Big Bang or Big Bust? The Role and Implications of Big Data in Evaluation
Touring the small but rapidly growing literature on Big Data analytics reveals a pervasive move toward what might be termed a data-driven empiricism. Big Data, it seems, as in the context of evaluation becomes almost mythological, something yet to be seen, yet something dearly desired, something to be both feared and befriended. More than that the application of preconceived theories, models, and hypotheses only serve to unduly restrict the analyses and the knowledge produced. Closely related to the first claim proclaiming the end of theory, some Big Data advocates have been quick to rejoice the shift from causation to correlation. In the example, the correlation between smoking and heart failure may be positive or negative or zero. And if circumstances are likely to change, even Big Data may be seen as a poor representation of the total population. Big Data is messy in the sense of originating from non-evaluation prompted behaviors and from being inherently fluid and fleeting.