1.3 Assessment Access and Bias

What Counts As Valid Evidence

An assessment is useful only if it supports the inference the teacher wants to make. Validity asks whether the task measures the intended skill. A social studies test full of idioms may not validly measure historical reasoning for a student who understands the topic but misses the figurative language. A math quiz with tiny print may measure visual access as much as proportional reasoning. Reliability, by contrast, asks whether scores are consistent; an unreliable rubric produces different scores for the same work. EAS items usually hinge on validity, not reliability.

When something unrelated to the target skill makes a task harder for some students, that is construct-irrelevant variance, and it is the central vocabulary of this competency. EAS questions often describe a mismatch between what students do in one setting and what the score suggests. The best answer neither ignores the score nor treats one score as the whole truth; it examines whether the task, timing, format, or rubric blocked students from showing what they know.

Multiple Sources of Evidence

One assessment can be wrong, incomplete, or too narrow. Teachers should triangulate three purposes:

• Diagnostic (before instruction): shows starting points and prior knowledge.
• Formative (during instruction): low-stakes checks that guide reteaching and grouping, such as exit tickets and conferences.
• Summative (after instruction): judges achievement against the standard.

Evidence can include discussion notes, exit tickets, short writes, conferences, performance tasks, quizzes, projects, observations, self-assessments, and work samples. The goal is not to discard formal tests but to compare evidence so the teacher can make a better instructional decision.

Rubric Alignment

A rubric should match the learning objective. If the objective is to explain causes of a historical event with evidence, most points should reward claim, evidence, reasoning, and accuracy. Conventions may matter, but they should not outweigh historical thinking unless writing conventions are a stated target.

Strong rubrics use observable language so students know what proficient work looks like before they start. Teachers can model scoring with an anonymous sample, let students compare work to criteria, and use the rubric for feedback rather than only a final grade.

Bias Checks

Assessment bias occurs when a task includes construct-irrelevant barriers that make it easier for some students and harder for others. Bias can hide in language, examples, names, images, assumed experiences, response format, technology access, or scoring expectations.

Reduce bias by asking practical questions: Are directions direct? Is untaught background knowledge required? Do examples stereotype or exclude? Can every student access the materials and technology? Do scoring patterns penalize one group for behavior, language variety, speed, or presentation rather than the target skill?

Bias review does not mean lowering the standard; it protects the standard from noise. If the task is science reasoning, unnecessary idioms should be removed. If the task is oral presentation, the rubric can score content, organization, and delivery, but it must not punish an accent or home dialect as though it were a lack of knowledge.

EAS Decision Pattern

When a scenario involves an assessment concern, use this order:

1. Name the intended learning target.
2. Ask whether the assessment directly measures that target (validity).
3. Look for construct-irrelevant barriers.
4. Gather more than one source of evidence.
5. Revise instruction, support, rubric language, or assessment format based on what the evidence shows.

What To Avoid

Avoid answers claiming fairness always means identical timing, identical format, or a single score. Also avoid discarding all assessments because one task had problems. EAS favors professional judgment: use data carefully, question weak inferences, and make the next assessment more aligned.

A Worked Scenario

A seventh-grade history class takes a unit test on causes of westward expansion. The class average is strong, but every student classified as an English language learner scores far below their daily performance. Before re-teaching content, the teacher audits the test and finds the items lean on idioms ("manifest destiny was the wind in their sails"), dense compound sentences, and a U.S.-geography assumption never taught in the unit. These are textbook examples of construct-irrelevant variance: the items measure reading-the-idiom and prior geography rather than historical causation.

The validity-protecting response is to rewrite future items in plain language, add a labeled map so geography is provided rather than assumed, and gather a second source of evidence such as a short oral or graphic-organizer task. The teacher does not lower the standard or strip historical vocabulary; instead, the noise is removed so the score finally reflects historical reasoning. On the EAS, the credited answer is almost always "investigate the source of the gap and triangulate," not "reteach," "re-test in the same format," or "trust the single score."

Quick Takeaways

• Validity is about measuring the right skill, not producing a neat score.
• Construct-irrelevant variance is the barrier to remove.
• Diagnostic, formative, and summative evidence prevent overreaction to one task.
• Rubrics should match the objective and be shared clearly.
• Bias checks remove irrelevant barriers while keeping rigor.