IAAI-CI study materials
IAAI-CI study resources

IAAI-CI: Training Evaluation, Testing, and Records

IAAI-CI · Work through your exam, one topic at a time.

All guide sections

Training Evaluation, Testing, and Records

This domain tests how you judge whether training worked and how you prove it later. Expect items on evaluation levels, test construction, item quality, remediation, test security, and the records an instructor must create and retain.

Kirkpatrick's four levels

Reaction (did learners find it useful), learning (did knowledge or skill change), behavior (did on-the-job performance change), results (did organizational outcomes improve). Reaction is the easiest and weakest evidence; a high smile-sheet score does not show competence. Plan evaluation when you write objectives, not after the class.

Validity and reliability

A test is valid when it measures the objective it claims to measure - items must match the objective's behavior and conditions. It is reliable when it produces consistent scores across administrations and raters. A test can be reliable and still invalid, consistently measuring the wrong thing. For performance tests, use a checklist with observable criteria and calibrate raters.

Item writing

One best answer, plausible distractors, no "all of the above," avoid negatives, avoid grammatical clues, and avoid trivia. Keep the stem short and put the detail in the options.

Records

Document attendance, objective-by-objective performance, scores, remediation, and instructor qualifications. Retain per your authority's policy and any accrediting body's requirement; protect learner privacy and secure test banks so items are not circulated.

Worked example

An item reads: "Which of the following is NOT a factor in origin determination?" The keyed answer is ventilation, but two other options are arguable. Reasoning: negative stems invite double negatives and ambiguity, and arguable distractors destroy validity. Rewrite positively: "Which factor most directly indicates the area of origin?" with one defensible best answer and three clearly wrong options.

Traps

Treating reaction data as proof of competence; confusing validity with reliability; keeping a flawed item because it "always worked"; failing to document remediation; leaving answer keys accessible.