Formative Assessment Guide: Types of Assessment, Formative vs. Summative, Standardized Testing, Grading and Accommodations

Formative assessment is the checking a teacher and students do while learning is still under way, so that what they find out changes what happens next: the next question, the next lesson, the next draft. The Council of Chief State School Officers defines it as “a planned, ongoing process used by all students and teachers during learning and teaching to elicit and use evidence of student learning” to improve students’ understanding and help them become self-directed learners. A 1998 review made it a priority: Paul Black and Dylan Wiliam reported that typical effect sizes in the formative assessment experiments they examined “were between 0.4 and 0.7,” larger than for most educational interventions. Later and stricter reviews have found smaller but still positive effects: a 2011 meta-analysis by Neal Kingston and Brooke Nash calculated a weighted mean of 0.20 (ERIC), and a 2024 meta-analysis of 118 studies found 0.25 overall and 0.22 in U.S. studies (Yao et al., 2024). Formative assessment sits alongside a great deal of other testing. In 2014–15, students in 66 large urban districts were required to take an average of 112.3 standardized tests between prekindergarten and grade 12 (Council of the Great City Schools, October 2015), and the results of the largest national test remain sobering: on the 2024 National Assessment of Educational Progress, around 40 percent of fourth graders scored below the NAEP Basic level in reading, the largest share since 2002 (National Assessment Governing Board, January 2025).

This guide explains the types of assessment, what formative assessment looks like and what research shows about it, how formative and summative assessment differ, what federal law requires of standardized testing and what NAEP is for, the evidence on testing both ways, opting out, standards-based grading, authentic assessment, assessment and AI, accommodations for students with disabilities and English learners, and using assessment data well. It is written for district and school leaders, assessment coordinators and teachers, with a section for families. It covers K–12 education in the United States. Each section links to edCircuit’s more detailed coverage.

Updated October 2026.

In this guide

• The basics: What formative assessment is · Types of assessment · Formative vs. summative assessment
• Evidence: What the research shows · Interim and benchmark assessment
• Standardized testing: What federal law requires · The evidence both ways · Opting out
• Grading and alternatives: Standards-based vs. traditional grading · Authentic and alternative assessment · Assessment and AI
• Fairness and use: Accommodations · Using assessment data well · Families · Questions to ask

What is formative assessment?

Formative assessment is assessment used to adjust teaching and learning while they are happening, as distinct from assessment used to report what has already been learned. What makes it formative is the use made of the evidence, not the kind of task. Black and Wiliam drew that line in 1998: assessment, they wrote, “becomes formative assessment when the evidence is actually used to adapt the teaching to meet student needs.” A quiz whose results are recorded and never acted on is not formative in that sense; a single well-chosen question that changes what a teacher does in the next five minutes is.

The working definition used by many state education agencies has changed over time. The Council of Chief State School Officers’ group on formative assessment, formed in 2006, first described it as a process used by teachers and students “during instruction” that provides feedback to adjust ongoing teaching and learning. In its 2018 revision, the group added that the process is planned and ongoing, that it is used by “all students” as well as teachers, that it concerns intended disciplinary learning outcomes, and that one of its purposes is to help students become self-directed learners (CCSSO, 2018). The revision treats students as users of the evidence, not only its source.

The most widely used description of what formative assessment involves comes from Dylan Wiliam and colleagues at Educational Testing Service, who worked with teachers in seven states in 2003 and 2004. Writing in ASCD’s Educational Leadership in November 2005, they set out five strategies that they called “nonnegotiable in that they define the territory of assessment for learning” (ASCD):

• clarifying and sharing learning intentions and criteria for success;
• engineering effective classroom discussions, questions and learning tasks;
• providing feedback that moves learners forward;
• activating students as the owners of their own learning; and
• activating students as instructional resources for one another.

