populous California, for example. It seems to me if you were truly concerned about overall dropout rates, you would put a good deal of attention on the schools and regions where dropouts are the greatest -including states like Texas, which has only one school (with high graduation rates) in the survey.
My most serious concern is that the authors largely accept the current way universities operate as a given. For example, they do not pay much attention to the serious decline in emphasis in teaching (and related activities) at many state universities. They do not talk about the decline in the instructional share of university budgets, the rise of vast university bureaucracies, or much about the possible tradeoffs between research emphasis and undergraduate performance. They do not emphasize that incentive systems lead faculty to shun undergraduate students in order to work with graduate students in completing research that is costly to produce and often of trivial importance to society. They fail to document the shocking decline in student academic effort over time, a decline that roughly parallels grade inflation. It is at least plausible that some or all of these developments impact significantly on college graduation rates.
Also, BCM quote other academics or President Obama in asserting a need for greater higher education attainment, when the evidence (as opposed to rhetoric) shows that the major mismatching problem today may be that college graduates increasingly are way over-trained for the jobs that they take. We are increasingly in the era of the college educated restaurant server, tree removal worker, or mail carrier. The proportion of new jobs truly needing traditional college educations is smaller than the proportion of the population we are educating. Our problem is not under-investment, but over-investment in traditional universities.
At the end of the book, the authors approvingly quote Kevin Carey, who perceptively notes that reduced emphasis on undergraduates may lead universities to ignore students who are floundering. If the authors had picked up and empirically examined that insight -exploring the academic advising function for example, they likely would have made a far better contribution to the question over how increases in graduation rates can be made at a reasonable cost.
In short, to me this is another book with all the academic trappings to demonstrate gravitas – lots of graphs, copious footnotes and regression equations (actually, found in an appendix available only online, an extreme annoyance) — but which in reality presents a highly skewed picture in order to advance the authors’ normative position.