A Comparison of Four Methods for the Analysis of N-of-1 Trials