I want to see changes in parameter X between two different groups having disorder Y but in different environmental conditions. How do I calculate changes in median with 95%CI for this parameter? I am using SPSS 21.0 for statistical analysis.
A binomial-based confidence interval for the median and other methods are described in:
http://support.sas.com/documentation/cdl/en/procstat/63963/HTML/default/viewer.htm#procstat_univariate_sect028.htm.
But personally I like to use Bootstrap, preferably with the BCa procedure (Bias corrected and accelerated). SPSS apparently does have a Bootstrap dialog where this option is available.
The median is the middle order statistic. Its pdf may be obtained non-parametrically as with other order statistics. However, the 95% confidence interval of the median depends on the particular assumed underlying distribution. This may notr have anything to do with the statistical package used for the analysis.
.
you might be interested in the following :
https://pdfs.semanticscholar.org/8fa8/fc57bd385695d12f18b1d1e9f3278339e98f.pdf
A Simple Confidence Interval for the Median
David J. Olive
and also :
http://exploringdatablog.blogspot.fr/2012/04/david-olives-median-confidence-interval.html
.
with SPSS, it seems you are left with bootstrapping :
http://www-01.ibm.com/support/docview.wss?uid=swg21476966
but Olive's CI is pretty straightforward to implement
.
A binomial-based confidence interval for the median and other methods are described in:
http://support.sas.com/documentation/cdl/en/procstat/63963/HTML/default/viewer.htm#procstat_univariate_sect028.htm.
But personally I like to use Bootstrap, preferably with the BCa procedure (Bias corrected and accelerated). SPSS apparently does have a Bootstrap dialog where this option is available.
I would use interquartile range (IQR). I think it is useful for describing a median based variable.
I just support Carine.
The best is to Bootstrap the median even though it is possible to apply a confidence interval on the basis of the binomial distribution.
In SPSS, you can use the Ratio statistics in the descriptive analysis dialogue to compute the medium with 95%CI
https://www.ibm.com/support/pages/obtaining-confidence-interval-median
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics