In a CBC survey, participants choose in total 2724 times level A and 2487 times level B (=52.3% A and 47.7% B).
The average utility in CBC/HB for A was -.02618144 and for B = .16109872.
Calculating the odds and thereafter the probabilities results in 49.3% for A and 54.0% for B.
Is this plausible? Is there literature on this effect?