Assessment and Intervention: Scores, Decisions, and Evidence
Scores and their uncertainty
Know the metric system cold: most cognitive and achievement composites use mean 100, SD 15; subtests mean 10, SD 3; T-scores mean 50, SD 10. A standard score is an interval, not a point-report the confidence band built from the SEM. Reliability caps validity, and the standard error of the difference (not of measurement) tells you whether two scores truly diverge. Percentile ranks are not percent correct, and grade equivalents overinterpret small differences at the tails.
Screening and classification logic
Sensitivity is the proportion of true cases correctly flagged; specificity is the proportion of non-cases correctly cleared. Positive predictive value depends heavily on base rate, so a good screener in a low-prevalence population still yields many false positives-hence multiple-gating in MTSS.
Behavior and intervention
Functional behavior assessment identifies antecedents, consequences, and the function (attention, escape, access to tangibles, sensory/automatic) maintaining behavior; a BIP that ignores function will fail. Strong academic interventions are explicit, systematic, scaffolded, and paired with frequent practice and feedback; progress monitoring with CBM plus a decision rule (for example, four consecutive points below the goal line, or a trend flatter than the goal slope) drives modification.
Traps
- Calling a 6-point subtest difference significant without checking the difference score's standard error.
- Confusing a behavior's topography (what it looks like) with its function (why it works).
- Writing goals that lack a baseline or measurable criterion.
- Treating Tier 3 as synonymous with special education eligibility.
- Changing an intervention after a single progress-monitoring point.
Rule of thumb
Ask two questions before interpreting any score: How much error surrounds it? and What is the base rate here? Both shift the conclusion more than the number itself.