Show treatment effect not due to chance (statistical power, tests of significance) | Large sample size | Much greater statistical power to exclude chance as an explanation in the sequential context |

To control for confounding factors | Randomization | Prior experience (everyone died regardless) |

Causation | Experimental change is causal | Plausible process change which can be replicated with similar results |

Speed of answer | Typically in years for a large trial | For a large change the results can be demonstrated for one patient |

Ethics | Would you volunteer for such a trial | No one foregoes the new treatment |

Use of knowledge from prior experience of outcome | Not used | Uses knowledge of prior experience |