Systematic error.
WHAT BIAS IS
A consistent distortion, rather than random error.
WHY IT IS WORSE THAN NOISE
More data does not fix it.
WHERE IT COMES FROM
Who is included and who is not How questions were asked Which records were kept Which period was chosen What you were looking for
COMMON KINDS IN BUSINESS
Analysing only current customers, and concluding customers are satisfied Surveying only those who responded Measuring only what is easy to measure Selecting the comparison that flatters
THAT FIRST ONE
Those who left are the ones with the answer.
WHAT TO ASK
Who is missing from this data? Would the missing records change the conclusion?
HOW TO REDUCE IT
Include the excluded where possible State what is not covered Look for evidence against your conclusion
THAT LAST ONE
The single most effective habit in analysis.
WHAT TO DISCLOSE
What the data does not cover.
