This post is a kind of extension to the previous “Solve problems with few, messy data” in which I exposed ways to overcome the lack of solid and sufficient data to start solving problems and improve.
I faced such a case with few, messy and incomplete data where I managed to gather some bits out of different sources. When checking for correlations and pattern similarity, some data from different sources were convergent while others contradict one another.
Instead of losing time searching for the causes of these confirmations / contradictions, there is a simple rule of thumb: if at least two sources tell the same, the information or data is assumed valid, if not we assume the information not trustworthy.
The rationale behind this rule goes back to redundant systems, designed to provide valid information even one of its subsystem fails.
In our case the rule may not withstand a thorough robustness check, but is good enough to allow quick decision with imperfect information.
What’s important during time-constraint process assessments, diagnoses or performance audits or even problem solving is not so much accuracy than sound logic to get things going.
Approximately right is better than precisely false.