News, Views and Careers for All of Higher Education
Nov. 10
The Collegiate Learning Assessment may be passing the largest test to date of whether it can measure growth in student learning. But the study of the CLA also found that many minority students and those who are not well prepared for college show smaller gains on the CLA — potentially reinforcing the concerns some have about how the test may be used.
The CLA has emerged in the last year or two as a key response for colleges facing demands that they demonstrate the “learning outcomes” and “value added” that take place on their campuses. Students who take the CLA are asked to complete a series of exercises to measure critical thinking, analytic thinking and written communication. The test is offered to small, representative groups of students as freshmen and to other groups later in their college careers, in an attempt to measure growth in learning.
The theory behind CLA and similar assessment tools is that colleges need to get away from measuring their excellence only by “input” measures (students’ incoming SAT scores, for example) or prestige and pay more attention to what actually takes place during college. So the CLA might find that Harvard University students have great skills upon arrival, but don’t grow much, while students at other colleges see much more learning. When the Voluntary System of Accountability was announced a year ago by two groups of public universities, the CLA was designated as one of the tests that could be used to measure student learning in a comparable way.
That of course begs the question of whether the CLA can measure growth in student learning, and the new research released Saturday suggests that it can. The study was based on tracking 2,300 students at 24 four-year colleges and universities, which were not named but included a broad range of institutions by standards of mission, competitiveness and demographics. The analysis was conducted by the Social Science Research Council. While the Council for Aid to Education, which runs the CLA, cooperated with the project, the council had no control over the study or release of its findings. For the students tracked, CLA scores and transcripts were analyzed at the beginning of the freshman year and at the end of the sophomore year. Additional studies are now planned as the students are tracked through the rest of their college careers and, perhaps, beyond.
Here are some of the findings to date:
Richard Arun, a professor of sociology and education at New York University and program director for education at the Social Science Research Council, said that the study was significant for “moving basic social science research, where you can look at value added, from K-12 education to higher education.” Although this approach has become common in elementary and secondary schools, it is “overdue” in higher education and this research suggests that it can be done.
Similarly, Roger Benjamin, president of the Council for Aid to Education, said that this research helps “to push this new testing paradigm, which gets us beyond the multiple choice test and speaks to actual cognitive outcomes.”
Some educators have worried that the CLA and similar tests would end up — like traditional measures of educational excellence — saying that flagship universities or elite liberal colleges do a better job than institutions that admit and work with students who have not been well prepared. The new research could well add to such fears as it finds the greatest gains at institutions with a well prepared student body in a traditional curriculum.
“That reality does give me concern,” said Benjamin. But he added that the CLA also demonstrated that colleges do not perform equally well at reaching minority students or students without a solid high school education. Institutions that serve such students benefit “if we really identify and describe the obstacles” and then focus on why some colleges perform better, he said.
The purpose of CLA is “to understand how a school understands where it stacks up, so that then they can improve skills in the classroom.” The idea isn’t to reduce colleges and their work to a number, he said.
Of course the concern of colleges and some testing critics is that however sophisticated an analysis the CLA’s creators envision, many politicians will look for a number, and may not credit the college making arduous but important gains with disadvantaged students.
Robert Schaeffer, public education director of the National Center for Fair and Open Testing, said that his group’s concerns about “value added” tests like the CLA “are less with the quality of those instruments than in how some proponents want the results to be used. Any attempt to impose one-size-fits-none measures on colleges and universities is sure to create even more problems than No Child Left Behind did in K-12 education.”
Want it on paper? Print this page.
Know someone who’d be interested? Forward this story.
Want to stay informed? Sign up for free daily news e-mail.
Advertisement
The inherent problem with CLA is simple. Learning occurs at the student level, not institutional. Asking a handful of students to take a test, then calculating an institution-level score borders on meaningless. The within-institution variation (as pointed out in the other article) is far greater than between institution variance.
Charles, at 8:15 am EST on November 10, 2008
Even if one could get by the question of whether the computer code for the “Make an Argument/Break an Argument” section of the CLA picks up all the major fallacies in argumentation and delivers reliable scoring in the mid-ranges of performance (a classic psychometric problem with unrestricted response testing), two major questions remain about our assumptions, fascination, and use of the CLA and similar tests.
First, whether the small samples of volunteers taking the exams the populations and institutions at issue. The study reported here opens up a door into this question in reporting results by major field. That instantly raises the question of whether the 100 volunteer test-takers at institution X reflect the distribution of majors at institution X. The same type of question can be raised about the representativeness of the test-taking sample with respect to age and family status, enrollment intensity, and other standard demographics. For a large institution, e.g. Arizona State, to claim representativeness of 100 or even 200 test-takers would require statistical gymnastics of Olympic caliber.
The second open question—allied to the first but more serious in light of institutional uses of such tests—-is the extent to which an institution can claim the results are wholly due to the student’s educational experience at that institution.For the most obvious case, roughly 60 percent of our undergraduates attend more than one school, some—particularly transfer students—-for significant portions of their undergraduate experience. Assuming transfer students are proportionately represented in an institution’s sample of volunteer test-takers, how much of the observed gains can an institution claim are by-products of its work? One could add life experiences—-jobs with intellectual content, prior military experience, extra-curricular activities pursued with intensity, lovers from other language and cultural backgrounds, and on and on along the paths of becoming an adult—-as contributing in different degrees to communication, argumentation, and organizational skills that have nothing to do with the student’s experience at the institution. Yet in our passion to put a number up on the board—-e.g. a 1.21 standard deviation improvement in performance, whatever that means—-we brush all these contributions under the rug, and claim everything for ourselves. It’s a very misleading arrogance.
Now, if we think the CLA—or other similar tests—truly measure what we do in higher education, then instead of playing value-added games with dubious assumptions, set passing scores at each institution (there are standard ways of doing this, and each institution could do so on its own), and require everybody to exceed the threshold as a condition of receiving a bachelor’s degree. Give them 3 shots to pass. Then we would have a reference point for what we think our degrees mean. I am sure that approach would scare the wits out of higher education, but it certainly would bring us face-to-face with the challenge of producing publicly-understandable benchmarksfor academic content and performance.
Cliff Adelman, Senior Associate at Institute for Higher Education Policy, at 8:50 am EST on November 10, 2008
These so-called “measures of learning” were concocted as a response to building pressures on institutions to demonstrate that student learning occurs — and now, voila!, “learning” conveniently appears.
Furthermore, are we to assume that NO learning takes place at community colleges (including AP/dual enrollment) and adult vo-tech programs, where college credit is earned but not tested for? Doesn’t this missing data also emphasize the institutional constructedness of these measures?
Glen S. McGhee, Dir., at Florida Higher Education Accountability Project, at 8:55 am EST on November 10, 2008
I don’t think many of us doubt that the CLA, when done in a truly longitudinal fashion, can measure gains in student learning. The biggest problem with the CLA as it is actually used/administered by the large majority of institutions is its dependence on using a cross-sectional design (freshmen and seniors, different groups of students, tested to measure gains) to estimate gain scores. Add to this the problems most of us have with recruiting students and motivating them to take this fairly long test, and it becomes pretty useless as a measure of how much is learned. Unless individual institutions find a way to do longitudinal testing of motivated, representative, students, the CLA will not be a useful instrument — even if it is one of those used by the VSA.
gerryd, at 9:40 am EST on November 10, 2008
I agree with the previous comments about the flaws of the CLA, but I would like to describe a more fundamental problem: none of us, including the CLA publishers, have bothered to define what it is the CLA tests. I’ve been collecting definitions of critical thinking, and so far have upwards of 50. I imagine there are dozens of others, and they differ from each other substantially.
What definition of critical thinking does the CLA use? I believe the developers won’t say, because it is proprietary information that would reveal to competitors how the test is constructed and scored. How can a quality as abstract and socially constructed as “critical thinking” be tested in any meaningful way if it is not defined?
Critical thinking is defined differently in different contexts. What passes for good critical thinking in an engineering department at Boeing, for example, may be quite different from what passes as good critical thinking at the Smithsonian, or the Pentagon, or the New York Times. Would the professors at College X score student writing on a critical-thinking the same way they do at College Y? Would Professor X and Professor Y, who share office space, score it the same way? Probably not, unless they have been engaged in norming exercises similar to the process of norming the scorers of standardized tests.
The same argument can be made for other higher-level cognitive skills. Although they are among the most important things we teach, we lack commonly accepted definitions that provide the foundation of good assessments. If in doubt, ask a few colleagues to grade a student essay for critical thinking. There are likely to be as many scores as scorers.
The answer to this dilemma, I believe, is to engage faculty members in a kind of “assessment community” where the scoring of actual samples of student work and subsequent discussion of rationales can develop commonly-understood defintions that are grounded in real-life examples. Gradually, such local communities can link to each other through shared websites to develop an assessment network among institutions with similar missions and student populations.
Without such commonly understood definitions, test results are meaningless, both for faculty who wish to improve teaching and curriculum, and for policymakers and parents who wish to compare institutions.
Lee Griffin, at 10:46 am EST on November 10, 2008
I agree with Lee. “Critical thinking” shows up in mission statements all over the place (including the goals of the AAC&U LEAP initiative), and the CLA wants to measure it. Without a defining context, it’s far too fuzzy. I discuss a more useful alternative (which we use at Coker College) at http://highered.blogspot.com/2008...-initiative-and-thinking-skills.html
David Eubanks, Director of Planning, Assessment, and Information Services. at Coker College, at 2:26 pm EST on December 18, 2008
Advertisement
or search for jobs directly.
Texas Woman’s University — Denton Campus Department: College of Arts and Sciences Title: Assistant Dean Job Code: 9AS01 Date ... see job
Unique opportunity with top Health and Science Medical Center in the Southwest see job
Walden University is focused exclusively on providing a superior university experience for adult professionals through the ... see job
The nation’s first university, Penn is a world-renowned leader in education, research, and innovation. Situated on a ... see job
The University of Minnesota is a premier employer and a talent magnet attracting leading faculty and staff from around the ... see job
The University of Minnesota is a premier employer and a talent magnet attracting leading faculty and staff from around the ... see job
The University of Minnesota is a premier employer and a talent magnet attracting leading faculty and staff from around the ... see job
The University of Minnesota is a premier employer and a talent magnet attracting leading faculty and staff from around the ... see job
Position Summary: The administrative assistant is responsible for and not limited to the following tasks: — ... see job
Madison Area Technical College (MATC) is the technical and community college for the greater Madison area. It is dedicated to ... see job
Did we read the same report???
“That of course begs the question of whether the CLA can measure growth in student learning, and the new research released Saturday suggests that it can.”
I read through the linked document, and there is nothing which documents the fundamental integrity of CLA as an assessment. At least in the PDF referred to in the file, there is a blithe assumption that CLA measures learning, and then a bunch of tables associating CLA measures with various socioeconomic indicators. Reads like a circular argument to me: “See here — we use this measure to make associations between the measure and common social-science variables. That proves that the measure is meaningful!” No, it doesn’t.
Sherman Dorn, Professor at University of South Florida, at 7:05 am EST on November 10, 2008