How can AI help D.C. evaluators write teacher evaluations under the DC Model Rubric?
How can AI help D.C. evaluators write teacher evaluations under the DC Model Rubric?
July 05, 2026 • 17 min read
Washington D.C. isn't a state, and it doesn't have a single teacher evaluation instrument used everywhere. The DC Model Teacher Evaluation System (MTES) — published by the Office of the State Superintendent of Education (OSSE) — is a voluntary framework of four domains and twenty-three components, scored on a four-level scale from Ineffective to Highly Effective, that local education agencies (LEAs) may adopt. DC Public Schools, the traditional public system, uses a separate instrument entirely, called IMPACT. AI can meaningfully help evaluators using the DC Model Rubric draft their observation write-ups — mapping evidence to the right components and generating feedback in the rubric's own language. What it can't do is tell you which instrument your particular LEA actually uses, since that's a locally made decision in D.C.'s unusually decentralized landscape.
What Washington D.C.'s teacher evaluation landscape actually is
To understand evaluation in D.C., start with what D.C. is structurally: a federal district, not a state, where public education is delivered through two parallel sectors. DC Public Schools (DCPS) is the traditional public school system, chancellor-led, serving roughly half of D.C.'s public school students. The other half attend public charter schools, each authorized as its own independent local education agency by the DC Public Charter School Board, with substantial autonomy over curriculum, staffing, and — relevant here — teacher evaluation. OSSE, the Office of the State Superintendent of Education, functions as D.C.'s state education agency, but it does not mandate a single evaluation instrument across every LEA the way most state departments of education do.
Into that landscape, OSSE built the DC Model Teacher Evaluation System. Launched on November 13, 2014, MTES was a collaborative project between OSSE, Thurgood Marshall Academy Public Charter School, and thirteen other LEAs, developed over nine months of planning meetings focused on professional development, rubric language norming, and stakeholder feedback. The result was a comprehensive teacher evaluation framework and associated rubric — explicitly described by OSSE as a tool LEAs may use, not one they're required to use. It's a resource, hosted alongside training materials, a communications toolkit, and a companion classroom observation guide, primarily adopted by charter LEAs that don't already run their own proprietary system.
DCPS, meanwhile, uses its own separate system called IMPACT. Enacted for the 2009–2010 school year, IMPACT predates MTES by several years and is one of the most closely studied teacher evaluation systems in the country — it combines classroom observation with other measures, including, for some grades and subjects, a value-added model built with Mathematica Policy Research. IMPACT is structured differently from the DC Model Rubric, with its own domains and components. Notably, IMPACT is currently the only DC evaluation system OSSE formally accepts for educator credential renewal purposes — a detail that matters if you're trying to understand how much institutional weight each system carries.
The four domains and twenty-three components
The DC Model Rubric organizes teaching practice into four domains:
Domain 1 — Learning Environment (3 components): positive relationships and respect; procedures and routines; and physical environment.
Domain 2 — Delivery of Instruction (6 components): culture of learning; student engagement; questioning techniques; execution of lesson; content knowledge; and use of assessments.
Domain 3 — Planning and Preparation (5 components): knowledge of content and pedagogy; knowledge of students; long-term instructional planning; daily instructional planning; and student assessment.
Domain 4 — Professional Foundations (9 components): department/grade level contributions; lesson plans; professional feedback; school contributions; professional development and individual growth; family contact and communication; record keeping; professionalism; and attendance.
That's twenty-three components in total — and Domain 4 stands out. At nine components, Professional Foundations is by far the largest and most operationally detailed domain in the rubric, covering everything from meeting deadlines to submitting lesson plans on time to maintaining substitute plans. It reads less like a typical "professionalism" domain in other state frameworks and more like a genuine accountability checklist — a distinctive feature of how the DC Model Rubric was built.
The four-level scale
Each component is rated Ineffective, Minimally Effective, Effective, or Highly Effective. Ineffective reflects significant gaps in practice with minimal or harmful impact on student learning. Minimally Effective describes a teacher attempting the practice with positive intent, but inconsistent or falling short in implementation. Effective is the professional standard: the teacher consistently meets expectations, delivering solid, expected professional practice with a positive impact on student learning. Highly Effective sits above that — practice that consistently impacts all students and, notably, often includes serving as a model or mentor for peers.
The distinction that matters most for evaluators: Effective is not a middling score. It represents genuine, consistent proficiency — the level the rubric is built for a solid teacher to land on. Highly Effective specifically calls for evidence of leadership initiative and mentorship, not just excellent individual classroom practice. A teacher can run an outstanding classroom and be accurately rated Effective; Highly Effective adds something beyond the individual classroom.
After the observation: where the work piles up
The observation itself is familiar territory for any experienced evaluator. The write-up that follows is where the time goes.
