[Verse 1] Picture scientists in white coats, testing theories all day long They need a method, need a framework, to prove if they are right or wrong First they state what they're assuming, call it null hypothesis Then they gather all their data, put assumptions to the test [Chorus] Null says nothing's really changing, status quo remains the same Alternative's the rebel shouting, "Something's different in this game!" P-value tells the probability, of seeing what we found If null were true, would we observe this? Numbers spinning all around Test that hypothesis, test that hypothesis Don't let assumptions fool ya Test that hypothesis, test that hypothesis Statistics are your ruler [Verse 2] Alpha level sets the threshold, typically it's five percent If p-value falls below it, null hypothesis gets bent Reject it when the evidence screams louder than the doubt But sometimes we make mistakes here, let me spell those errors out [Chorus] Null says nothing's really changing, status quo remains the same Alternative's the rebel shouting, "Something's different in this game!" P-value tells the probability, of seeing what we found If null were true, would we observe this? Numbers spinning all around Test that hypothesis, test that hypothesis Don't let assumptions fool ya Test that hypothesis, test that hypothesis Statistics are your ruler [Bridge] Type One error's like a false alarm, rejecting truth by mistake Finding patterns where there are none, believing lies that data fake Type Two error misses signals, accepts null when it's wrong Like ignoring smoke detectors when the fire's burning strong [Verse 3] In finance we test portfolio returns, do managers beat the street? Null assumes they're just average, alternative says they compete Gather months of performance data, calculate that crucial p If it's tiny, then we're confident, skill's what we truly see [Final Chorus] Null says nothing's really changing, status quo remains the same Alternative's the rebel shouting, "Something's different in this game!" P-value tells the probability, of seeing what we found Type One and Two errors lurking, when conclusions aren't sound Test that hypothesis, test that hypothesis Framework guides your thinking Test that hypothesis, test that hypothesis Statistical stars are blinking
← Point Estimates and Confidence Intervals | Common Statistical Tests: t-tests and z-tests →