|
[Boom, chukka, boom]
Detect an effect
that isn’t really there.
It happens a lot,
oh yeah, like you care.
That’s Type I, baby
—the test statistic’s big.
But do you know what?
It only looks sig!
Type I error!
Uhhh! False positive!
Uhhh! False positive!
You say to yourself,
“Alpha level too high . . .
gotta crank it down
so very little gets by.”
—But now the dust settles
and you’ve got a new mess,
’cause with alpha too small
you might not see an effect that really is
there. [Rhythm disintegrates temporarily at this point.]
Type II error!
Uhhh! False negative!
Uhhh! Low power!
Booooo!
|