The article described techniques that remain common: “hinge-point” questions that decide which way a lesson goes next; having every student answer at once, on mini whiteboards or with cards labeled A to D; calling on students at random rather than waiting for raised hands; and exit passes at the end of a lesson. On feedback, the authors cited research by Butler (1988) finding that students given a grade and a comment ignore the comment. They also stressed that teachers had to choose and adapt techniques for their own students, which makes formative assessment mainly a matter of teaching practice rather than of tests or software. edCircuit’s coverage of formative assessment in classrooms:

• 4 Strategies for Formative Assessment, which describes anticipation guides, Four Corners, graffiti walls and self-rating exit slips
• Designing Teaching and Understanding Learners with Dylan Wiliam
• 3 Ways to Use Exit Slips in Your Classroom
• Do We Actually Know What Formative Assessment Is?

What are the main types of assessment?

Schools commonly distinguish four types of assessment by purpose and timing: diagnostic assessment before or at the start of learning, formative assessment during it, interim or benchmark assessment at intervals through the year, and summative assessment at the end of a unit, course or year. The purpose, not the format, decides which type an assessment is.

The distinction that causes the most confusion is between formative and interim assessment. In a 2009 article in Educational Measurement: Issues and Practice, Marianne Perie, Scott Marion and Brian Gong observed that local assessment systems were being marketed as formative, benchmark and predictive, and that many products sold as formative assessments were in fact interim assessments, not the kinds of practice Black and Wiliam had studied. They described interim assessments as given on a larger scale than formative assessments, with less flexibility, and aggregated to the school or district level to inform policy (Perie, Marion and Gong, 2009).

The table is a reading aid rather than an official classification; the categories overlap, and the terms are used loosely by vendors and districts alike. Federal law blurs one line deliberately: the Every Student Succeeds Act lets a state give its annual test either as a single summative assessment or as multiple statewide interim assessments during the year that result in a single summative score (20 U.S.C. 6311(b)(2)(B)(viii)). For district leaders, the useful check is whether each assessment on the calendar has a stated purpose, a known user and a decision it informs. edCircuit’s articles on assessment design:

• Assessing Your Assessments
• 3 Keys to Testing and Assessment Success

Formative vs. summative assessment: what is the difference?

Formative assessment is used to improve learning while it is still happening; summative assessment is used to judge and report what has been learned at the end of a period. The two are often described as assessment for learning and assessment of learning. The difference lies in use, so the same task can be either: a unit test is summative when its score goes into the gradebook and the class moves on, and formative when the teacher uses the results to reteach and students revise before the unit closes.

Black and Wiliam warned in 1998 that high-stakes external tests “always dominate teaching and assessment” and give teachers poor models for formative work, because they summarize achievement rather than diagnose it. Wiliam and his ETS colleagues added in 2005 that tests given every six to ten weeks often arrive too late to change instruction for the students who took them. For edCircuit’s discussion of assessment for learning, see 5 Approaches to Improve Assessment and Promote Learning.

What does research show about formative assessment?

Research consistently finds that formative assessment has a positive effect on student achievement, but the more rigorous reviews put that effect at roughly 0.2 to 0.25 standard deviations, well below the 0.4 to 0.7 range often quoted from Black and Wiliam, and the quality of much of the evidence is low. The main sources, in order of publication:

