roc analysis in spss

noob

New Member
#1
Hello,

I am analyzing some data from 2 diagnostic tests and I want to do a ROC analysis in SPSS. However I don't have any experience with this, so please correct me if I'm doing something wrong.

TEST A (GOLD STANDAARD)

score range: 0 - 10
outcome: healty or not healty
healty patients (=pts) = cut off score ≤ 2
not healty pts = score >2

I want to evaluate the performance (predictive value, optimal cut off, sensitivity, specificity) of test B to distinguish healty from not-healty pts, this compared to TEST A.

TEST B --> score range: 0 - 15

DATASET:
TEST A (GOLD STANDAARD): continuous data
TEST B: continuous data

ANALYSIS in SPSS 17:

For ROC analysis I GUESS :)confused:) I have to transform the data for TEST A to a dichotomous outcome. right??
--> healty= 0 ; not-healty= 1

SPSS:

ANALYZE --> ROC CURVE
TEST VARIABLE: results from TEST B (continuous data)
STATE VARIABLE: results from TEST A (dichotomous data)
VALUE OF STATE VARIABLE: 1 (don't understand this 100%, but I want to know who's not healty so 1)


Do you think this is correct?
Are there other things I have to consider?


Thank you for your time.
 

noob

New Member
#2
correct

Hello,

2e question:

I computed the data: spss gives me sensitivity and 1-specificity results for different cut off values for Test B.
These cut off values are 1.5 , 2.5 , 3.5 ...

How can I get results for cut off values 2 , 3 , 4 ...


I think my post should be under 'Epidemiology and Biostatistics'. My fault.
 
Last edited: