Within the volumes of big data there are often a lot of small bits of evidence that are contradictory to even clearly data-supported facts. Generally, this data noise can be seen as such and, in the context of the body of data, it is clearly outweighed. When data is selectively chosen from vast sources, however, a picture can often be created to support a desired view, decision or argument that would not be supported by a more rigorously controlled method.
Data confabulation can be used both intentionally and unintentionally to promote the user’s viewpoint. When a decision is made before data is examined, there is a danger of falling prey to confirmation bias even when people are trying to be honest. The term confabulation comes from the field of psychology, where it refers to the tendency of humans to selectively remember, misinterpret or create memories to support a decision, belief or sentiment.