• Black and Wiliam, 1998. For “Inside the Black Box,” published in Phi Delta Kappan in October 1998, the authors drew on 250 of about 580 sources they identified and reported that innovations strengthening formative assessment produced significant and often substantial gains, with typical effect sizes between 0.4 and 0.7, and that low achievers often gained more than others. They also wrote that they did not conclude formative assessment was “yet another ‘magic bullet’ for education,” because change in classroom practice comes slowly.
• Kingston and Nash, 2011. Of more than 300 K–12 studies reviewed, many had “severely flawed research designs” and only 13 yielded usable effect sizes (42 in all). The median effect was 0.25 and the weighted mean 0.20. Effects appeared larger in English language arts (0.32) than in mathematics (0.17) or science (0.09), and programs based on professional development (0.30) or computer-based formative systems (0.28) appeared more effective than other approaches. They concluded that the often-claimed effect of 0.40 to 0.70 “is not supported by the existing research base” (Educational Measurement: Issues and Practice, 2011).
• The response to Kingston and Nash, 2012. Derek Briggs, Maria Araceli Ruiz-Primo, Erin Furtak, Lorrie Shepard and Yue Yin questioned how Kingston and Nash selected and combined studies and concluded that “considerable uncertainty remains” about formative assessment’s effect on achievement (Briggs et al., 2012).
• Bennett, 2011. Randy Elliot Bennett of ETS argued that “formative assessment” does not yet name a well-defined set of practices and that the commonly quoted effect sizes derive from “untraceable, flawed, dated, or unpublished sources” (Assessment in Education, 2011).
• Regional Educational Laboratory Central, 2017. A federal review found 23 studies of elementary school formative assessment rigorous enough to support causal conclusions; 19 provided effect sizes. Formative assessment had a positive overall effect, with mean effects of 0.36 in mathematics (six studies), 0.22 in reading (seven) and 0.21 in writing (six). Effects were larger when a teacher or a computer program directed the process than when students did (Klute et al., REL 2017-259, February 2017).
• Reading, 2022. A meta-analysis of 48 studies with 116,051 K–12 students found a positive but modest effect on reading achievement (+0.19). Studies with 250 or fewer students produced significantly larger effects than larger studies (Xuan, Cheung and Sun, Frontiers in Psychology, August 2022).
• K–12, 2024. Yuankun Yao and colleagues analyzed 258 effect sizes from 118 studies worldwide and found an overall effect of 0.25, and 0.22 in U.S. studies. Effects were similar across types of formative assessment and subjects, and larger in high school and in quasi-experimental studies (Educational Research and Evaluation, 2024).
• An umbrella review, 2024. Andrew Sortwell and colleagues examined 13 meta-analyses of formative assessment in K–12. They found effects ranging from trivial to large and none negative, but rated the certainty of the evidence very low in nine of the meta-analyses, low in three and moderate in one, and concluded that this “precludes robust recommendations” about which formative assessment strategies work best (Sustainability, September 2024).

Read together, the sources show that formative assessment, done well, helps students learn and that no review found it harmful, but the benefit is smaller than the figures still widely quoted, depends heavily on how the practice is carried out, and tends to look larger in small studies. A claim that a product will deliver effects of 0.4 to 0.7 deserves caution. And because Kingston and Nash found larger effects for approaches built on professional development, and the federal review found larger effects when teachers directed the process, formative assessment is best treated as teacher skill developed over time rather than as a purchase.

What are interim and benchmark assessments, and do they help?

Interim and benchmark assessments are tests given at set points during the year, often in fall, winter and spring, to check whether students are on track for end-of-year standards, and rigorous studies of them have produced mixed results. They are widespread: in the CGCS study, the average student took about eight standardized tests a year, typically including three “formative” exams in each of two subjects (CGCS, 2015), tests Perie and colleagues would call interim.

Three randomized studies show the range of outcomes:

The studies point in the same practical direction. A benchmark test produces information; achievement changes only if teachers and leaders change instruction because of it. In the Illinois trial, classroom practice did not measurably change, and neither did reading scores. Districts adding or renewing a benchmark system can ask what decisions the results will drive and whether the tests duplicate what state and classroom assessments already provide. For edCircuit’s coverage of data-informed intervention, see 3 Critical Elements of an Effective MTSS Program and AI in Early Student Assessment and Intervention.

What does federal law require for standardized testing?

