Dear all,
I asked you few days ago about confidence intervals for medians, and I'd like to thank you all the help I received.
Now I have a new question, I have a database where I have clinical figures ungrossed (unadjusted for missing) and grossed(adjusted for missing data). I want to know if there is a relevant difference between the grossed and ungrossed medians.
I wanted to compare both using confidence intervals, and if they overlap in the confidence intervals and i would not have any statistically evidence to prefer the grossed median. So I'd use the ungrossed median.
(When I have a large number of observation and the number of missing is relative small I have almost no difference between both medians)
However my database contains the whole population, so I'm not dealing with a sample, although I have missing data.
Can I proceed in this way? Am I making any wrong assumption (comparing the medians, whole populations vs sample)?
Thank very much for all your help,
Kind regards,
Jorge
---------------------------------
With Yahoo! Mail you can get a bigger mailbox -- choose a size that fits your needs
|