At a glance
Directed self-placement lets college students pick their own entry-level courses. A review of the evidence on completion, equity and the hidden cost of advising.
Placement into a first college course was, for most of the twentieth century, something done to students. A score arrived, a cut-off was applied, and a schedule followed. Directed self-placement (DSP) reverses that order, and it has been reversing it for longer than most policy summaries admit. Writing studies scholars set out the idea in the late 1990s, arguing that students who are told plainly what each course demands will choose more wisely than a placement examination chooses on their behalf. The practice has since travelled from first-year composition into mathematics, into engineering prerequisites, and into statewide advising systems in the United States. Its names have multiplied with its ambitions: informed self-placement (ISP), guided self-placement, and, in some administrative usage, direct enrollment.
The mechanics are simple enough to describe in a paragraph. The college publishes course descriptions, prerequisite expectations and realistic workloads. The incoming student reads them, answers a short questionnaire about prior coursework and confidence, and registers at the level they judge appropriate. Sometimes a brief reading and writing task sits inside that questionnaire. Sometimes a counselor or adviser reviews the decision before the schedule locks. What never happens, in a pure version, is a test score deciding the question.
That simplicity conceals a difficult empirical problem, and three decades of research have not settled it. The question is not whether students should be treated as adults. It is whether a college that hands over the decision has also handed over the information, the time and the advising that make the decision a good one — and the published evidence suggests the answer varies enormously from campus to campus.
A Spectrum, Not a Single Policy
Colleges use the same vocabulary for practices that share almost nothing. The Center for the Analysis of Postsecondary Readiness (CAPR), a research unit at Teachers College, Columbia University, plots placement methods along a guidance-choice continuum with four positions. Traditional placement offers low choice and low guidance. Self-placement offers high choice and low guidance. Informed self-placement offers both. Informed placement offers guidance while keeping the decision with the institution.
Most of the disagreement in the literature comes from researchers comparing institutions that sit in different quadrants while using the same label. A college that requires a reflective essay and a ninety-minute advising conversation is doing something different from a college that posts a course list and a link.
The reform has spread unevenly. As recently as 2016, one estimate put the share of open-access two-year colleges that had formally experimented with offering students the choice at roughly one percent. The published record is also overwhelmingly American, and much of it concerns two-year colleges and first-year writing. That narrow base should temper any confident claim about what the model would do in a system where placement is settled centrally by national examination — a structure common in Europe and much of Asia, and one that directs self-placement has barely touched.
The Strongest Case for Handing the Decision Over
The case against placement tests rests on something firmer than fashion. Standardized placement exams are weak predictors of how students perform in college courses, and the sequences they feed — one, two, sometimes three remedial terms before a student earns a single credit — are where large numbers of students stop enrolling altogether. Under-placement is not a neutral error. A student who could have passed a gateway course but was routed to a prerequisite loses a term, pays tuition for it, and often leaves.
Tiffany Morton's 2022 review for CAPR, synthesising the studies then available, found that colleges using informed self-placement generally enrol fewer students in developmental courses and more in gateway courses. Performance outcomes were either unchanged or better, even when larger numbers of students started at the higher level. Students and faculty reported positive experiences.
That last finding deserves a caveat the review itself implies. Students reported confidence in their choices regardless of how those choices turned out. Satisfaction is not the same as accuracy, and a policy defended on the grounds that students feel good about it is defended weakly. The completion data carry more weight, and the completion data are, in the main, encouraging without being decisive. Morton's conclusion was blunt: because no large-scale rigorous evaluation existed, the impact of self-placement on student outcomes could not be known.
Photo by Brett Jordan on Unsplash
What the Validity Studies Found
The most searching test of the model came from the University of Michigan, which introduced directed self-placement in 1999 in place of a portfolio system and used it for a decade. Anne Ruggles Gere, Laura Aull, Timothy Green and Anne Porter examined records from 2003 to 2008 and concluded that the arrangement did not have strong validity. Students' answers on the self-placement questions did not correlate well with other measures of their achievement, and the students with the greatest need for developmental work were the ones whose choices served them least well.
It does not follow, however, that the Michigan study discredits the model. Its authors were explicit that validity is always local — bound to a particular instrument, a particular student body, a particular curriculum — and they offered their framework precisely so that other institutions could test their own versions rather than inherit a verdict. What the study does establish is that a questionnaire can look like guidance while functioning as a formality.
Aull's later work sharpened the point about instrument design. In a study of a selective private university in the southern United States, published in Assessing Writing, she found that adding a reading and writing task to the self-placement process changed how students rated their own proficiency — and that the differences were statistically significant only for certain groups of students. The same instrument was measuring different things in different populations.
