Smith College Haikus II

P-values are great.
They help us to reject nulls
or fail to reject.

Simpson's paradox
Is only the beginning
Getting dangerous

Regression's wonder -
Multiple in nature with
"y" against "x"-es

standard error or
standard deviation or
root mse hmmm

Am I near normal?
Is my n large enough to
make inferences?

Statistics is fun
Interpretive dance of math
Useful for research

Found correlation
Careful, it's not causation
Variables may lurk

Models recently
Are getting frightfully thin
They fail the F-test

Stats are not psychic
Can't go beyond our data
Don't extrapolate

Shape, center, spread, range
I would like a picture please
describes intro stats

To be normal or
not to be normal is the
major question here

r-squared adjusts
Check model's validity with
"p" "se" values.

Null hypothesis
Reject or not to reject
I am on that quest

More than one outcome
is possible, thats why we
do a two tailed test.