Errors that produce wrong conclusions.
CONFUSING CORRELATION WITH CAUSE
Two things moving together may share a cause, or be coincidence.
LOOKING ONLY AT TOTALS
An average hides the distribution beneath it.
IGNORING THE DENOMINATOR
Fifty complaints means nothing without knowing how many customers.
SMALL SAMPLES
Conclusions from a handful of records.
SELECTING THE PERIOD THAT SUITS
Choosing dates that show what you hoped.
COMPARING UNLIKE PERIODS
December against February, in a seasonal business.
SURVIVORSHIP
Analysing only customers who stayed, and concluding customers are satisfied.
IGNORING WHAT IS MISSING
Absent records are data too.
ANCHORING ON THE FIRST NUMBER
Then fitting everything else around it.
CONFIRMING WHAT YOU EXPECTED
Looking until you find support, then stopping.
WHAT PREVENTS MOST OF THESE
Asking what would disprove your conclusion, and checking that too.
