How is writing scored on Colorado’s CMAS English language arts test?

What does a Colorado student actually write for the state test?

Colorado’s state English language arts test is the Colorado Measures of Academic Success, or CMAS, taken every year in Grades 3 through 8. It runs as three units of 90 minutes, carries 21 to 25 items worth 65 to 80 points, and sorts students into five levels from Did Not Yet Meet Expectations to Exceeded Expectations.

Extended writing appears as Prose Constructed Response items, known as PCRs, in three passage-based task types: the Literary Analysis Task, the Research Simulation Task and the Narrative Task. Each student writes exactly two PCRs per ELA form, and the Writing claim is built from the trait scores on both.

High school students write nothing extended for the state. Colorado’s high school measures are the PSAT in Grades 9 and 10 and the SAT in Grade 11, and since spring 2024 the digital SAT’s Reading and Writing section has been 54 questions with no essay. A student finishing CMAS in Grade 8 has produced their last state-scored piece of writing.

What are the two dimensions on the CMAS writing rubric?

There are two named traits, and the names shift with the task. For the Research Simulation Task and the Literary Analysis Task in Grades 6 to 8, the scoring guide gives Reading Comprehension and Written Expression on 0 to 4, and Knowledge of Language and Conventions on 0 to 3. The conventions dimension has no score point 4 column. On the Narrative Task the top dimension is simply Written Expression, because the reading dimension is not scored for elicited narrative stories.

Grade 3 caps both dimensions at 3, and the 4-point scale first appears at Grade 4, on the second response. The footnotes matter too. Tone is not assessed in Grade 6, and coherence, clarity and cohesion are tied to the grade-level writing standards rather than judged in the abstract, so identical rubric wording means different things at Grade 6 and Grade 8.

The naming is inconsistent across documents. Score reports use Written Expression (WE) and Knowledge and Use of Language Conventions (WKL), while the guide a rater reads says Reading Comprehension and Written Expression.

Why do conventions count for so little of the writing score?

Because Colorado says so, in writing. Its technical report states that the Writing claim sums the individual trait scores, with Written Expression multiplied by three to emphasize the additional importance of the content of the writing as reflected in the standards. That is a deliberate decision to hold grammar and mechanics below content.

The arithmetic is blunt. Using Grades 6 to 8 point values on a single response, a student scoring 4 on Written Expression and 0 on conventions carries 12 points. A student who writes cleanly but says little, at 1 on Written Expression and a full 3 on conventions, carries 6. Across a whole Grades 6 to 8 form, conventions contribute at most 6 of the 30 Writing points, a fifth. At Grade 3, where both dimensions cap at 3, they reach a quarter.

Correct sentences will not rescue a thin argument, and an argument that lands survives a fair amount of surface error. If your department spends its grading energy on comma splices, move it. Claim, evidence and reasoning is where four fifths of the Grades 6 to 8 writing score sits.

Why does one rubric judgment count twice?

Look at the name of the top dimension on evidence-based tasks: Reading Comprehension and Written Expression. It is one judgment routed to two places. The technical report is direct: some PCR items also carry reading literature or reading informational points, and the reading points for those items were a duplicate of the Written Expression score.

So a single rater decision lands in a Reading subclaim and in the Writing claim at once. A student who misreads the passage is charged twice, and one who reads accurately is credited twice. That makes accurate use of the source the highest-value thing to teach on a Research Simulation Task, well ahead of polish.

What do the published CMAS score distributions show?

Colorado publishes item-level distributions, and they are sobering. On the 2025 operational Grade 3 form, the first Prose Constructed Response scored on Written Expression out of 3 points: 46.49 percent scored 0, 45.20 percent scored 1, 7.23 percent scored 2, and 0.09 percent scored 3. Nine in ten third graders landed on the bottom two points, and fewer than one in a thousand reached the top point.

Grade 4 loosens a little. On the second response, scored out of 4, the distribution runs 30.87, 54.42, 12.07, 1.75 and 0.32 percent. Statewide in 2026, 40.8 percent of Grade 3 students met or exceeded expectations in English language arts, rising to 49.9 percent at Grade 5 and back to 42.2 at Grade 8.

The lesson for a classroom rubric is calibration. A scale that hands out top bands freely is not describing what the state scale describes. Moving a student from 1 to 2 on Written Expression puts them ahead of most of the state on that trait.

How reliable is the human scoring, and why is writing not automated?

Colorado already runs an automated scoring engine in production. The 2025 administration used a mix of machine, human and automated scoring, with a small number of mathematics constructed responses scored by the Math Reasoning Engine. The ELA Prose Constructed Responses were scored on two trait dimensions using human scoring. The state has the technology and does not point it at writing.

Human scoring carries published tolerances, and they loosen as the rubric gets longer. One and two point items require 90 percent exact agreement between raters. Three point items require 80 percent, four point items 70 percent, and items worth five or more 65 percent. Ten percent of responses on each item go to a second reader.

So on the 4-point writing dimension, Colorado’s own accepted floor is 70 percent exact agreement between two trained raters working from the same guide. Some disagreement is built into the instrument.

What about students who write in Spanish?

Colorado Spanish Language Arts, or CSLA, is a separate Spanish-language form first administered in 2016. It covers Grades 3 and 4 only, and stayed there for spring 2026. The Grade 4 scoring guide carries two parallel rubrics, one English and one Spanish, with dimensions named Reading Comprensión y Expresión Escrita, Expresión Escrita, and Conocimiento del lenguaje y convenciones. The conventions dimension scores the conventions of standard Spanish.

Scoring works differently too. CSLA is administered on paper only, its scorers must be proficient in written Spanish and English, and they may only score during daily online meetings with supervisory staff, Monday to Friday in business hours. Most other human-scored items go to distributed scorers who work seven days a week with extended evening hours.

The eligibility rules create an odd edge. A Spanish-speaking newcomer in Grade 4 receiving Spanish language arts instruction takes CSLA and writes. The same student one grade higher is exempt from English language arts and writes nothing. House Bill 25-1278, signed on May 23, 2025, expands Spanish reading and writing tests to Grades 5 through 8 once 1,500 English language learners statewide receive native-language support within an assessed grade.

How should a department turn this into its own grading rubric?

Four changes make a classroom rubric behave like the state one. Weight content three times conventions. Score reading accuracy and written expression as one judgment on evidence-based tasks. Keep the top band rare enough that it means something. And double-grade ten percent of a class set with a colleague to see whether you clear 70 percent exact agreement.

Holding a three-to-one weighting steady across five sections by hand is where the hours go. Jeddle’s expert essay grading applies the success criteria you wrote rather than a generic template, so the last script gets the same weighting as the first. JeddAI drafts the comment and score, and you edit or reject every one before a student sees it.

Students should be able to name the criteria too. A Grade 8 class that knows the top dimension rewards accurate use of the passage can aim at it. Subject-specific study guides give them somewhere to rehearse the two responses they meet on test day.