Assessment and Testing
Assessment items test psychometric reasoning: what a score means, how much error surrounds it, whether a test is appropriate, and who may see results.
Reliability and validity
Reliability is consistency: test-retest, alternate forms, internal consistency (Cronbach's alpha, split-half), and inter-rater agreement. The standard error of measurement is SEM = SD × √(1 − r). Validity is accuracy of inference: content, construct, and criterion-related (concurrent and predictive). Face validity is appearance, not evidence. A test can be reliable but invalid; it cannot be valid without being reliable.
Scores and norms
Percentile ranks show position in a norm group, not percent correct. Standard scores include z (mean 0, SD 1), T (mean 50, SD 10), stanines (1–9), and IQ-type scores (mean 100, SD 15). Norms matter: a score means little without a representative comparison group. Grade equivalents overstate precision and are easily misread.
Instruments
WAIS/WISC for cognitive ability; MMPI-3 and NEO-PI-3 for personality; MCMI for clinical syndromes; Rorschach and TAT as projective measures; behavior rating scales and achievement tests for school and clinical decisions. Select using the Mental Measurements Yearbook and Tests in Print.
Ethics
ACA Code Section E requires competence, informed consent, attention to cultural bias, appropriate accommodations, and careful release of results. Clients generally receive explanations and their own results; raw data go to qualified professionals.
Worked example
A test has SD = 10 and reliability r = .80. SEM = 10 × √.20 ≈ 4.5. The reasoning: a single score is a band, roughly ±4.5 points at one SEM, so a 2-point difference between two administrations is noise, not change.
Traps that cost marks
Equating reliability with validity, overinterpreting grade equivalents, ignoring SEM, using tests with unrepresentative norms, and releasing raw data to unqualified recipients.