Table 3.
Analysis of the children participating in CHS. We show parameter estimates (standard errors) using maximum likelihood on the original cohort (OC) and maximum conditional likelihood on the outcome-dependent sample (ODS). The OC pertains to 1286 subjects, and the ODS pertains to the 698 subjects who exhibited response variation among all subjects. We also display the observed ρx for each covariate
| Covariate | ρx | OC | ODS |
| Intercept | 1 | − 1.235 (0.079) | − 1.476 (0.202) |
| Asthmatic | 1 | 0.112 (0.091) | 0.291 (0.182) |
| p = 0.15† | 0.110 | 0.225 | |
| p = 0.23† | 0.093 | 0.191 | |
| p = 0.30† | 0.085 | 0.175 | |
| Within-subject, 4-day average ozone (per 10 ppb) | 0 | − 0.007 (0.046) | 0.037 (0.049) |
| p = 0.15† | 0.045 | 0.047 | |
| p = 0.23† | 0.046 | 0.048 | |
| p = 0.30† | 0.047 | 0.049 | |
| Asthmatic × within-subject ozone | 0 | 0.037 (0.064) | 0.039 (0.061) |
| p = 0.15† | 0.077 | 0.076 | |
| p = 0.23† | 0.065 | 0.064 | |
| p = 0.30† | 0.060 | 0.059 | |
| Between-subject, 4-day average ozone (per 10 ppb) | 1 | 0.048 (0.057) | 0.113 (0.080) |
| Thirty-day average ozone (per 10 ppb) | 0.72 | − 0.165 (0.06) | − 0.231 (0.067) |
| Male | 1 | − 0.468 (0.079) | − 0.563 (0.171) |
| Noncaucasian | 1 | − 0.377 (0.116) | − 0.377 (0.245) |
| Grade 7 (versus grade 4) | 1 | 0.183 (0.081) | 0.008 (0.168) |
| Secondhand smoke exposure | 1 | − 0.039 (0.081) | 0.104 (0.167) |
| Dependence model parameters | |||
| γ | 0.603 (0.110) | 0.603 (0.115) | |
| log(σ) | − 1.427 (0.848) | − 1.427 (0.627) |
†
Standard error estimates from our feasibility check, where p is the presumed proportion of asthmatic children in the OC.