Assessment
What the questions actually test
Assessment items reward precision about technical vocabulary and about purpose.
Validity and reliability. Validity asks whether an instrument measures what it claims for a given use; reliability asks whether it produces consistent results. A test can be reliable but invalid (a scale that consistently reads five pounds heavy). A test cannot be valid without being reliable. Fairness and freedom from bias are part of validity: an item that assumes background knowledge some students lack measures background, not the target skill.
Purposes. Formative assessment happens during learning and informs instruction; summative assessment happens after and certifies learning. A common trap describes a quiz given mid-unit, then calls it summative because it was graded. Feedback quality matters more than feedback quantity: specific, timely, actionable, and focused on the task rather than the student's character.
Types and scoring. Norm-referenced tests compare students to each other; criterion-referenced tests compare performance to a defined standard. A percentile rank of 70 means the student scored at or above 70 percent of the norm group, not that the student answered 70 percent correctly. Rubrics clarify criteria but only if they are shared before the work begins and describe observable quality. Portfolios and performance tasks support authentic assessment of complex outcomes.
Accommodations and modifications. Accommodations change how a student demonstrates learning (extended time, read-aloud, separate setting) without changing what is measured. Modifications change the content or standard. Mixing these up is a frequent trap.
Worked example
A teacher worries a unit test was unfair because several students who usually perform well did poorly. The defensible first step is to examine the items for ambiguity and construct-irrelevant difficulty, that is, to question validity, before deciding students did not study.