Dear R-help,> prop.test(9, 137, p=0.066)> prop.test(9, 137, p=0.05)give two different 95% confidence intervals. I thought the confidence interval calculation should be independent of testing calculations (and thus the null hypothesis)? Splus 2000 has similar problems but give slightly different answer. Using R1.3.0 on windows. Mai Zhou -.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.- r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html Send "info", "help", or "[un]subscribe" (in the "body", not the subject !) To: r-help-request at stat.math.ethz.ch _._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._
mai at ms.uky.edu wrote:> > Dear R-help, > > > prop.test(9, 137, p=0.066) > > > prop.test(9, 137, p=0.05) > > give two different 95% confidence intervals. > I thought the confidence interval calculation > should be independent of testing calculations (and thus > the null hypothesis)?You can transform a two sided test problem into a confidence interval and vice versa. Why should the confidence intervals be the same for different probabilities of success in this case?> Splus 2000 has similar problems but give slightly different > answer.Uwe Ligges -.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.- r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html Send "info", "help", or "[un]subscribe" (in the "body", not the subject !) To: r-help-request at stat.math.ethz.ch _._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._
On Mon, 24 Sep 2001 mai at ms.uky.edu wrote:> I think the confidence > interval are those values of the parameter that have a large > P-value, so independent of the specific null hypothesis > that we are testing. i.e. we can construct a conf. int. from > data without doing a test. The fact that we (later) also > want a test should not change the calculation of the > confidence interval. >OK. You are so right! It's too late to work! It happens because Yates' continuity correction is used, which can be turned off: prop.test(..., correct=FALSE) [BTW: The same in S-Plus] Uwe Ligges -.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.- r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html Send "info", "help", or "[un]subscribe" (in the "body", not the subject !) To: r-help-request at stat.math.ethz.ch _._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._
Possibly Parallel Threads
- prop.test() and the simultaneous confidence interval for multiple proportions in R
- Confidence interval calculation in prop.test
- Confidence interval calculation in prop.test (PR#9325)
- prop.test() references
- Confidence interval for the difference between proportions - method used in prop.test()