Assessment and Data-Based Decision Making
Assessment questions test whether you can choose measures, interpret scores correctly, and turn data into decisions.
Scores and psychometrics
- Standard scores mean 100, SD 15; T-scores 50/10; scaled scores 10/3; z-scores 0/1.
- Reliability (test-retest, internal consistency, inter-rater) caps validity. The SEM drives confidence intervals; a 95% interval is roughly ±1.96 SEM.
- Validity: content, construct, criterion (concurrent, predictive), and consequential validity for high-stakes decisions.
- Common tools: WISC-V (FSIQ, VCI, VSI, FRI, WMI, PSI), WJ IV, WIAT-4, KTEA-3, BASC-3, Conners-4, Vineland-3, ABAS-3.
- SLD models: ability–achievement discrepancy, RTI/dual discrepancy, and patterns of strengths and weaknesses. IDEIA 2004 permits RTI and bars states from requiring discrepancy.
Progress monitoring and FBA
CBM probes are brief, standardized, and repeated (e.g., words read correct per minute). Slope = (last − first) ÷ weeks. Compare slope to the aimline and to normative growth, and use decision rules (e.g., four consecutive points below the aimline) before changing instruction. FBA defines behavior operationally, uses ABC recording, and produces a functional hypothesis (attention, escape, tangible, sensory) that drives the BIP.
Worked example
Fall CBM = 42 wrcm; after 8 weeks, 58. Slope = (58 − 42) ÷ 8 = 2.0 wrcm/week, against benchmark growth of about 1.0–1.5. The student remains below benchmark but is closing the gap, so keep the intervention and raise the goal; changing it would be an error.
Traps
- Reading subtest scatter as a processing disorder without base-rate data.
- Calling a confidence interval a child characteristic rather than score precision.
- Judging an intervention from one data point.