Federal law requires every state to test all public school students in reading or language arts and mathematics each year in grades 3 through 8 and once in high school, and in science once in each of three grade bands, and to take part in NAEP in reading and mathematics in grades 4 and 8. The requirements are in the Every Student Succeeds Act (ESSA) of 2015, at 20 U.S.C. 6311(b)(2):

• Grades and subjects. Mathematics and reading or language arts in each of grades 3 through 8 and at least once in grades 9 through 12; science not less than once in grades 3 through 5, 6 through 9 and 10 through 12. States may test other subjects at their discretion.
• Multiple measures. The tests must involve multiple up-to-date measures, including measures of higher-order thinking, and “may be partially delivered in the form of portfolios, projects, or extended performance tasks.”
• Format options. A state may use one summative test or several interim tests that produce a single summative score, and may use computer adaptive tests. A district may, with state approval, use a nationally recognized high school assessment in place of the state’s high school test.
• Reports to parents and teachers. The tests must produce individual student reports, including diagnostic information, delivered “as soon as is practicable” in an understandable format and, to the extent practicable, in a language parents can understand, and must allow results to be broken down by student group.
• Participation. States must annually measure the achievement of at least 95 percent of all students, and of each student group, on the reading and mathematics tests, and explain how that requirement factors into school accountability (20 U.S.C. 6311(c)(4)(E)).

ESSA also created an Innovative Assessment Demonstration Authority, which lets approved states try systems of competency-based, interim or performance-based assessments for up to five years, with no more than seven states in the initial period (20 U.S.C. 6364).

NAEP’s role. The National Assessment of Educational Progress, known as the Nation’s Report Card, is a different kind of test. It is given to random samples of students and reports results for the nation and states, not for individual students. Under its authorizing law, NAEP must assess reading and mathematics in grades 4 and 8 at least every two years; participation is voluntary for students, schools and districts; and federal use of NAEP to “rank, compare, or otherwise evaluate individual students or teachers,” or to reward or sanction them, is prohibited (20 U.S.C. 9622). ESSA requires each state receiving Title I funds to take part in the biennial grade 4 and 8 reading and mathematics assessments if the federal government pays the cost. NAEP therefore serves as a common yardstick for comparing states.

Recent federal actions affect both kinds of testing:

The department has said it will keep comparable statewide testing. Kirsten Baesler, assistant secretary for elementary and secondary education, told Education Week in March 2026: “We’re prohibited from waiving requirements that would harm transparency. It’s very clear that state assessments need to have a standard comparison.” All4Ed’s review of 2026 state bills found proposals to cut testing to the federal minimum or let districts choose among tests, all of which it listed as failed, and others that would use state results to place students in advanced courses or provide tutoring. District leaders should read their own state’s statute and any approved waiver directly; the summaries here are limited to the sources this guide examined.

Standardized testing: what does the evidence show, both ways?

Standardized testing tied to accountability has produced measurable gains in some subjects, especially mathematics, and has also narrowed what is taught and taken time from instruction; the evidence supports both claims. The main findings:

• Achievement gains. Thomas Dee and Brian Jacob compared NAEP trends in states that had school accountability before No Child Left Behind with those that did not. They found that the law produced statistically significant increases in fourth graders’ average mathematics performance (an effect size of 0.23 by 2007), evidence of improvement in eighth-grade mathematics, particularly among low-achieving groups, and no evidence of gains in fourth-grade reading (Journal of Policy Analysis and Management, 2011).
• Information on gaps. Because ESSA requires results by race and ethnicity, income, disability, English proficiency, gender and migrant status, state tests are the main source of comparable information on how different groups of students are served. In 2024 NAEP, the lowest-performing students scored about 100 points below the highest performers (NAGB).
• Narrowing of the curriculum. In a nationally representative survey of 349 districts in 2006–07, the Center on Education Policy found that 44 percent had added time for English language arts or mathematics at the expense of social studies, science, art and music, physical education, recess or lunch (CEP, as reported in Arts Education Policy Review, 2008).
• Time and volume. In the Council of the Great City Schools’ 2015 study of 66 urban districts, students were required to take an average of 112.3 tests from prekindergarten through grade 12; 401 different tests were given across the districts; mandated testing took about 4.22 days, or 2.34 percent of school time, in eighth grade, the heaviest year, not counting preparation. The council found no correlation between the amount of mandated testing time and districts’ NAEP reading and mathematics scores (CGCS, October 2015).
• Prediction. In college admissions, some evidence has moved toward tests. Explaining its decision to require the SAT or ACT again beginning with the Class of 2029, Dartmouth College said its researchers found standardized testing “among the most reliable indicators for success in Dartmouth’s curriculum, surpassing high school GPA,” and especially useful for identifying high-achieving applicants from low-income, first-generation and rural backgrounds (Dartmouth). That finding concerns one highly selective college and should not be generalized to all institutions.

