Table 2.
Percentage of replicates in which the epidemic growth is better described by either a time-squared or exponential model of growth. ΔAIC values comparing maximum-likelihood fits of an exponential model of epidemic growth and a time-squared model of growth. For each comparison we computed the difference between the AIC for the exponential model and the time-squared model (ΔAIC = AICexponential − AICtime-squared). (Smaller AIC values are associated with a better fit.) Positive ΔAIC values indicate a smaller AIC value for the time-squared and a better fit by the time-squared model of growth, while negative ΔAIC indicate a better fit by the exponential model of growth. (Absolute number of replicates better described by each growth model are shown parenthetically.)
| LDT rate (ϕLDT) | N replicatesa | time-squared growth | exponential growth |
|---|---|---|---|
| 0 | 40 | 72.5% (29) | 27.5% (11) |
| 3.75 × 10−8 | 40 | 20% (8) | 80% (32) |
| 7.5 × 10−8 | 38 | 5.3% (2) | 94.7% (36) |
| 1.5 × 10−7 | 37 | 18.9% (7) | 81.1% (30) |
| 3.0 × 10−7 | 36 | 8.3% (3) | 91.7% (33) |
| 6.0 × 10−7 | 38 | 10.5% (4) | 89.5% (34) |
aThe number of replicate runs varies because the model is stochastic, and in some runs the epidemic goes extinct before reaching the end of the 20 year simulation.