Neighbourhood, district, and municipality variation contributes little to achievement (under 2% of variance collectively) and does not interact with student EA-PGI¶

Claim · All claims
Evidence · 1 study (2 entries) · associational r2–r3 · q3
Currency · newest study 2022 · not yet checked against its sources · not reviewed by a person

Subclaims¶

q3 i? Residential area main effects on achievement are small: ~1% variance for municipalities, ~1% for neighbourhoods, <1% for districts. → Rosa Cheesman 2022 q3 i? The best-fitting model included random slopes for schools but not for residential areas, indicating EA-PGI effects do not depend on neighbourhoods, districts, or municipalities. → Rosa Cheesman 2022 (2)

Evidence¶

Rosa Cheesman 2022¶

Rosa Cheesman, Nicolai T. Borgen, Torkild H. Lyngstad, Espen M. Eilertsen, Ziada Ayorech, Fartein A. Torvik, Ole A. Andreassen, Henrik D. Zachrisson and Eivind Ystrom. (2022). A population-wide gene-environment interaction study on how genes, schools, and residential areas shape achievement. npj Science of Learning 7:29. https://doi.org/10.1038/s41539-022-00145-8

q3 · i? · associational · r3

Intraclass correlations from the multilevel models fitted to the 23,471-student sample show "1% for municipalities, 1% for neighbourhoods, and <1% for districts" variance explained in achievement by residential levels.

"The variance explained in achievement by residential areas was 1% for municipalities, 1% for neighbourhoods, and <1% for districts (see Supplementary Table 4a for intraclass correlations)."

Rosa Cheesman 2022 (2)¶

Rosa Cheesman, Nicolai T. Borgen, Torkild H. Lyngstad, Espen M. Eilertsen, Ziada Ayorech, Fartein A. Torvik, Ole A. Andreassen, Henrik D. Zachrisson and Eivind Ystrom. (2022). A population-wide gene-environment interaction study on how genes, schools, and residential areas shape achievement. npj Science of Learning 7:29. https://doi.org/10.1038/s41539-022-00145-8

q3 · i? · associational · r2

AIC-based model comparison among 11 nested multilevel models showed random EA-PGI slopes improved fit for schools only; residential areas retained intercepts only, so EA-PGI-by-residential-area interactions were not supported.

"The best-fitting model – Model 3a in Table 1 – included random slopes and intercepts for schools, but only random intercepts (not slopes) for residential areas."

Discussion¶

Related Claims¶

• Students' EA-PGI interacts with schools such that school differences in achievement are wider among students with lower EA-PGI (compensation, not Scarr-Rowe suppression) — related
• Between-school variation in EA-PGI effects is substantial: the average EA-PGI effect is 0.22 with slope SD 0.034, and variance explained by EA-PGI ranges from ~2% in the weakest-effect schools to ~8% in the strongest — related
• Five measured school sociodemographic characteristics do not explain the EA-PGI-by-school interaction — related
• The EA-PGI-by-school interaction is stronger for maths and reading than for English — related