The evidence does not settle the argument, because the two sides are often measuring different things. Supporters point to comparable information, attention to underserved groups and gains in mathematics; critics point to lost instructional time, narrowed curriculum and pressure on students and teachers. Much of the testing burden lies in the local layer federal law does not require: the CGCS study found that many tests originated at state and local levels, and that testing time depended as much on how often tests were given as on how many there were. edCircuit’s coverage of standardized testing:

• Navigating State Testing Amid Shifting Federal Winds
• Has Standardized Testing Gone Too Far?
• Beyond Standardized Testing
• Assessment Testing Is Outdated in Modern Education
• One Big Idea: Ray McNulty on Replacing Standardized Testing

Can families opt out of state tests?

Whether a family can opt a child out of state tests depends on state and local law; federal law neither grants nor removes that right, but it requires states to count participation. ESSA states that nothing in its assessment provisions preempts “a State or local law regarding the decision of a parent to not have the parent’s child participate in the academic assessments” (20 U.S.C. 6311(b)(2)(K)). At the same time, a state must measure at least 95 percent of students, and when participation falls below that level, the state must include in the denominator for achievement the greater of 95 percent of students or the number who tested. A school with many opt-outs can therefore see its reported proficiency rate fall.

Opt-out rates vary widely. The CGCS study found that rates in its member districts with data were typically below 1 percent in 2015, with noticeable exceptions. New York was one: Matthew Chingos, analyzing district data compiled by an advocacy group for the spring 2015 tests, estimated that 21 percent of students in 648 districts opted out, with rates above 50 percent in 13 percent of districts. Districts serving more disadvantaged students had lower opt-out rates, and, once income was taken into account, districts with lower test scores had higher rates (Education Next, June 2015). Chingos called the conclusions preliminary.

For families, opting out spares a child the test but also forgoes the individual score report ESSA requires, and widespread opt-out reduces what a school and state know about how groups of students are doing.

Standards-based grading vs. traditional grading: what is the difference, and which works better?

Traditional grading usually averages points or percentages from tests, assignments, homework and sometimes behavior into a single letter grade per course; standards-based grading reports a student’s proficiency on each of a set of learning standards, usually on a short scale, and often lets later evidence replace earlier attempts. The research on which approach serves students better is limited, and the results so far run in both directions.

The table describes typical practice rather than any one district’s policy. The evidence:

• What grades measure. A 2016 review of more than a century of research, by Susan Brookhart, Thomas Guskey and six colleagues, found that early 20th-century studies “generally condemned teachers’ grades as unreliable,” and that more recent studies suggest report card grades assess a multidimensional construct containing both cognitive and noncognitive factors reflecting what teachers value in student work (Review of Educational Research, 2016).
• A randomized trial. In a cluster randomized trial of a standards-based grading system called Proficiency-based Assessment and Reassessment of Learning Outcomes, involving 2,736 ninth graders in 14 treatment and 15 control schools, students in treatment schools scored 0.33 standard deviations higher on end-of-course algebra and geometry tests. The program did not change how much students valued mathematics or expected to succeed in it, and students who already had higher expectancies and value benefited more (Kramer et al., Journal of Research on Educational Effectiveness). The system combined grading changes with formative feedback and reassessment for full credit, so its effect cannot be credited to the grading scale alone.
• A contrary finding. A comparison of two similar Midwestern high schools found that grade point averages were unchanged and ACT scores “may be negatively impacted” where the school used standards-based grading (Townsley and Varga, 2018). The study compared two schools rather than assigning schools at random.
• Implementation. Laura Link and Thomas Guskey found “little consensus on what ‘standards-based grading’ actually means,” that implementation is widely inconsistent, and that confusion and resistance were leading some educators to abandon it; they argued that it can serve its main purpose, communicating students’ performance to students and parents, when clear criteria are met consistently (Link and Guskey, Theory Into Practice, 2022).

