Skip to main content

Table 2 AIC models for autism spectrum characteristics, within whole sample

From: Gene functional networks and autism spectrum characteristics in young people with intellectual disability: a dimensional phenotyping study

Component Models N variables AIC weight AICc ΔAIC Residual deviance
Inflexibility Anxiety + FNG + Hyperactivity + Vineland 4 0.221 100.21 0 17.81
Anxiety + FNG + Hyperactivity 3 0.202 100.39 0.18 18.98
Anxiety + FNG + Hyperactivity + (FNG × Hyperactivity) 5 0.175 100.68 0.466 18
Social understanding Anxiety + FNG + Hyperactivity + Inattention + Vineland + FNG × Hyperactivity + FNG × Inattention 9 0.174 114.93 0 20.26
FNG + Hyperactivity + Inattention + Vineland + FNG × Hyperactivity + FNG × Inattention 8 0.133 115.47 0.537 21.98
Anxiety + FNG + Gender + Hyperactivity + Inattention + Vineland + FNG × Hyperactivity + FNG × Inattention 10 0.131 115.49 0.564 19.05
Social motivation Hyperactivity + Inattention 2 0.34 127.12 0 36.37
Hyperactivity 1 0.3328 127.19 0.071 38.43
Age + Hyperactivity 2 0.199 128.19 1.07 37.24
  1. Summary of the best set of three similarly supported models, for each ASC dimension. N variables = number of parameters for each model, AIC weight = is the probability of each model being the best model, or relative evidence for each model. These estimates are computed by normalising model likelihoods. AICc = AIC criterion of model selection, corrected for smaller sample size, ΔAIC = AIC difference between the best fitting model (equal to zero) and the second best model. Residual deviance = distance between the data and the model