A one-factor unidimensional model is the most reasonable fit for MAP computerized adaptive Reading and Mathematics tests across grades, terms, and ten states, though Michigan Mathematics misses the RMSEA criterion¶

Claim · All claims
Evidence · 1 study · associational r2 · q2
Currency · newest study 2013 · checked against its sources 2026-10-15 (automated, 1 of 1 entry) · not reviewed by a person · overdue for a recheck since 2016: tool and model accuracy changes quickly

Subclaims¶

q2 i? CFA of goal-level scores shows the one-factor (unidimensional) model is the most reasonable model for MAP tests in the ten studied states; RMSEAs for Michigan Mathematics did not meet the Hu and Bentler criteria. → Shudong Wang 2013

Evidence¶

Shudong Wang 2013¶

Shudong Wang, Marty McCall, Hong Jiao, & Gregg Harris. (2013). Construct Validity and Measurement Invariance of Computerized Adaptive Testing: Application to Measures of Academic Progress (MAP) Using Confirmatory Factor Analysis. Journal of Educational and Developmental Psychology, 3(1). https://doi.org/10.5539/jedp.v3n1p88

q2 · i? · associational · r2

CFA (Proc TCALIS, maximum likelihood) on goal scores from MAP Reading and Mathematics tests in ten states, grades 3-9, five academic terms. Fit indexes met Hu and Bentler criteria, and "the one-factor (unidimensional) model is the most reasonable model". Michigan Mathematics RMSEAs were exceptions. In the later test of invariance across grades, RMSEAs and SRMRs for Michigan Mathematics missed the criteria (Table 9: RMSEA 0.08–0.11, SRMR up to 0.14 under the loading-plus-variance constraint), and metric invariance was supported for every test except Michigan Mathematics.

"All values of fit indexes (except RMSEAs for Michigan MAP Mathematics tests) satisfy the Hu and Bentler (1999) criteria and show that each model fits data extremely well for different content areas, grades, terms, and states. Overall results suggest that the one-factor (unidimensional) model is the most reasonable model for MAP tests in these 10 states."

Discussion¶

Related Claims¶

• The authors conclude that MAP reading and mathematics CAT constructs are invariant across five time points in ten states (except Indiana mathematics), though their mathematics tables show strong-invariance CFI drops above 0.01 in most states — related
• MAP tests show metric invariance across grades for all studied tests except Michigan Mathematics tests — related
• Achievement dimensionality appears to change from pretest to end-of-course testing, questioning unidimensional ICC measurement of individual growth during instruction — related
• Chi-square difference tests show significant loading invariance constraints for Kansas Reading but all constraints significant for Michigan Mathematics — related
• Quadratic latent growth fits MAP longitudinal achievement data better than linear growth across ten states — related
• Item parceling at the testlet level substantially improves model fit, allowing partial (Rasch, 2PL) or full (bifactor) recovery of CAT constructs — related
• State epistemic curiosity and situational interest items load on a single latent factor in confirmatory factor analysis — related
• Establishing longitudinal factorial invariance is neither sufficient nor necessary for examining developmental change — related
• Tutor ratings do not entirely tap one underlying trait; one- and two-factor models fit unsatisfactorily — related
• Confirmatory factor analysis on FICSMath data confirms the 4C/ID Model's four components fit secondary-tertiary mathematics instructional experiences (CFI = 0.907, RMSEA = 0.050) — related
• MLGM model fit failed for the Indiana mathematics test, the sole exception to invariance support — related
• Immediate knowledge of results during computerized achievement testing produces essentially no systematic differences in achievement estimates or response dimensionality — related
• Global-fit evidence does not clearly favor unidimensional over multidimensional academic resilience measurement models, as fit-index pass rates diverge by index — related
• Early numeracy shows a two-factor structure (NRS and CS) on the Dutch ENT-3, with the two-factor model outperforming a one-factor model at several time points — related
• The five well-being measures form a robust unidimensional structure replicating across both studies — related