Families’ reliance on grades adds a practical concern. In a fall 2023 Learning Heroes and Gallup survey of nearly 2,000 K–12 public school parents, 79 percent said their child received mostly B’s or better and 88 percent said their child was at or above grade level in reading, figures the authors contrasted with test results suggesting fewer than half of students perform at grade level (Learning Heroes). Whatever grading system a district uses, the question is whether a grade tells a family accurately what a student has learned. edCircuit’s coverage of grading:

• Educators Taking a Critical Look at Letter Grades
• How Effective Graders Think
• What is the Impact of Grades?
• Assessment in the AI Era: Rethinking How We Grade

Authentic and alternative assessment: where do they fit?

Authentic assessment asks students to carry out tasks that resemble the real work of a discipline, such as conducting an investigation, writing and revising for a real audience or defending an argument, and alternative assessment is a broader term for any approach other than conventional tests, including portfolios, projects, presentations and performance tasks. Grant Wiggins, writing for the ERIC Clearinghouse in 1990, put the core idea this way: “Assessment is authentic when we directly examine student performance on worthy intellectual tasks.” He contrasted it with traditional tests that rely on “indirect or proxy ‘items,’” and noted that the choice is “not ‘either-or’”: conventional tests may be adequate for monitoring performance, while authentic tasks matter more for improving it.

Federal law leaves room for these approaches in state testing. ESSA allows state tests to be delivered partly through portfolios, projects or extended performance tasks, and its innovative assessment authority permits performance-based systems. The evidence on using them at scale is mixed:

• Reliability. In Vermont’s statewide portfolio program, an early example, rater agreement on the quality of student work in 1991–92 was low, with reliability coefficients of 0.33 to 0.43, “low enough to seriously limit the use” of individual results, although statewide averages were quite reliable because of the large numbers of students (Koretz et al., interim report, 1992). Teachers in the same evaluation reported positive effects on instruction (Koretz et al., 1994).
• Achievement. New Hampshire’s Performance Assessment of Competency Education (PACE) pilot received a federal waiver beginning in 2014–15 to use performance assessments in place of the state test in most required grades and subjects. A descriptive study of its first three years found that students in PACE schools tended to show small positive effects on grade 8 and 11 state tests in mathematics and English language arts compared with students in non-PACE schools, with small positive differential effects for lower-achieving students (Evans, Education Policy Analysis Archives, 2019). The design was not randomized.

The lessons for schools are practical. Performance tasks need clear rubrics, scorer training and checks on agreement between scorers if their results are to carry weight beyond the classroom, and they take more teacher time to design and score. edCircuit covers both approaches:

• The Rise and Impact of Authentic Assessment in Education
• Alternative Forms of Assessment in Education
• 10 Reasons Why You Should Implement Digital Student Portfolios

How is artificial intelligence changing assessment?

Artificial intelligence is affecting assessment in two directions: tools that score student work or give feedback promise faster formative assessment, while generative AI makes it harder to know whether take-home work reflects a student’s own thinking. Federal guidance and research so far counsel caution on both.

On AI as an assessment tool, the U.S. Department of Education’s Office of Educational Technology, in its May 2023 report Artificial Intelligence and the Future of Teaching and Learning, identified formative assessment as a possible leading area for AI in schools because of the importance of fast feedback. It used automated essay scoring as an example of both promise and limits: such systems can analyze some features of essays but can be misled by an essay’s length or by appropriate keywords placed in sentences that make no sense. The report concluded that “AI may support but not replace high-quality, human-led processes and practices of formative assessment in schools.”