A single DC Model Rubric evaluation asks the evaluator to take fragmentary notes and turn them into evidence mapped across as many as twenty-three components in four domains — including all nine granular Professional Foundations components — decide a best-fit rating for each on a four-level scale, and draft feedback in the rubric's language that a teacher can act on. Multiply that across a caseload, and the sheer component count in Domain 4 alone makes DC Model Rubric documentation one of the more time-intensive evaluations to produce by hand among the frameworks EvalScribe supports.
The observation takes a class period. Turning it into a complete, twenty-three-component write-up takes considerably longer — and the granularity that makes a DC Model Rubric evaluation genuinely useful is exactly what makes it slow to produce manually.
Where AI helps with the DC Model Rubric write-up
The strongest fit between current AI and DC Model Rubric documentation is the mapping-and-translation problem, made more acute here by the rubric's unusually high component count.
AI handles three parts of that reliably. First, it turns fragmentary notes into coherent, evidence-anchored prose in the rubric's voice — the difference between "kids explained their thinking to each other" and a sentence tied to Domain 2's questioning techniques component. Second, it maps evidence to the relevant components across all four domains, including evidence that touches more than one — a well-planned, differentiated lesson can support Domain 3's daily instructional planning component and Domain 2's student engagement component at once. Third, it maintains consistency across all twenty-three components, including the nine operational components in Professional Foundations, so nothing gets collapsed into vague generalities simply because there are so many boxes to fill.
For D.C. specifically, the payoff is in taking a rubric with real breadth — three domains' worth of instructional components plus a uniquely detailed professionalism domain — and drafting all of it with the specificity each individual component deserves, rather than the generic summary a rushed write-up tends to produce.
Where AI doesn't help — and what stays with the evaluator
The honest scope here is shaped directly by D.C.'s unusual structure.
AI cannot tell you which evaluation system your LEA actually uses. Because D.C. has no single mandated instrument, that's a fact only your school or LEA can confirm — MTES, DCPS's IMPACT, or something else entirely, depending on whether you're in a traditional DCPS school or one of D.C.'s many independently authorized charter LEAs.
AI cannot replicate IMPACT. The DC Model Rubric and IMPACT are structurally different instruments; a tool built around one cannot substitute for the other, even though both happen to use similar-sounding rating labels.
AI cannot verify the operational facts in Professional Foundations. Attendance records, deadline compliance, and family contact logs are factual matters that belong to the evaluator's and school's own records — an AI tool can help you write them up, but it can't independently confirm them.
AI cannot verify genuine Highly Effective evidence — mentorship, leadership initiative, and consistent impact across all students — from a well-written paragraph alone. That judgment depends on what was actually observed and known about a teacher's broader contribution, not on how polished the resulting write-up sounds.
And AI cannot supply the institutional knowledge a seasoned D.C. evaluator brings — how their specific LEA's evaluation policy is bargained, what their charter's board actually weighs in employment decisions, how their school's culture interprets components like School Contributions. That context shapes a rating in ways no chatbot reconstructs.
What AI does well, it does well. Pasting notes into a general chatbot functions as a glorified search engine — but for a twenty-three-component rubric that needs each component individually and specifically addressed, the gap between "sounds polished" and "actually complete" is significant.
Observable evidence: what each domain looks like in practice
Each domain surfaces in things an evaluator actually sees, hears, and reviews.
Learning Environment shows in fair, culturally responsive interactions between teacher and students, clear and consistently enforced routines that protect instructional time, and a classroom that's clean, organized, and arranged to support access for every student. Delivery of Instruction shows in genuine teacher enthusiasm paired with explicit belief in every student's capability, differentiated and well-paced instruction, questioning that moves beyond recall into student-to-student discussion, objective-driven lessons with real closure, accurate and rigorous content delivery, and assessment data that visibly shapes instruction in real time.
Planning and Preparation shows in instructional plans grounded in deep content and pedagogical knowledge, explicit awareness of individual students' needs including those with IEPs and English Language Learner status, coherent long-term plans aligned to standards and external assessments, daily plans with anticipated misconceptions built in, and assessments designed with clear criteria and backwards planning. Professional Foundations shows in reliable meeting attendance and deadline compliance, complete and timely lesson plan submission, receptiveness to feedback, visible participation in school-wide initiatives, applied professional learning, documented and proactive family communication, accurate record keeping, consistently professional conduct, and dependable attendance with quality substitute plans.
Common scoring mistakes under the DC Model Rubric
Even experienced D.C. evaluators fall into a few recognizable patterns.
The first, and most consequential, is assuming the DC Model Rubric applies without checking. Given how many different evaluation systems operate across D.C.'s LEAs, it's easy to default to a familiar rubric that turns out not to be the one your specific school or charter network actually uses.
The second is conflating the DC Model Rubric with IMPACT. They share similar-sounding rating labels (Ineffective, Minimally Effective, Effective, Highly Effective), which makes it tempting to treat them as interchangeable. They're not — the domains, components, and underlying design are genuinely different instruments.
