Statistics: Get to Know Them

7. Define what you consider to be significant results before you become mired down in interpreting data after the fact.

8. When you test, isolate only one item to change in your test (e.g. carrier, reply, copy).

9. Be careful of projecting data in early days (before 28 days), because a number of items can affect the projections (e.g. number of Mondays, multiple mail dates, different post offices, actual days of return versus reported days of return).