Table 4. Estimated ERR (βˆ) and null information at β = 0 for each of the analysis methods using data with case and location outcome and location-group simulated from the WECARE case-control sets. Location-group likelihoods were used for the location-group data sets.
| Analysis method | Location of origin data | Location group sim ulation | ||
|---|---|---|---|---|
| Precise location | Markov growth | Completely unknown | ||
| A. Estimated ERR/Gy (βˆ) (95% confidence intervalsa) | ||||
| CCML | 0.89 (0.60-1.25) | 1.05 (0.59-1.67) | – | – |
| CL | 0.91 (0.10-5.48) | 0.34 (-0.10-∞) | 0.65 (-0.01-5.00) | 0.34 (-0.10-∞) |
| CCAL | 0.88 (0.59-1.24) | 1.03 (0.58-1.65) | 0.89 (0.59-1.24) | 0.88 (0.58-1.25) |
| Location group average | – | – | 0.85 (0.56-1.19) | 0.77 (0.52-1.08) |
| B. Null information per case (Percent relative to location of origin-CCAL) | ||||
| CCML | 0.60 (94%) | 0.32 (49%) | – | – |
| CL | 0.17 (27%) | 0.09 (15%) | 0.14 (22%) | 0.09 (15%) |
| CCAL | 0.64 (100%) | 0.34 (53%) | 0.60 (94%) | 0.56 (87%) |
| Location group average | – | – | 0.59 (93%) | 0.63 (99%)b |
a
based on the profile likelihood.
b
not valid since estimate is biased.