3. Results
The model checking procedure allowed to detect, in the majority of
cases, the failure of the logistic model (θ¼1) to satisfactorily fit the
data in cases of low (σobs¼0.05) and middle (σobs¼0.1) noise levels
(Fig. 1A and B). Indeed, 94% of the pmin values obtained on data
generated with the theta-logistic (θ¼4) were lower than the 5 percentile
of pmin values obtained on data generated with the logistic
model (θ¼1), in the low noise case (Fig. 1A). This number was still
equal to 64% in the middle noise case (Fig. 1B). In contrast, this model
checking procedure failed in the large noise case (Fig. 1C).