Assessment and Data Use
Domain 4 (Assessment and Data Use) tests measurement concepts, appraisal, and the use of data to support student outcomes. Items ask about reliability and validity, score interpretation, appropriate test use, and data-driven program decisions.
Measurement essentials
Reliability is consistency: test-retest, internal consistency, and inter-rater reliability. Validity is the evidence that a test measures what it claims: content, construct, and criterion validity. The standard error of measurement estimates score fluctuation and should temper high-stakes decisions. Norm-referenced tests compare students to a norm group; criterion-referenced tests compare performance to a standard. Standard scores include z-scores (mean 0, SD 1), T-scores (mean 50, SD 10), stanines (1–9), NCEs (mean 50, SD 21.06), and percentile ranks. Grade-equivalent scores are easily misread and should not be interpreted as grade placement.
Use accommodations to remove barriers without changing what is measured; modifications change expectations. Test bias, language load, and cultural fairness require scrutiny. Only qualified personnel administer and interpret certain instruments, with informed consent and confidentiality.
Data types and program use
ASCA distinguishes process data (what was done), perception data (what stakeholders know or believe), and outcome data (what changed). A school data profile summarizes enrollment, attendance, discipline, grades, course-taking, and graduation. Disaggregation by subgroup reveals equity gaps. Pre/post comparisons, results reports, and program audits turn data into improvement.
Florida context
Florida's assessment system includes the FAST in ELA and mathematics, statewide science assessments, and end-of-course assessments such as Algebra 1, Geometry, Biology, Civics, and U.S. History. Counselors also interpret PSAT/SAT/ACT, AP, dual enrollment, and industry certification results. Multiple measures-not a single score-should inform placement, promotion, and counseling recommendations.
Worked example
A reading test has a mean of 60 and SD of 10. A student scores 70. The counselor computes z = (70 − 60) / 10 = +1.0, which corresponds to a T-score of 60, a percentile rank near 84, and stanine 7. The reasoning: the score is one standard deviation above the mean; percentile rank is not percent correct, and the counselor reports the confidence band rather than treating 70 as an exact true score.
Traps
Reading percentile rank as percent correct; treating grade equivalents as placement; confusing reliability with validity; inferring causation from correlation; basing high-stakes decisions on one score; and ignoring the standard error of measurement.