Twenty-seven users completed a task, and each user did so with a combination of two variables, S and V, which can each have a value of "yes" or "no". Pass/fail data was collected for each user with the following results:
S=yes, V=yes: 7/8 users passed
S=yes, V=no: 1/4 users passed
S=no, V=yes: 10/10 users passed
S=no, V=no: 4/5 users passed
How would one determine which combination of variables is the best (with statistical significance)? If that's not possible, can we show that one combination is statistically significantly worse than the others? What method would be used?