The same type of probabilities can be used to predict performance for a quadrant cursor path, where the predicted correct entry probability to reach position k would be π quadrant(k) π row(k) π column(k) and the predicted error entry probability is 1 minus that product.
Multivariate Cox regression modeling was done for DFS, EFS and OS using a forward-selection stepwise modeling process (with a forward selection method with entry probability of P = .01.01
The entry probability was 0.05, and none of the variables were forced back into the model.
In the absence of other information (e.g., Buckeridge, et al. [ 16] do not report their probabilities) we set the ED utilization probabilities to be: P d = 0.25 (weekday entry probability at the prodromal stage); P w = 0.4 (weekend entry probability at the prodromal stage) and P f = 0.80 (entry probability at the fulminant stage).
All regression analyses reported here used the stepwise method and were conducted with the default entry probability of p < 0.05 and the default removal probability of p > 0.10.
Univariate and forward stepwise (entry probability = 0.05, removal probability = 0.1) logistic regression were subsequently performed to assess the contribution of each associated marker and to identify the most predictive multivariate model, respectively.
The final set of significant factors associated with awareness was identified through the use of a forward stepwise selection procedure with entry probability of 0.05 and removal probability of 0.10.
Unconditional logistic regression analysis with stepwise forward selection (probability for entry 0.05; probability for removal 0.10) of variables showed that those with risk effect were older age, male sex, rural inhabitant, farmer or fisher occupation, lower income, consumption of tobacco, alcohol, fermented fish sauce, pickled vegetable, and sowbelly.
Backward stepwise regression was performed with variable entry when probability was less than 0.05 and removal when probability exceeded 0.10.
