6
$\begingroup$

The latest edition of The Lancet has a paper reporting on mortality among children under the age of 5 around the world. It is entitled "Global, regional, and national causes of under-5 mortality in 2000–15: an updated systematic analysis with implications for the Sustainable Development Goals." In the abstract, they report something called a "95% uncertainty range" when stating the number of preterm birth complications noted in the study.

What is the difference between a "95% uncertainty range" and a "95% confidence interval"? Are these in fact the same thing? And is there some underlying philosophical reason to prefer the term "uncertainty range" to "confidence interval"? If so, what are the pros and cons of each term?

Improve this question
$\endgroup$
3
  • 1
    $\begingroup$ I took a look at the paper. They seem to be using models to determine age specific mortality and evaluate the models using cross-validation and construct bootstrap uncertainty intervals. They mention using 2.5th percentile and 97.5th percentile to construct the uncertainty intervals. They do not use the term confidence intervals but it looks like these interval might be bootstrap percentile two-sided 95 percent confidence intervals. $\endgroup$
    – Michael R. Chernick
    Commented Dec 16, 2016 at 18:20
  • 2
    $\begingroup$ Andrew Gelman recently re-sparked discussion of the labelling of confidence-interval-like intervals: andrewgelman.com/2016/11/26/… $\endgroup$
    – Michael Lew
    Commented Dec 16, 2016 at 20:08
  • $\begingroup$ See also: journals.lww.com/epidem/Fulltext/2012/09000/… $\endgroup$
    – Fomite
    Commented Apr 11, 2018 at 21:48

1 Answer 1

Reset to default
3
$\begingroup$

If you look at some other papers from the project "Global burden of diseases", you will see that data imputation for missing cases have been carried out using proxy. As in case of modelling the global burden of diarrhoea, due to unavailability of data from systematic surveillance in countries like Liberia, Libya, Afghanistan, Syria and Haiti; estimates have been carried out by extrapolating spatiotemporal trends from countries or region with better data matching for the geographical and socio-economic condition. Hence, it is likely that due to the uncertainty associated with the estimates, instead of using confidence interval, the term uncertainty interval might have been used.

https://www.sciencedirect.com/science/article/pii/S1473309917302761

https://www.sciencedirect.com/science/article/pii/S1473309917303365

Improve this answer
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.