3 4 2 Simpson's Paradox

[brendansob]

One of the better vids explaining this. I like how you do not jump to a conclusion about the data - that's the whole point! Based on the vagueries of sample sizes and aggregation, would you conclude that a) Lowell is the better batter; or b) Ellsbury is the better batter, BUT he batted less in 2007 when perhaps batting was easier (e.g. poorer quality pitchers) thus effecting his overall average??? This is the key to interpreting data effected by Simpon's Paradox in my opinion.

[Noxoreos]

Are you referring to real numbers in the example? Those numbers are important to know for the conclusion and should be pointed out before saying someone is wrong. Without the weights those percentages are just that: percentages. Without weights no one is able to make a conclusion, except of assuming that the weights of those percentages were equal. And in that case that would actually be right and the paradox would not apply.