See this Perspectives piece in Science this week: http://www.sciencemag.org/content/347/6228/1314.short

Jeff Leek and Roger Peng describe how ‘mistaking the type of question being considered [when drawing conclusions from an analysis] is the most common error in data analysis’.

They describe 6 kinds of data analysis: descriptive, exploratory, inferential, predictive, causal, and mechanistic. They provide a flow chart describing how to determine the kind of analysis you’re doing and the error that is made if you mismatch the real and perceived question.

It’s a good quick read!