Qualitatively similar results were obtained when we carried out alternative non-parametric tests (Wilcoxon-Mann-Whitney test for binary predictor variables; Jonckheere Terpestra test for trends) on our data.
An ideal binary predictor would have zero false predictions and so the desired point on the ROC curve is top left corner where sensitivity is one and (1-specificity) is zero.
Here we employed Weights-of-Evidence (WOE) Bayesian modeling to investigate the spatial relationship between historical fire events in the Chaharmahal-Bakhtiari Province of Iran, using a wide range of binary predictor variables (i.e., presence or absence of a variable characteristic or condition) that represent topography, climate, and human activities.
Using the binary predictor "Involved" and "Uninvolved," an ANOVA of motivation to learn more programming as the dependent variable showed a significant relationship between programming involvement and motivation to learn programming, F 1, 484) = 97.81, p < .001, η p 2 = .17.17
These factors were then used to build a binary predictor of radiation exposure.
We set as the response variable the total number of copulations per male-female dyad, and as the binary predictor variable, whether or not the male had transferred meat to the respective female over the entire study period.
We noted the occasional absence of an attending physician during some observations of internal medicine, so we incorporated the presence of an attending physician as a binary predictor variable in each model.
The additional analyses were performed to validate the poor performance of the supervised binary predictor when the model was built using total body radiation exposure and used to predict partial body radiation exposure.
Gender was used as a single binary predictor.
Univariate comparison of the binary predictor variables against all course outcomes was conducted with t-tests.
This was split to a binary predictor based around a cut-off of 500 people/km.
