Linear regression outperformed equipercentile and mean-sigma equating for predicting ARM scores and was selected as the final linking model¶

Claim · All claims
Evidence · 1 study · associational r2 · q2
Currency · newest study 2024 · not yet checked against its sources · not reviewed by a person · recheck by 2027: tool and model accuracy changes quickly

Subclaims¶

q2 i? Among three linking methods compared by RMSE on the common student sample, linear regression showed the lowest error for grades 2–5 in both terms, though equipercentile was slightly better for grade 1 fall. → NWEA Psychometrics and Analytics 2024

Evidence¶

NWEA Psychometrics and Analytics 2024¶

NWEA Psychometrics and Analytics. (2024). Predicting Amira Reading Mastery Based on NWEA MAP Reading Fluency Benchmark Assessment Scores. https://www.nwea.org/research/publication/predicting-amira-reading-mastery-based-on-nwea-map-reading-fluency-benchmark-assessment-scores/

q2 · i? · associational · r2

Model comparison on the common student sample using Root-Mean-Square Error (Table 3.7); e.g., grade 3 winter RMSE 0.35 for regression versus 0.39 equipercentile and 0.37 mean-sigma. The report notes differences in performance are minimal across models.

"Although the equipercentile method showed slightly better performance for grade 1 data, the linear regression model consistently outperformed all other models for the remaining grades and was therefore selected as the final prediction model."

Discussion¶

Related Claims¶

• EISA-based MAP Growth testing produced slightly lower fall Mathematics scores but higher winter and spring Mathematics scores than traditional MAP Growth, and equipercentile linking was used to align the two versions for norming — related
• Regression analysis shows the MTCS method significantly degrades dilation-parameter precision relative to the Ratio of Eigenvalues method, while simulation factors explain most RMSE variation — related
• Least Squares procedures consistently outperform the MTCS method for estimating MIRT translation parameters m1 and m2 — related
• The developed MIRT equating methods behave as unbiased, effective, and consistent estimators of transformation parameters — related
• The Ratio of Trace method consistently yields the most precise estimates of the MIRT dilation parameter k across all simulated equating situations — related
• A linear regression on skill variables (n, dim, pc) predicts the minimum RSS value for BKT-BF training with high predictive ability — related
• MAP Reading Fluency SWCPM scores correlate highly with Amira Reading Mastery scores (r = 0.90–0.96) across grades 1–5 in fall and winter — related
• BFpack handles common statistical analyses with default priors and data missing at random — related
• Equated WCPM scores generally correlate more highly with MAP Growth Reading RIT scores than raw WCPM scores, with one exception — related
• Equated WCPM scores from non-reference passages show smaller residuals against reference-passage scores than raw scores, with one passage exception — related
• SWCPM score ranges corresponding to ARM performance levels show monotonic progression across terms within grades, except in first grade — related
• MAP Reading Fluency SWCPM scores classify students into ARM proficiency categories with exact match rates of 74.6–90.8%, kappa 0.60–0.85, and overall accuracy 0.81–0.96 — related