How to Lie with Statistics《For the sake of Statistics, be skeptical》

From time to time I receive "fill online questionnaire" requests from random people; and I am often curious about the statistic meaning behind those results. Basically speaking those summarized results are not statistically reasonable at all; partly due to the obvious fact that sample population doesn't represent study population.

This book was written 50+ years ago yet it is classy and old school. The author listed many pitfalls which could readily be ignored:

1) AVERAGE: are we talking about MEAN MEDIAN or MODE?

2) BUILT-IN BIAS: if you find out people brush their teeth 1.02 times a day will you consider the possible hidden truth that there are some people embarrassed by not brushing teeth so they lie.

3) ONE DIMENSIONAL PICTURE: unlike multi-dimensional charts numbers normally appear on the y-axises are shown by the sizes of objects. It is easy for people to see the improvements or decreases; nevertheless hard to tell the exact changes in terms of precise measurements.

4) STATISTICULATE: all sorts of tricks are used to deceive people suppose you earn $100/day and there's a 50% pay cut at first then one year later your boss increase your salary by 50% you get $75 instead of the original $100.

5) QUESTIONS WORTH GIVING A THOUGHT TO:

-Who says so?

-How does he know?

-Did someone change the object?

-Does it make sense?

p.s.The author mentioned in the book that Time magazine always has satisfying statistical graphs.