Inferring candidate main effects

[emjun]

As of Dec. 03, relying primarily on conceptual graph to suggest main effects. Could consider additions or alternatives, such as :

• checking for type of edge (causal, correlational) in conceptual graph
• looking at the dataset
• may depend on "task" for explanation or prediction