Lyric copyright by Joy Jordan
Across the nation
There is a need for standard deviation
For explaining variation
For measuring correlation
But don't always believe what they say
in that USA today
'Cause you know that correlation don't imply causalateh
Is it a random sample?
Is the number of people ample?
Or is the response voluntary?
If so, this could get hairy!
Is the sampling biased?
Is the variation vast?
Then the results aren't legit
And we can kick your es-timate.
Chorus(think Eminem's "Lose Yourself" chorus):
Lose yourself in the data, the theory
You want it, you know you'll never let it go -- whoa.
Stats class is helpful, is necessary
You'll use it everywhere you go -- whoa.
Now here's a story
Of the residuals' glory.
Do the errors seem to fan?
Then rethink the regression, man.
Outliers can have an effect
That you can't neglect.
So first graph the data --
You'll be happy lay-tah.
Hey -- we have the same birthday.
But we might see an effect that portends of consequence
When actually it's coincidence
Can we really be that dense?
The meaning of life, you ask?
I'm up to the task.
Hear the roll of the drum...
It's the Central Limit Theorum.
Now what if we want to estimate the mean of the population?
Do we just give an estimate, or is there more explanation?
If you say the confidence is in the interval not the method,
Then you haven't learned the lesson -- which is what I've dreaded.
When you're hearing someone rant
About a result that's significant
Ask if the assumptions are met
Otherwise it's probably crap that you'll get.
If you've been studying all term
Then there's no reason to squirm.
You should have this stuff down pat,
'Cause statistics is where it's at!