On detecting AI-written work, a 2023 study by Weixin Liang and four colleagues found that several widely used detectors “consistently misclassify non-native English writing samples as AI-generated,” while correctly identifying native writing, and that simple prompting strategies could bypass the detectors. The authors cautioned against using such detectors in evaluative or educational settings (Liang et al., 2023). For schools, the implication is that a detector score should not by itself be the basis for an accusation, especially against English learners.

edCircuit’s coverage describes schools responding by changing assessment design rather than relying on detection: more work begun in class, drafts and checkpoints that show how a piece developed, short conversations in which students explain their reasoning, and clear statements of when AI use is allowed. Those approaches are themselves formative assessment practices, which is one reason the two topics now overlap. edCircuit’s coverage of AI and assessment:

• AI Detection in Schools: Rethinking Assessment
• AI Assessments in Schools: Faster Feedback, Better Learning, Smarter Classrooms
• Homework in the AI Era: Why Schools Are Rethinking It
• AI in Education: A Guide for Schools and Colleges

What accommodations do students with disabilities and English learners receive on tests?

Students with disabilities and English learners must be included in state tests, with accommodations that let them show what they know without changing what the test measures; a small group of students with the most significant cognitive disabilities may take alternate assessments.

Students with disabilities. Federal special education regulations require states to include all children with disabilities in state and district-wide assessments, with accommodations or alternate assessments as set out in each child’s individualized education program (IEP). State guidelines must identify only accommodations that “do not invalidate the score,” and IEP teams must choose only from those (34 CFR 300.160). ESSA requires accommodations, including assistive technology, for students with disabilities under IDEA and for students with disabilities who receive accommodations under other federal laws. For students with the most significant cognitive disabilities, states may use alternate assessments aligned with alternate achievement standards, but the number assessed this way may not exceed 1 percent of all students tested in a subject statewide. Parents must be told how taking the alternate test may affect completing a regular diploma (20 U.S.C. 6311(b)(2)(D)).

English learners. ESSA requires that English learners be assessed in a valid and reliable manner with appropriate accommodations, including, to the extent practicable, tests in the language and form most likely to yield accurate data, until they reach English proficiency. A student who has attended U.S. schools for three or more consecutive years must be tested in reading or language arts in English, although a district may decide case by case to test in another language for up to two more years. States must also assess the English proficiency of all English learners every year, and have some flexibility for recently arrived English learners enrolled for less than 12 months (20 U.S.C. 6311(b)(2)–(3)).

What research shows. A 2005 review of studies on test accommodations by Stephen Sireci, Stanley Scarpati and Shuhong Li found two consistent findings: extended time tended to improve the performance of all students, with students with disabilities gaining relatively more, and reading math tests aloud was associated with higher scores for some students with disabilities (Review of Educational Research, 2005). In a randomized study of Texas third graders, an accommodated reading test given in two sessions, with proper nouns and question stems read aloud, raised scores only for students with decoding difficulties, increasing their odds of passing sevenfold (Fletcher et al., Exceptional Children, 2006). The findings support choosing accommodations for each student’s specific need rather than applying them uniformly. For more on IEPs, 504 plans and parents’ rights, see edCircuit’s Special Education Guide.

How can schools use assessment data well?

Schools use assessment data well when they treat it as part of a regular cycle of looking at evidence, changing instruction and checking again; the research base for specific data-use practices, however, is thin. The federal practice guide Using Student Achievement Data to Support Instructional Decision Making makes five recommendations: make data part of an ongoing cycle of instructional improvement; teach students to examine their own data and set learning goals; establish a clear vision for schoolwide data use; provide supports that foster a data-driven culture; and develop and maintain a districtwide data system. The panel rated the evidence for all five as low, meaning it found no body of research showing effects on achievement, while stressing that this did not make the recommendations unimportant (Hamilton et al., IES, NCEE 2009-4067).

