hello everyone!
I have a set of data related to certain treatment response, I wanted to calculate the 95% CI of the objective response rate, please anyone know how to do that using SPSS? thanks in advanced !
Meet us half way!
How is treatment response measured? 0/1 – non-response/response, or on an ordered scale, or a numeric scale.
And, critically, what statistic are you using to summarise the response?
As Ronán Michael Conroy has noted, you have not provided enough information. If you are asking about how to compute a 95% CI for a binomial proportion, the attached syntax file may be helpful. It has code for 5 different methods:
1. Clopper-Pearson "exact" method
2. Wald method
3. Adjusted Wald method (Agresti & Coull, 1998)
4. Wilson score method
5. Jeffreys method
HTH.
I might add to Bruce Weaver 's helpful syntax file that the choice of method is dependent on your sample size. With N < 100, the Jeffrey's method seems to get closest to nominal coverage, while with N > 100 the Agresti-Coull method is preferable. See
Brown, Lawrence D., et al. “Interval Estimation for a Binomial Proportion.” Statistical Science, vol. 16, no. 2, 2001, pp. 101–117. JSTOR, www.jstor.org/stable/2676784. Accessed 9 Apr. 2021.
It's also worth noting that the so-called 'exact' method isn't exactly what you thought an exact method would be.
My go-to resource for this topic is Robert Newcombe's book--or at least it used to be before I started working from home (WFH).
- https://www.routledge.com/Confidence-Intervals-for-Proportions-and-Related-Measures-of-Effect-Size/Newcombe/p/book/9780367576707
If you click on the Support Material link on that page, it will take you to the bottom of the page where you have the option to download a file called K10649.zip. It contains an Excel workbook implementing all (or at least most) of the methods Newcombe describes in his book.
IIRC, Newcombe's book includes graphs showing that both the Agresti-Coull and Wilson score intervals perform pretty well across the whole range of possible proportions (0 to 1) and for any sample size too. So if one wanted to choose one (approximate) method to use in most circumstances, I think either of those would be a good choice. (But notice again that I opened this paragraph by saying IIRC.) ;-)
HTH.
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Software
- SPSS