LJHSCE flashcards

32 cards. Read the term and the definition here. The deck at the top of this page is optional.

Cut score

The minimum score required to pass a criterion-referenced exam. It is set by expert judgment methods such as modified Angoff, not by a fixed failure rate.

Modified Angoff method

A standard-setting procedure in which experts estimate the probability that a minimally competent candidate answers each item correctly; the summed estimates across items define the recommended cut score.

Item p-value

The proportion of examinees answering an item correctly. Values near 1.0 mean the item is easy; values near 0.0 mean it is hard.

Point-biserial discrimination

The correlation between performance on one item (right or wrong) and total test score. Strong positive values mean high scorers got the item right; negative values flag a possible key error.

Classical test theory model

Observed score equals true score plus error (X = T + E). Reliability is the ratio of true-score variance to observed-score variance.

Standard error of measurement

The standard deviation of observed scores multiplied by the square root of one minus reliability. It estimates the spread of scores a single candidate would obtain across repeated testing.

Cronbach's alpha

An internal-consistency reliability estimate based on item variances and total variance. It is a lower bound on reliability when items are not essentially tau-equivalent.

Norm versus criterion referencing

Norm-referenced scores compare a candidate to a comparison group; criterion-referenced scores compare performance to a defined standard. The two interpretations answer different questions and are not interchangeable.

Single-best-answer item

An item format with one defensible correct option and several plausible distractors. Each distractor should attract candidates holding a specific, identifiable misconception.

Distractor analysis

A review of the proportion of candidates selecting each incorrect option. A distractor chosen by almost no one provides no information and should be revised.

Negative stem

An item stem requiring the candidate to identify what is false or not true. Negative wording raises misreading rates and should be used sparingly and marked clearly.

Absolute qualifiers

Words such as always, never, and only that make a statement absolute. They let candidates eliminate options by test-wiseness rather than content knowledge.

Construct-irrelevant content

Material in an item that affects difficulty but is not part of the knowledge being measured, such as obscure idioms or local references. It is a fairness defect.

Item key verification

An independent second review confirming that the marked correct answer is actually correct. A wrong key is the most damaging item defect because it penalizes the best-prepared candidates.

Grammatical parallelism

A rule that all options in an item share the same grammatical form and similar length. Violations create unintended cues, since the longest option is often the key.

Item-writing blueprint alignment

The requirement that each item map to a specific objective in the published blueprint, so that form assembly can match the intended content weighting.

Raw versus scaled score

A raw score is the count of correct answers; a scaled score is a transformation used to equate forms and report on a stable metric across administrations.

Test equating

A statistical procedure that adjusts for difficulty differences between forms so scores are comparable. Common designs use common anchor items or randomly equivalent groups.

Standard error of difference

The standard error of measurement multiplied by the square root of two. Score differences smaller than this band are not meaningfully interpretable.

Percentile rank

The percentage of the norm group scoring at or below a given score. Percentile ranks are ordinal, so averaging or subtracting them is statistically invalid.

T-score

A standardized score with a mean of 50 and a standard deviation of 10. It is a linear transformation of a z-score, which has mean 0 and standard deviation 1.

Stanine

A nine-band standardized scale with a mean of 5 and a standard deviation of about 2. Stanines are coarse, so small differences within a band are not reported.

Reliability in speeded tests

Internal consistency underestimates reliability in speeded tests because unreached items reflect time limits rather than knowledge. Alternate-form or test-retest designs are preferred.

Cohen's kappa

A chance-corrected agreement statistic for two raters scoring constructed responses. It removes agreement expected by chance, unlike a simple percentage-agreement figure.

Test security plan

Documented controls covering item exposure, secure transmission, access logging, and response to suspected compromise. Live-item exposure is the primary threat to bank value.

Accommodation versus modification

An accommodation changes administration conditions without changing what is measured; a modification changes the construct itself and must be reported as a nonstandard administration.

Irregularity report

A record of an administration event such as a late start, power loss, or suspected cheating. It is adjudicated before scores are released.

Pretest item

An unscored item embedded in a live form to gather difficulty and discrimination data. It must be excluded from candidate scores and identified in the answer key.

Chain of custody

A signed transfer log for secure test materials at each handoff. An unbroken record is the evidence that compromise did not occur.

Practice analysis

A survey of the target role that defines the content domain and its weighting. The exam blueprint is derived from it rather than from the items already available.

Recertification requirement

The continuing education or periodic assessment a credential holder must complete to remain certified. The interval and units must be published in advance.

Conflict-of-interest screening

A governance control that excludes subject-matter experts with competing interests, such as those who train candidates for the exam, from key-writing and standard-setting roles.