The Equity Finding That Will Not Go Away
Holly Kosiewicz and Federick Ngo exploited an accident: a community college mistakenly allowed students to place themselves into or out of developmental mathematics. More students enrolled directly in gateway and transfer-level courses, which is what advocates predicted. But greater shares of women, of Black students, and of Hispanic students chose the lowest levels of mathematics compared with their test-placed peers — and the positive outcomes of self-placement accrued mostly to White, Asian and male students. The study appeared in the American Educational Research Journal in 2020, and its authors reached for stereotype vulnerability and behavioural decision theory to explain the pattern.
The finding is single-institution and single-subject, and it should be read with that caution. It still poses the question the model's supporters have to answer. When choice is introduced into a system where some students have been told, repeatedly and implicitly, that mathematics is not for them, choice can reproduce the very sorting that reform was meant to undo. Later work by Purnell and Burdman found that first-generation students and those with low confidence in mathematics made less optimal self-placement decisions, which points at the same mechanism from a different angle.
It does not follow that the test was fairer. Under-placement by examination was also patterned by race and income, and the remedial sequences it produced had dismal completion rates. The realistic comparison is between two imperfect instruments, one of which at least has the virtue of being visible.
Advising Bears the Cost, and Nobody Budgets for It
The clearest account of what self-placement demands comes from Florida, which required its college system to adopt placement based on multiple measures of student achievement under a 2013 statute. Amanda Nix and Shouping Hu interviewed 275 advisers across 19 institutions about implementation between 2014 and 2019, and their case study of the Florida College System records three recurring problems: heavier adviser workloads, insufficient information reaching students, and student resistance to the recommendation on offer.
Colleges responded with group advising sessions, advanced preparation for advisers, and self-diagnostic tools students could use before registering. Those are not trivial administrative adjustments. They are the difference between informed self-placement and a course catalogue with a shrug.
Nix and Hu also note a wrinkle that should interest anyone writing policy. Their review of the literature found four variants in practice: full student choice, limited choice, what researchers call veiled choice — where a choice exists formally but is steered — and informed self-placement with no choice at all. Earlier work on the Florida reform found colleges varied in how faithfully they followed the letter and spirit of the statute, with some quietly circumventing it.
Photo by Bernd 📷 Dittrich on Unsplash
Discipline Matters More Than the Model
Writing and mathematics are not the same placement problem, and the divergence is instructive. A 2024 doctoral study at the University of Arizona by Angela Garcia compared students placed by directed self-placement with those placed by traditional means at one community college. Students who chose their own level were more likely to complete their required college-level mathematics within two semesters. The exception was students majoring in science, technology, engineering or mathematics, for whom traditional placement appeared to hold an edge.
An engineering case points the same way. A study of 39 sections of an introductory circuit analysis course taught between 2013 and 2018 found that sections using a self-placement policy outperformed those that did not. Yet prior grade point average remained the strongest predictor of success in both — which means the self-placement sections may have been doing something other than self-placement, whether better teaching, a different mix of students, or a lighter prerequisite chain.
Jason Godfrey's regression discontinuity analysis, posted to arXiv in late 2025, tested the causal question directly using more than 20,000 placement records in first-year writing. Directed self-placement did not damage grades or pass rates. But class, race and gender differences persisted across the programme, which suggests placement technique is one lever among many and a modest one.
California offers the largest natural laboratory, though its reforms are adjacent rather than identical. Assembly Bill 705, enacted in 2017, required colleges to maximise the chance that students start and finish transfer-level English and mathematics within a year, using high school coursework and grades as primary criteria. The Public Policy Institute of California's analysis reports that 58 percent of first-time mathematics students completed a transfer-level course on their first attempt in fall 2024, a seven-point gain since fall 2022, when access was already near universal. The gains have been concentrated in statistics and liberal arts mathematics; college algebra, trigonometry and precalculus have improved more slowly. The same body of work finds no evidence that opening the courses diluted them.
The Precedent the Decision Sets
What is being decided, when a college adopts directed self-placement, is not only who sits in which classroom next term. It is whether the institution's judgement about a student's preparation is a service it owes or a gate it operates — and that precedent will govern the next reform, and the one after.
Three questions are worth putting to any committee now drafting such a policy. Does the guidance match the choice, in staff time and in honest information about failure rates as well as success stories? Are the subgroup outcomes being tracked separately, or will the college learn in five years that choice widened a gap it never measured? And has anyone costed the advising, because the Florida evidence suggests the labour lands on people already carrying a full caseload.
The gap that should trouble the field most is the one Morton identified: no large-scale rigorous evaluation has yet been mounted. Descriptive studies from single institutions, however careful, cannot tell a system what will happen when the policy goes statewide. Until that study exists, colleges will keep making the decision on principle and anecdote — and the principle, on the evidence available, is defensible without being proven.
- higher-education-policy
- higher-education-policy
- student-advising
- student-advising
- directed-self-placement

