very fascinating, thank you!

]]>Hi Daniel,

Thanks for the great question. ðŸ˜Š

I think that the bias reduction property of jackknife estimators is interesting. However, I actually think the real value of the jackknife in modern applications is the construction of pseudo-observations and their use in variance estimation.

In the case of regression, I’m not so sure the jackknife with respect to bias would provide many advantages since most coefficient estimators are unbiased.

**However, for estimands without regression frameworks, pseudo-observations can be treated as the response in a GLM to adjust for covariates and estimate treatment effects! **

This paper provides some cool examples of the application of pseudo-observations in survival analysis:

Andersen, P. K., & Pohar Perme, M. (2009). Pseudo-observations in survival analysis. Statistical Methods in Medical Research, 19(1), 71â€“99. doi:10.1177/0962280209105020

I’m hoping to follow-up this blog post with another discussing some applications of pseudo-observations… eventually…

Cheers,

Emma

Hi Glen,

I’m glad you found the blog post helpful. ðŸ™‚

I’m part of a research group working on estimating nonparametric treatment effects in two-arm clinical trials.

The effect we are interested in is equivalent to the area under the receiver operating curve, for which estimation has long been of interest to Radiology.

*Hanley, J. A., & McNeil, B. J. (1982). The meaning and use of the area under a receiver operating characteristic (ROC) curve. Radiology, 143(1), 29â€“36. doi:10.1148/radiology.143.1.7063747*

This result was employed by

*DeLong, E. R., DeLong, D. M., & Clarke-Pearson, D. L. (1988). Comparing the Areas under Two or More Correlated Receiver Operating Characteristic Curves: A Nonparametric Approach. Biometrics, 44(3), 837. doi:10.2307/2531595 *

to derive variance and covariance estimators for multiple AUCs, and seems to have been rediscovered several times in different ways!

Cheers,

Emma