Model | AIC | ΔAIC | Inference |
Straight line, no intercept, with lognormal heteroscedastic error | 13,919.56 | 352.96 | No empirical support |
Three-parameter power function with lognormal, heteroscedastic error | 13,825.72 | 259.12 | No empirical support |
Straight line, intercept, with lognormal heteroscedastic error | 13,823.78 | 257.18 | No empirical support |
Two-parameter power function with lognormal, heteroscedastic error | 13,823.74 | 257.14 | No empirical support |
Straight line, no intercept, with normal heteroscedastic error | 13,702.02 | 135.42 | No empirical support |
Straight line, no intercept, with normal homoscedastic error | 13,696.38 | 129.78 | No empirical support |
Three-parameter power function with normal, heteroscedastic error | 13,610.86 | 44.26 | No empirical support |
Three-parameter power function with normal, homoscedastic error | 13,604.72 | 38.12 | No empirical support |
Straight line, intercept, with normal, heteroscedastic error | 13,568.30 | 1.70 | Essentially equivalent |
Straight line, intercept, with normal, homoscedastic error | 13,567.62 | 1.02 | Essentially equivalent |
Two-parameter power function with normal, homoscedastic error | 13,567.60 | 1.00 | Essentially equivalent |
Two-parameter power function with normal, heteroscedastic error | 13,566.60 | 0 | Best |
Qualitative terms for the relative difference (ΔAIC) from the estimated best model (ie, the model with the lowest AIC value; ΔAIC=0) were assigned according to the following scale: 0–2, essentially equivalent; 2–7, plausible alternative; 7–14, weak support; >14, no empirical support.10
AIC, Akaike’s information criterion; Δ AIC, Akaike difference.