Analyzing Give-N data by knower-level yields greater statistical power than analyzing by percent correct, plausibly explaining the discrepancy with Ansari et al. (2003).¶

Claim · All claims
Evidence · 1 study · associational r2 · q2
Currency · newest study 2012 · not yet checked against its sources · not reviewed by a person

Subclaims¶

q2 i? In a 10,000-iteration simulation, knower-level analysis detected a significant correlation about 12% more often than percent-correct analysis of the same data. → Negen 2012

Evidence¶

Negen 2012¶

Negen, J., & Sarnecka, B. W. (2012). Number-Concept Acquisition and General Vocabulary Development. Child Development. https://doi.org/10.1111/j.1467-8624.2012.01815.x

q2 · i? · associational · r2

Computer simulation (Experiment 3) using the Lee & Sarnecka generative model to create Give-N data for 28 simulated children, repeated 10,000 times; both observed correlations (this study's and Ansari et al.'s) fell within the resulting 95% sampling-distribution interval.

"Analyzing Give-N data by inferred knower-level was more powerful than analyzing the same data by percent correct, detecting a significant correlation about 12% more often. It also tended to produce larger sample correlations."

Discussion¶

Related Claims¶

• The Excel sheet's knower-level inference closely matches fully Bayesian inference, with no discrepancies in posterior maxima — related
• Knower levels were longitudinally stable across sessions, replicating Wynn, while the CP-Non-Mapper/CP-Mapper distinction fluctuated across sessions — related
• Fully Bayesian inference on the calibration data set assigned 423 sessions across six knower levels and estimated an evidence value v of 16.94 — related
• Nonintervention (observational) studies in educational psychology are producing recommendations for practice at increasing rates — related
• The primary study (Lehmann et al., 2016) had only about 45% power to detect its observed interaction effect; roughly 150 participants would be needed for 80% power — related