@DataScienceInR Connect the dots by introducing the 'concordance statistic': https://t.co/8p2a9uJMA8
When comparing performance of the same logistic model in different populations, a higher c-statistic is to be expected for the population with greater variation in the explanatory variable, even if the odds ratio is transportable across populations. https:
RT @ESteyerberg: @TPA_Debray @Richard_D_Riley 5.The link between calibration slope and discrimination can mathematically be defined. https:…
@TPA_Debray @Richard_D_Riley 5.The link between calibration slope and discrimination can mathematically be defined. https://t.co/zxiFrIJ7Y3. Discrimination is determined by the slope and the case-mix in a validation setting: https://t.co/AKAz3PXhRN; https:
This is under binormal model, ref: https://t.co/haPAEPZOrD, by Peter Austin and @ESteyerberg.
@toxicpath @Grumpy_Hoosier @InuWolfie @SteveStuWill @michaelshermer There is a direct relationship between cohens d and c-statistics that has nothing to do with absolute value (ie size of the mean). Recall cohens d only has to do with differences in mean.
Interpreting the concordance statistic of a logistic regression model: relation to the varian... https://t.co/ojjsTvwt0j