Specifications grading
A way of grading a course without scoring every piece of work on a number line. Each assignment is judged pass or fail against a written standard; students get a small budget of tokens to revise or replace work that did not pass; and the final grade depends on which combinations of passing work a student completes, with those combinations announced up front. The aim is to keep a try-revise-resubmit rhythm alive inside schools that still demand a single letter grade at the end.
A grading scheme, articulated in Linda Nilson's 2015 book, that combines three features: each assignment is evaluated on a pass / fail basis against a published specification rather than scored on a continuous scale; students receive a bounded budget of tokens or bundles allowing revision or replacement of work that did not meet the specification; and the course grade is determined by which combinations of acceptable submissions the student produced over the term, with the combinations published at the start. The scheme is intended to preserve a formative revise-and-resubmit loop inside institutions that require a terminal letter grade.
Etymology§
The term specifications grading is Nilson's coinage in the 2015 Stylus volume. The scheme's structural elements have antecedents: pass / fail grading inside selective institutions has a long history; mastery-learning's test-correct-retest loop (Bloom, 1968 onwards) is the acquisition-side antecedent; contract grading in writing- studies (Peter Elbow, 1993; Asao Inoue, 2014 onwards) is the closest practical relative. What Nilson assembled was the integration of these into a single coherent scheme designed specifically for university teachers facing institutional grade-book requirements.
The scheme's three structural features address three different problems with conventional grading. Pass / fail evaluation against a specification eliminates the continuous-scale problem that Butler 1988 identified — there is no numerical score for the student to attend to in preference to the comment content, so the formative information has a clearer path to the learner's attention. The specification has to be written carefully and operationally enough that meets / does not meet is a defensible judgement; this is the implementation difficulty the scheme requires teachers to do up-front work on.
Tokens or bundles institutionalise the revise-and-resubmit loop. In a typical implementation, each student receives a small number of tokens at the start of the term (often three to five), and a token can be spent to revise a piece of work that did not meet specification or to replace one assignment with another. The bounded budget prevents the loop from running indefinitely while making real the opportunity to close the gap on the most important work. The structural commitment matters: in conventional grading, the opportunity to revise is at the teacher's discretion and is often unused because students do not believe revision will change the grade; tokens make the opportunity contractually real.
The combination-based final grade makes the relationship between what work was done and what grade is earned transparent at the start of the term. A typical scheme will specify, for example, that a B requires meeting the specification on assignments X, Y, and Z and one additional elective; an A requires the same plus a longer final project meeting an additional specification. Students choose their own target grade and the corresponding work; the grade follows the work without intermediate calculation. The transparency removes the what does the teacher want problem that conventional grading produces and substitutes a what work do I want to do question.
Between the two poles of grading reform, the scheme sits on the procedural side: it keeps a contractual grade-determination structure, where the more radical ungrading move removes that structure and treats the terminal grade as a conversation. Practitioners move between the two, and the choice is typically a function of institutional constraint. Documented implementations run across mathematics (Robert Talbert), biology (Stephanie Chasteen), philosophy, foreign-language instruction, and writing studies, mostly in US higher education since 2015; as with the rest of the grading-reform literature, the evidence is largely self-reported rather than drawn from controlled comparison.