type I error alfa estimates.xlsx

Type I error probability estimate, or the significance level estimate, for a simulation-based test of difference in the means of two gamma-distributed variables. The level depends on the parameter set-up for the two gamma distributions. Each sig.level estimate is for a given parameter combination, given sample size. The combinations considered are for either shape parameter to be 1 .. 30 and for the scale 1 parameter to be 1..30. The other distribution's scale 2 parameter is such that the two gamma distributions have the same expected value. Each sig.level is estimated by running the proposed test five thousand times. Six parameter scenarios were considered: scenario1 ... both shape parameters are 1..10, scenario2..both shape params are 11..20, scenario3..both shape params are 21..30, scenario4..shape 1 is 1..10, shape 2 is 11..20, scenario5..shape 1 is 1..10, shape2 is 21..30, scenario6..shape 1 is 11..20, shape 2 is 21..30. Scale 1 is always 1..30 in all the scenarios. The sample sizes (equal) and the scenario gave rise to the data file name. For example, a_sc2_40.xlsx contains sig.level estimates for sample sizes of 40 and all scenario 2 parameter combinations (each number in the file relates to one parameter combination).