I didn't notice that you were using the -singletonA option. Neither ETAC nor Clustsim will work with that option, and I thought I had put in a test to make sure that type of situation would print out an error message and stop. I guess I did not do that (but will do it in a moment).
In your case, you want to test the mean of the -setB datasets against the constant 0.5. The only way to do that with ETAC or Clustsim at this time is to manually subtract 0.5 from each of the datasets, and then do a 1-sample test of the results (against 0). The subtraction is easily done with the 3dcalc program.