How often teachers use different kinds of data varies. In a 2021 study of 353 Nebraska schools by the Regional Educational Laboratory Central, 29 percent of teachers reported not using summative data, 21 percent not using interim data and 6 percent not using formative data. Those who used them drew on formative data almost weekly and interim data about monthly (Wilkerson et al., REL 2021-054).

The studies in this guide suggest several practical rules:

• Match the data to the decision. State results suit program and resource decisions; benchmark results suit regrouping and checking progress; classroom formative evidence suits tomorrow’s lesson. Using state scores to plan daily instruction asks too much of them.
• Close the loop. The Illinois benchmark trial shows that data delivered without changed instruction changes little. Data meetings should end with a specific instructional decision and a date to check its result.
• Look below the averages. ESSA requires results by student group and itemized score analyses so that teachers can see which skills and which students need attention.

edCircuit’s coverage of data and accountability includes District Leaders Working to Change Accountability Measures and AI and Achievement Gaps in Education.

What should families know about assessment and testing?

Families get the most from school assessment when they look at more than one measure, ask what each one is for, and talk with teachers about what the results mean for their child. Practical steps:

• Compare grades with other measures. As the 2023 Learning Heroes survey shows, report cards alone can mislead. Families can compare report card grades with state test and benchmark results and ask the teacher directly whether their child is working at grade level.
• Read the state score report. Federal law requires schools to provide individual reports on state tests, in an understandable format and, where practicable, in a language parents understand. Families can ask for it if it does not arrive.
• Know the rules on opting out. Whether a family may opt out of state tests depends on state and local law, and opting out also means losing the individual report.
• Ask about accommodations. For a child with an IEP or 504 plan, or an English learner, families can ask which testing accommodations the child will receive and how they were chosen.
• Keep tests in proportion. A single test is one piece of evidence. edCircuit’s The Role of Teachers and Parents in Helping Students Cope with Test Anxiety and How Teachers Can Prepare for End of Year Parent-Teacher Conferences are written with families as well as educators in mind.

Questions to ask about assessment in your schools

For school and district leaders and assessment coordinators

• Do we have an inventory of every assessment our students take, with its purpose, the decision it informs and the time it takes, and which could be dropped?
• Which of the products we call “formative” are actually interim assessments, and are we using each for the purpose it was designed for?
• When benchmark results arrive, what specifically changes in instruction, and how do we check whether it did?
• Are we supporting formative assessment as a teaching practice through sustained professional learning, rather than as a product purchase?
• Do our grades tell families accurately whether their children are working at grade level, and do grades and test results tell the same story?
• If we are moving to standards-based grading, have we defined it precisely, piloted it and explained it to families and colleges?
• Are testing accommodations for students with disabilities and English learners chosen for each student’s needs, and do they avoid invalidating scores?

For teachers

• Do my students know what they are learning and what good work looks like before they begin?
• Do I plan questions that reveal what every student is thinking, not only those who raise their hands?
• Does my feedback tell students what to do next, and do I give time in class to act on it?
• Do I change what I teach tomorrow because of what I learned today?
• Do my grades reflect what students have learned, kept separate from behavior and work habits?

Assessment that serves learning

Assessment does many jobs in schools, and the evidence supports each when the assessment is matched to its job and its results are used. Formative assessment helps students learn, though by less than often claimed and only when it changes teaching; benchmark tests help only when they lead to different instruction; state tests give comparable information on how groups of students are served, at a real cost in time and curriculum; and grades work when they tell families accurately what has been learned. The districts most likely to get value from assessment are those that know why each test is given, give teachers the skill and time to act on everyday evidence, keep accommodations fair, and treat no single number as the whole account of a student.

Related reading

• 4 Strategies for Formative Assessment
• Designing Teaching and Understanding Learners with Dylan Wiliam
• Navigating State Testing Amid Shifting Federal Winds
• The Rise and Impact of Authentic Assessment in Education
• Special Education Guide: IEPs, 504 Plans, IDEA and Parents’ Rights

Subscribe to edCircuit to stay up to date on all of our shows, podcasts, news, and thought leadership articles.

More on Assessment and Testing