Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: st: Sensitivity and specificity graphs for binary vars
From
Nick Cox <[email protected]>
To
[email protected]
Subject
Re: st: Sensitivity and specificity graphs for binary vars
Date
Wed, 23 May 2012 14:03:36 +0100
What kind of graph do you want? With a bar graph for those numbers you
get one very large bar and three small ones. Or you could plot the
fractions on a logit scale. There's ROC stuff too. You probably need
to spell what is customary for your kind of biostatistical audience.
(I'm not a bio- or medical statistician. Or even a statistician.)
Nick
On Wed, May 23, 2012 at 1:51 PM, King, Carina <[email protected]> wrote:
> I realise any graphs are not really necessary here, I just much prefer visual presentations where they're possible (and obviously only if they are appropriate!). I was hoping for some help with the detection error trade-off graph mainly, as I couldn't really find anything about making them in Stata? Or whether it would be the best thing? Thanks!
Nick Cox
> Why do you need or want a graph at all?(*) All the information is summarised efficiently by those four totals.
> (*) Longstanding Statalist members are asked not to quote me saying this out of context.
On Wed, May 23, 2012 at 1:29 PM, King, Carina <[email protected]> wrote:
>
>> I have two binary variables, one being a 'gold standard' indicator of infection and the other based on a coded predictor. I have calculated the sensitivity and specificity (using -diagt-) of the coded predictor for identifying the 'gold standard' infections, but want to represent it graphically. I have looked at the ROC commands, including -roctab- and -roccomp- but don't think these are right as both my variables binary variables. My questions are whether a Detection Error Trade-Off graph is appropriate? Or is there another type of graph that would show the sensitivity and specificity well? And are they possible in Stata?
>>
>> I am using Stata SE 11
>>
>> Two-way table of my data:
>>
>> Negative (gold standard) Positive (gold standard)
>> Negative (coded) 542 (true neg) 9 (false neg) Positive
>> (coded) 21 (false pos) 26 (true pos)
*
* For searches and help try:
* http://www.stata.com/help.cgi?search
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/