Hi Pavel,

there nothing like a "c statistic" (IMHO). What you mean is simply the confidence intervall for a certain parameter. What is this parameter--that means, what does e.g the first 0.71 mean? Is it standardized mean difference? Normally, this is what you enter in a data set together with the N that underlies the estimation of the parameter.

HTH

Holger

Dear Holger, c-statistic is a parameter I want to evaluate in my meta-analysis. 0.71 is a c-statistic of the particular prediction model and assess its accuracy (please see Table 3 of the following paper). I asked the previous question because I want to acknowledge how should I code c-statistics with their confidence intervals from different studies to analyze them in meta-analysis.

https://link.springer.com/article/10.1007/s40620-021-01220-w/tables/3

Hi Pavel,

you are right. ChapGPT as the rescue taught me:

"The c-statistic (also known as the concordance statistic or the area under the receiver operating characteristic curve) is a measure of the discrimination ability of a predictive model in epidemiology. It quantifies the ability of the model to correctly distinguish between individuals who experience an event (e.g., develop a disease) and those who do not. A c-statistic of 0.5 indicates random prediction, while a c-statistic of 1.0 indicates perfect prediction."

Unfortunately, I cannot tell whether this thing as a defined sampling distribution that can be used to estimate meta-analytical SE and CIs...

But otheres here will probably know.

Good luck!

Holger