How is the STAAR essay scored, and what does the rubric measure?

What does a Texas student actually have to write on STAAR?

Since the 2022-2023 school year every Texas reading language arts test carries an extended constructed response, or essay: Grades 3 through 8, plus English I and English II. Students write in response to a reading selection, in one of two modes, informational or argumentative. From Grade 6 through English II a prompt may instead ask for correspondence, a letter written to a specific audience, scored on the same rubric.

The change came from House Bill 3906, passed by the 86th Texas Legislature in 2019. The essay component shifted from writing a response to a stand-alone prompt to writing a response based on a reading passage. The same bill capped multiple choice at no more than 75 percent of the points on a STAAR test, which is why constructed response now sits on every form.

The prompt wording is stable enough to teach directly. The spring 2024 English I prompt listed five reminders: clearly state your thesis, organize your writing, develop your ideas in detail, use evidence from the excerpt in your response, and use correct spelling, capitalization, punctuation, and grammar. The Grade 5 prompt from spring 2025 runs the same five reminders, but asks for a central idea rather than a thesis and points students to the selection rather than the excerpt. Neither gives a word count.

Two shorter tasks sit alongside the essay: a reading-domain short constructed response scored 0 to 2, and a writing-domain one scored 0 to 1. Short constructed responses also appear on Grade 5 and Grade 8 science, Grade 8 social studies, Biology and U.S. History.

What do the two rubric traits actually measure?

Four rubrics are published, all dated Fall 2022, split by grade band and mode: Grades 3 to 5 and Grades 6 to English II, each in an argumentative or opinion version and an informational version. Every one of them has two traits and nothing else, so teachers arriving from a six-trait writing rubric tend to over-prepare on categories Texas does not score.

The first trait is worth 0 to 3. The Grades 6 to English II rubric heads it Development and Organization of Ideas; the Grades 3 to 5 rubric prints the same words in the other order. Four named sub-criteria sit inside it, reading at the top score point: the argument, opinion or central idea is clear and fully developed; organization is effective; evidence is specific, well chosen, and relevant; and expression of ideas is clear and effective.

The second trait is Conventions, worth 0 to 2, defined as consistent command of grade-level-appropriate conventions including correct sentence construction, punctuation, capitalization, grammar and spelling. That is the whole rubric: five points from two columns.

One rule catches people out. If a response receives a score point 0 in the Development and Organization of Ideas trait, it also earns 0 in Conventions. Clean sentences that do not answer the question earn nothing. Students should hear that plainly.

Two discriminators are grade-banded inside the trait rather than given their own row. For Grades 8 through English II, counterarguments must be identified and refuted for 3 points, identified but not refuted at 2, and not identified at 1. For paired passages in Grades 6 through English II, top-band evidence is drawn from both texts, while evidence from one text sits at 1. Released tests carry annotated exemplars at every score point, and those annotations, read alongside your own study guides for the subject, teach both discriminators faster than any general advice about essay structure.

Who actually scores the essay?

This is the part with no real equivalent in other states. Responses are initially scored by an automated scoring engine and at least 25 percent are then routed to human scorers. Every English-language constructed response gets a machine first pass.

Spanish is carved out completely. Responses to short and extended constructed response questions on STAAR Spanish assessments are 100 percent human scored and do not get scored by the engine. Same state, same rubric, different scoring technology, split by language of administration.

The engine is trained on student responses and human scores from field-test data, to emulate how humans would score each question. The stated acceptance bar is that it must agree with human scorers at the same rate human scorers agree with one another.

There is also a published list of what the engine will not score. Condition codes flag a response that is blank, uses too few words, uses mostly duplicated text, is written in another language, consists primarily of stimulus material, uses vocabulary that does not overlap with the vocabulary in the subset of responses used to train the engine, or uses language patterns reflective of off-topic or off-task responses. Those responses, and any the engine flags as low confidence, go to trained human scorers. Low confidence usually means a response on the border between two score points. A human score always becomes the score of record.

One number surprises people. Extended constructed responses are scored by two different scorers and the scores are summed, so students may receive up to 10 points for the essay, even though the published rubric tops out at 5.

What changes when House Bill 8 replaces STAAR?

House Bill 8, passed in 2025, replaces STAAR with a program referred to in statute as the Student Success Tool. It starts with the 2027-2028 school year, and STAAR remains the live program through 2026-2027. The replacement runs beginning, middle and end of year windows for Grades 3 through 8, with the first two optional for Algebra I, Biology, English I and U.S. History.

For English departments the biggest item is that the English II end-of-course assessment is being eliminated, as an assessment and as a graduation requirement, from 2027-2028. The classes of 2026 and 2027 still have to take and pass it. English I survives, and the refreshed 2028 accountability system drops English I to English II growth.

Two statutory details are specific to writing. Section 39.02301(i) of the Education Code puts the writing component of the end-of-year reading language arts test in its own window, April 1 to April 15, so responses can be scored in time for the reporting deadline in subsection (l), which is two business days after a window closes. Section 39.02301 also requires an automatic rescore whenever the agency determines that a one-point improvement would move a student to the next performance level.

Do not plan against a rubric that does not exist yet. Draft blueprints and rubrics for the new program are listed as coming soon. For 2026-2027 the Fall 2022 rubrics are the ones that count.

What does this mean for how you grade writing in class?

Grade classroom writing on the two traits Texas actually uses rather than a longer rubric from elsewhere. Score development and organization out of 3, conventions out of 2, and apply the zero rule honestly. Then double the total, so the number on the page matches a score report.

Build the grade-banded items in early. Counterargument work starts counting at Grade 8, so Grade 7 is where it should be taught rather than assessed. Paired-passage synthesis starts counting at Grade 6. And since the condition codes cover off-topic responses, copied stimulus material and responses that are too short, the best instruction for a struggling writer is to answer the question in their own words at reasonable length.

The volume problem is real. An essay on every reading language arts form from Grade 3 up, plus short constructed responses across science and social studies, is more student writing than most departments have graded before. Jeddle exists for that problem. You supply the rubric, the tool drafts a score and comments against it, and you edit and approve before a student sees anything. A department that wants expert essay grading against the two-trait rubric can load the official descriptors verbatim and reuse them all year.

Keep the human decision where the state keeps it. On the state test a human score is the score of record whenever a person reads a response. The same applies to any classroom drafting tool: JeddAI produces a first pass and the teacher owns the grade.