Which test to compare a procedural outcome based on age in a group?

[stuartmcintosh10]

I have a dataset of 200 participants. Procedural outcome is either failed or successful. For each participant, I have an age in years.
I am hoping to perform a statistical analysis to determine if there is increased success or failure of procedure depending on age

Which statistical test would be best for this?

thanks

 

[hlsmith]

Not too sure on how savvy you are, but I would do two things. First a generalized additive model, regressing outcome on age to see if there is a nonlinear relationship. If no, then followed by a general linear model, identity link and binomial distribution. The latter could be substituted with a logistic regression.

 

[Karabiner]

Maybe you compare the failure versus success groups with respect to mean age (t-test);
or with respect to whether the ages in the success group tends to have lower (or higher,
respectively) than the ages in the failure group (Wilcoxon rank sum test).
You could also create a box plot, to compare age between groups.

With kind regards

Karabiner

 

[stuartmcintosh10]

many thanks for your help. much appreciated. When I try logistic regression, for some reason SPSS does not include "age" in the equation

 

[hlsmith]

What does your output or errors look like?

 

[stuartmcintosh10]

so when doing the logistic regression, I get the following: only age appears in step 1 but not 0
Step 0 :Variables in the equation: constant; B -1.41; S.E 0.228; Wald 38.79; df 1; sig <0.001

Variables not in equation: age; score 1.377; df 1; sig. 0.241

Step 1: variables in the equation: age; B -0.0261; S.E 0.023; Wald 1.358; df 1; sig 0.244
constant constant; B 0.261; S.E 1.441; Wald 0.033; df 1; sig 0.856