Using Above-Level Testing to Track Growth in Academic Achievement in Gifted Students
Warne RT · 2014
grade Clongitudinalindependentunreplicated
Sample
224 middle-school students, 435 total test administrations
Population
US middle-school students in a gifted programme, tested above level on the Iowa Tests of Basic Skills and Iowa Tests of Educational Development
Design
Longitudinal hierarchical linear modelling of repeated above-level administrations - the first attempt to subject above-level testing to psychometric scrutiny rather than assume it. Graded C. Two limits the archive records honestly: the sample is one gifted programme rather than a representative population, and the specific quantity this topic most wants - how much a grade-level test understates growth for a high scorer - is in the full text, which was not obtained. That number is recorded debt.
Key findings
The paper's own framing is the finding that matters most here: "Although widely accepted in gifted education, above-level testing has not been subject to careful psychometric scrutiny." A practice this archive has good theoretical reason to recommend has almost no empirical validation behind it, and this is the study that says so. What the study does report is substantial differences between students from overrepresented ethnicities (White and Asian American) and underrepresented ones (Hispanic and African American) in BOTH initial above-level scores and rate of score gain; sex differences only in the rate of reading gain; and socioeconomic differences that had no unique effect beyond ethnicity. Those group differences are a caution about using above-level testing as a screening gate, which is exactly how talent searches use it.
Genetic confound
Medium. Growth trajectories in an already selected gifted sample, with no genetically informative design; group differences in intercept and slope are reported descriptively and the paper does not attempt to attribute them.
Replication notes
No identified replication. Warne (2012) documents that this is close to the only modern psychometric work on above-level achievement testing at all.
Effects
| Outcome | Metric | Value | Measure | Timing | Vs | Horizon | Class |
|---|---|---|---|---|---|---|---|
| Psychometric validation status of above-level testing | assessment | widely accepted in practice but "not been subject to careful psychometric scrutiny" | standardized | not applicable | none | not-applicable | domain-skill |
| Group differences in above-level score intercepts and growth rates | HLM parameters | substantial differences by ethnicity in both initial scores and rate of gain; sex differences only in reading growth rate; SES not unique beyond ethnicity | standardized | repeated administrations across middle school | none | over-2yr | domain-skill |
Cited by
- Placement and mastery diagnosis — deciding what to teach next from evidence of current skillmoderate supportconf: mediumgc: medium