Addi van Bergen

Supplementary material chapter 2 187 Statistical analysis Confounding, match-ing & stratification etc. Results per indicator*** Correlations and effect estimates per indicator $ Combined result $$ Methodological limitations # Study quality ## Linear regression Adj for gender, age. age 2 , ethnicity, migrant, marital status, job status, educational level, social class, region and (transitions in) SAH and LLTI. Mediator/ moderator: rural vs urban; car access, mobile phone ownership, internet use Effect on SP (wave 3) -Good=>poor: + -Poor=>good: + -Stable poor: + Effect on SP (wave 3) c) GH1 ß (95% CI) -Stable good is reference -Good=>poor 0.76 (0.49- 1.02) ^^^ -Poor=>good 0.61 (0.32- 0.90) ^^^ -Stable poor 0.95 (0.72- 1.18) ^^^ + 2 3 5 + Effect on SP (wave 3) -Good=>poor: ns -Poor=>good: ns -Stable poor: + Effect on SP (wave 3) c) GH2 ß (95% CI) -Stable no LLTI is reference -Stable LLTI 0.22 (0.02- 0.42) ^ Logistic regression + SP Effect on GH3 (wave 4) SP: OR adj =1.15 (1.09-1.21) + SP Effect on GH4 (wave 4) SP: OR adj =1.07 (1.02-1.12) Bivariate correlation + S E1 E2 P2 P3 ns P1 P4 P5 C1 C2 (p<0.001) Kendall tau_b: S:0.06; E1:0.07; E2:0.08; P2:0.05; P3:0.12 +/0 2 4 5 6 7 + Concentration index: % contribution to health inequality Other factors in model: gender, age, education, employment, urbanicity, region, deprivation + S2 S4 E1 E2 P ns S1 S3 S5 S2 3.87%; S4 2.58%; E1: 29.85%, E2:2.61%, P: 8.56% + 3 4 5 + + S1 S2 S3 S4 E1 E2 P ns S5 S1 0.23%; S2 3.68%; S3 0.23%; S4 4.05%; E1: 29.73%, E2:3.74%, P: 7.17% + S2 S3 S4 E1 E2 P ns S1 S5 S2 4.40%; S3 0.11% S4 4.43%; E1: 32.56%, E2:4.02%,P: 8.01%