Random-intercept and lag2 panel models rest on different assumptions and offer alternative perspectives on controlling covariates¶

Claim · All claims
Evidence · 1 study · associational r2 · q1
Currency · newest study 2024 · checked against its sources 2026-10-21 (automated, 1 of 1 entry) · not reviewed by a person

Subclaims¶

q1 i? RI models can control fixed unmeasured covariates stable over time but rely on strong, not easily tested assumptions, while lag2 models adjust for unmeasured confounders without those assumptions; the two address different questions. → Marsh 2024

Evidence¶

Marsh 2024¶

Marsh, H. W., Guo, J., Pekrun, R., Lüdtke, O., & Núñez-Regueiro, F. (2024). Cracking Chicken-Egg Conundrums: Juxtaposing Contemporaneous and Lagged Reciprocal Effects Models of Academic Self-Concept and Achievement's Directional Ordering. Educational Psychology Review, 36. https://doi.org/10.1007/s10648-024-09887-w

q1 · i? · associational · r2

Theoretical-methodological argument in the research-hypotheses section, not an empirical test. The article argues RI models depend on strong modeling assumptions (e.g., no nonlinear effects) that are not easily tested and might overcorrect cross-lagged effects if misspecified, whereas lag2 paths help adjust for unmeasured confounders without those assumptions.

"RI and lag2 models are based on different assumptions, address different questions, and offer alternative perspectives on the control of covariates."

Discussion¶

Related Claims¶

• In German secondary-school mathematics, effects of self-concept on achievement are lagged (lag1) whereas effects of achievement on self-concept are more contemporaneous (lag0) — related
• A taxonomy of panel models incorporating lag0, lag1, and lag2 effects, random intercepts, and covariates supports reciprocal effects between academic self-concept and achievement — related
• RI-CLPMs control unmeasured time-invariant covariates better than CLPMs, but neither controls unmeasured time-varying covariates well — related
• Adding lag-2 paths to CLPMs substantially improves fit and provides stronger controls for preexisting differences, positioning extended CLPMs as viable alternatives to RI-CLPMs — related
• Simulation evidence shows RI-CLPM estimates are biased when the true model is a CLPM with lag-2 effects, and fit statistics cannot decide between the two — related
• Reciprocal effects are robust to time-invariant confounders, but the French lagged self-concept effect did not survive the lag2 carryover stress test — related
• The ESDM's unidirectional trait-ACH-to-trait-ASC path is not empirically identified, and the confounder-based RI-LTVC alternative shows poor fit with no support for a time-varying confounder — related
• Deviations from the convergence assumption produce unreliable cross-lagged findings in school motivation research — related
• Next-generation cross-lagged models relax the convergence assumption in school motivation panel data — related