Help with analysis methods...

#1
Hello all.

I was wondering if you can help me. I have a set of results of two separate tests (lets say Test A and Test B) that detect the same thing (lets say Disease X).
After testing the same group of people (N=400) with both tests I have the results.
Test A yielded 150 positive results and 250 Negative
Test B yielded 120 positive results and 280 Negative (all the 120 results are included in the 150 positive ones of Test A)

What would be the best method to go about analyzing those results for statistical significance?

Ay help would be greatly appreciated!!

Thank you so much

Jer
 
#5
Yup! And the point is to compare the proportions between the two (as far as sensitivity goes). Is there anything I should use (besides descriptives frequencies etc. like binomial test for proportions?

Any ideas?
 

hlsmith

Not a robit
#6
What was your hypothesis or study objective? Was the outcome based on a threshold for a continuous variable (e.g. CB4 count, bioassay, etc.)?

If test A is the gold-standard, well then you would calculate the following for Test B:

Sensitivity
Specificity
Positive Predictive Value
Negative Predictive Value
Positive Likelihood
Negative Likelihood
Accuracy
ROC Curve
and maybe Lift

-all with their respective confidence intervals.
 
#7
Agree. Also, you can run a one-sample test for proportions using the sample Test B success rate as a random variable. The sample would be those infected according to Test A.