Fits of Gaussian, gamma, and inverse Gaussian distributions to middle times of rat behavioral data from the peak-interval procedure, indicating clear superiority of the inverse Gaussian fits
| Gaussian | Gamma | Inverse Gaussian | |
|---|---|---|---|
| 30 s | log likhd., −1803.46 | log likhd., −1787.62 | log likhd., −1784.44 |
| μN = 30.91 | k = 22.98 | μ = 30.91 | |
| σ = 6.55 | θ = 1.35 | η = 692.70 | |
| 45 s | log likhd., −1812.96 | log likhd., −1801.06 | log likhd., −1799.42 |
| μN =45.90 | k = 27.65 | μ = 45.90 | |
| σ = 8.84 | θ = 1.66 | η = 1239.40 | |
| 60 s | log likhd., −1381.05 | log likhd., −1371.13 | log likhd., −1370.27 |
| μN = 62.47 | k = 23.40 | μ = 62.47 | |
| σ = 13.12 | θ = 2.67 | η = 1417.48 |
Larger log likelihood (likhd.) values indicate better fits; all models considered have equal complexity (two parameters). Parameters are as follows: Gaussian, mean μN, standard deviation σ; gamma, shape k, scale θ; inverse Gaussian, mean μ, shape η.