The third is treating Effective as underwhelming. As with several other state and local scales built this way, Effective is the professional standard, not a disappointing middle score. Reserving Highly Effective for genuine leadership and mentorship evidence keeps that top rating meaningful.
The fourth is shortchanging Professional Foundations. With nine components, it's the domain most likely to get compressed into a few generic sentences under time pressure — but each component (attendance, record keeping, family communication, and the rest) has its own distinct rubric language and deserves individual attention.
The fifth is losing precision across such a large component count. Twenty-three components is a lot to hold in working memory during a single observation write-up; evidence that would clearly support one specific component sometimes gets folded vaguely into a nearby one instead.
D.C. in context — no single system, by design
Washington D.C.'s evaluation landscape doesn't resemble any single state's. Most states pick one instrument (mandated or model) and apply it broadly, even when local flexibility exists. D.C.'s structure — a chancellor-run traditional system running its own proprietary IMPACT alongside dozens of independently authorized charter LEAs, some of which adopt OSSE's voluntary DC Model Rubric and some of which build or adopt something else — produces a genuinely fragmented landscape by design, a direct consequence of D.C.'s charter sector educating close to half of its public school students under real institutional autonomy.
For evaluators and administrators working across this landscape, the practical takeaway is simple: confirm which instrument governs your specific school before assuming. The DC Model Rubric is a strong, well-built option used by a meaningful share of D.C.'s charter LEAs — but it's one option among several, not a district-wide default the way OTES 2.0 is for most Ohio districts or the CCT Rubric is for most Connecticut districts.
How EvalScribe handles the DC Model Rubric
EvalScribe is built around the actual DC Model Teacher Evaluation System — all four domains, all twenty-three components — rather than approximating them with generic teaching-evaluation prose. Every component carries the rubric's own language across the four levels, the look-fors an evaluator watches for, and the distinctions that separate Minimally Effective from Effective and Effective from Highly Effective. An evaluator captures notes by typing, dictating, or photographing handwriting (Smart Scan OCR converts it to text), and EvalScribe maps that evidence to the relevant components — including all nine Professional Foundations components individually — and drafts best-fit ratings and feedback in the rubric's voice, treating Effective as the genuine professional standard it's designed to be.
A few honest scope notes, because D.C.'s landscape demands them. First, the DC Model Teacher Evaluation System is a voluntary framework, not a district-wide mandate; EvalScribe is built around it specifically, and it's most directly useful for the LEAs — largely charter schools — that have adopted it. Second, DCPS's separate IMPACT system uses different domains and components; native IMPACT support is on our roadmap for a future update, and we're glad to hear from you at [email protected] if that would help your school sooner. Third, IMPACT remains the only DC evaluation system OSSE accepts for educator credential renewal, which is worth knowing regardless of which instrument your school uses day to day.
Beta testers report saving 30 to 60 minutes per evaluation versus writing the documentation traditionally — a savings that compounds given how many components the DC Model Rubric asks an evaluator to address individually. More detail on the D.C. workflow is available at evalscribe.com/washingtondc.
Frequently asked questions about the DC Model Rubric and AI
Is the DC Model Rubric used across all of Washington D.C.? No. It's a voluntary framework LEAs may adopt, used by a number of D.C.'s independent public charter LEAs. DCPS uses its own separate system, IMPACT.
What's the difference between the DC Model Rubric and IMPACT? They're different instruments with different domains and components. MTES was built in 2014 by OSSE with charter LEAs; IMPACT is DCPS's own system, in place since 2009–2010.
What does Effective mean if it's not the top rating? It's the professional standard — solid, expected practice with a positive impact on learning. Highly Effective requires evidence of exceptional performance and leadership or mentorship beyond the individual classroom.
Does AI replace evaluator judgment? No. EvalScribe drafts best-fit ratings and evidence-based feedback; the evaluator reviews, edits, and finalizes every rating.
Does EvalScribe support IMPACT? Not yet — native IMPACT support is on our roadmap for a future update. EvalScribe's D.C. framework today is the DC Model Teacher Evaluation System.
Where can I read the official sources? OSSE's DC Model Teacher Evaluation System Resources page hosts the framework, rubric, and training materials. DCPS documents IMPACT separately at dcps.dc.gov.
If you're evaluating teachers in D.C. under the DC Model Rubric, see how EvalScribe handles the four-domain, twenty-three-component workflow at evalscribe.com/washingtondc.
References
Office of the State Superintendent of Education, DC Model Teacher Evaluation System Resources
DCPS, IMPACT: The DCPS Evaluation and Feedback System for School-Based Personnel
OSSE, Credential Renewal (educator credential renewal)
Related articles
Page maintained by Anthony D. Neely, Ph.D. — practicing K-12 educator with nearly 20 years in the classroom, 2025–2026 Walker County Distinguished Teacher of the Year, and co-founder of EvalScribe. Framework details verified against the DC Model Teacher Evaluation System and OSSE sources.
Last updated on July 28, 2026.
