Table 2.

Comparison of our results of modeling the relationship between CO and REC (Equation (1)) with those of NIOSH/NCI.

Model	AICa	β (slope)	Source
Fixed common intercept, fixed common slope	586.6 586.5 579.9	0.47 0.47 0.43	Our results NIOSH-NCIb using our imputed values
Fixed mine-specific intercepts, fixed common slope	519.4 510.0	0.44 0.40	Our results using our imputed values
Fixed mine-specific intercepts, random mine specific slopes	522.7 516.8c 512.6	Our results NIOSH-NCI using our imputed values
Fixed mine-specific intercepts, random common slope	524.2 516.8c 514.9	0.43 0.58d 0.39	Our results NIOSH-NCI using our imputed values

^aMeasure of model fit (smaller indicates better fit). Our estimates of AIC and slope come from the maximum likelihood fitting alternative in the MIXED Procedure (SAS).

^bThese results are from a model that is only described by NIOSH/NCI as the “regression model” but we assume from the similarity of results that they are referring to this model.

^cIt is not clear to us whether this AIC was calculated using a common random slope or mine-specific random slopes.

^dIt is not clear to us whether this slope was calculated using this model (preferred approach) or represents an average of mine-specific random slopes (not preferred).