How is Florida’s B.E.S.T. Writing scored, and what caps the Development score?

What is B.E.S.T. Writing, and how is it different from FAST?

Florida’s writing assessment is B.E.S.T. Writing, short for Benchmarks for Excellent Student Thinking. It is a standalone test, not part of FAST. FAST is the Florida Assessment of Student Thinking, the state’s Coordinated Screening and Progress Monitoring System, and it covers ELA Reading and Mathematics across three events a year. Writing sits outside that cycle: students take it once, in the spring, and it is explicitly not PM1, PM2 or PM3.

The program is newer than most people assume. The State Board adopted the B.E.S.T. Standards in February 2020, and March 2022 legislation replaced the Florida Standards Assessments in ELA and Mathematics with FAST under s. 1008.25(9), F.S. Writing was field tested in spring 2023 and first administered operationally in spring 2024, making 2026 only the third operational year.

The task is fixed. Every student in Grades 4 through 10 reads two to four passages, then produces one typed multiparagraph essay using evidence from those sources, in 120 minutes, computer-based, with paper only for IEP and 504 accommodations. Used Writing Planning Sheets are secure materials, tracked on a chain-of-custody form and locked in storage.

Two quirks belong to writing alone. A student still working when time expires may continue for up to half a typical school day, where every other Florida statewide test allows a full one. And writing is the single exception to the state’s 24-hour reporting rule in the Florida Reporting System. The 2026-27 schedule lists a Writing window of March 29 to April 9, 2027.

What does the 12-point rubric actually score?

Three domains worth four points each: Purpose/Structure, Development and Language. There is no separate conventions domain and no separate score for reading the sources. What a student does with the passages surfaces inside Development.

There are four rubrics rather than one, split by grade band and mode: Grades 4-6 Argumentation, Grades 4-6 Expository, Grades 7-10 Argumentation, Grades 7-10 Expository. The score point labels are plain. A 4 is above grade-level accomplishment demonstrated, a 3 is within the range of grade-level performance, a 2 is approaching that range, a 1 is below it. Scoring is holistic by domain: a response earns a score point by demonstrating most of the descriptors there, not all of them.

Check the version before you circulate a rubric. The standalone rubric PDFs still carry draft-era front matter describing a spring 2023 field test and calling the rubrics drafts pending standard setting. The versions inside the 2026 release documents rewrote that in past tense and dropped the caveat. If a colleague believes the rubric is provisional, that is why.

Why does a missing citation cap Development at 2?

All four rubrics carry the same footnote, and it is the sharpest line in the document. Citation, it says, “is not a holistic consideration”. Without citation, the highest score a response can earn in Development is a 2.

Set that against the rest of the rubric and the oddity stands out. Development is otherwise judged holistically, on the weight and integration of evidence, on elaboration, on reasoning. This rule is mechanical and absolute. A Grade 10 essay with a controlled position, well-chosen evidence and real elaboration is still held to 2 out of 4 in Development if the student never attributes anything to a source. Two of twelve points, lost to a habit rather than to thinking.

Since every student writes from provided passages, attribution is the scoreable act, not a stylistic nicety. It is also the highest-yield thing to check on a practice essay: unlike most rubric judgments it is binary and takes seconds to spot.

What changes between the Grades 4-6 and Grades 7-10 rubrics?

The band break at Grade 7 is a change in criteria, not in wording. The argumentative anchor term shifts from Claim to Position, footnoted as Claim in Grade 7 benchmarks. Transitions move from varied strategies that enhance the progression to purposeful strategies that create cohesion. Rhetorical techniques carry a Grade 6 footnote in the 4-6 rubrics, so Grades 4 and 5 are not expected to use them at all.

Counterclaims are the difference that changes a score. In Grades 4-6 a counterclaim may be present, and it is mentioned only at score point 4. In Grades 7-10 it is scored at every point, from fully addressed down to absent or confusing. A Grade 7 student who omits one is held down in Development; a Grade 6 student writing the same essay is not. In a middle school teaching both grades, that gap is invisible unless someone reads both rubrics.

• Grades 4-6: counterclaim optional, mentioned only at the top score point
• Grades 7-10: counterclaim scored at all four points
• Grades 4 and 5: rhetorical techniques not expected
• Grade 7 onward: Position replaces Claim as the anchor term

How are the essays actually scored, by people or by machine?

Both, in a published ratio. An automated engine scores the majority of responses, trained on Florida student work already scored by at least two humans after the field test. State documentation names the scoring sub-contractor as Data Recognition Corporation, with human scorers trained by department ELA content specialists.

The quality control is unusually specific. The engine emits a confidence score for each response, and low-confidence responses are always rescored by a human. Up to 40 percent of responses are human-scored, counting unusual ones and those below the confidence threshold. A random 5 percent at each grade goes to humans purely as an engine-versus-human check. Agreement is tracked with Quadratic Weighted Kappa and Standardized Mean Difference. The engine can also be retrained partway through an administration, and papers it scored up to that point can be reset and scored again.

Districts have recourse too: a score inquiry asks ELA specialists to confirm a response was scored correctly, with the prompt, source material and annotated anchor sets supplied. All of it is set out in the state’s assessment guide, which is what to cite when a parent asks whether a machine graded the essay.

What does a B.E.S.T. Writing score actually count for?

Less than teachers expect. Reported measures are a raw score out of 12 plus the three domain scores out of 4. There is no scale score, no achievement level and no proficiency label. Every other statewide test reports a scale score and one of five achievement levels, with Level 3 as passing.

Three consequences follow. Writing is not combined with FAST ELA Reading into an overall ELA score, an explicit break from FSA Writing, which did contribute. It does not factor into graduation requirements, where state-level consequences attach to Grade 3 ELA Reading, Grade 10 ELA Reading and the Algebra 1 EOC. And it appears nowhere in the school grade calculation, which runs to as many as twelve components.

The compensation is transparency. Statute requires publication of a statewide assessment only every three years, but the department releases the writing prompts and student responses every year, per grade. The 2026 documents went up in late June, about eleven weeks after the testing window closed, each carrying the prompt, the source passages and the rubric used.

How should a department prepare students during the year?

Preparation follows the rubric, not a test-prep packet. Give students source sets of two to four passages rather than open prompts, because reading and using sources is the task. Teach both modes, since only one is assessed per grade per year and it moves. Make citation automatic on every draft. Released prompts, the annotated anchor sets built by Florida educator rangefinder committees, and the state scoring samplers are the cheapest calibration material there is, and they pair with your own study guides.

Calibrate as a team before the window, not after it. Score five anonymized responses independently against one band rubric and argue out the disagreements. The spread is usually widest in Development, exactly where the citation footnote and the counterclaim expectations do the most damage.

The practical limit is feedback volume. Wanting a source-based essay every few weeks means grading a great many of them, and by hand the practice frequency collapses first. That is where a tool like Jeddle fits: JeddAI drafts rubric-scored comments against criteria you set up, and you review each one before a student sees it. Teams wanting expert essay feedback on Florida tasks should encode the citation rule into the rubric, since a generic one will not apply the state’s cap.