Why wouldn't you throw out data points that appeared to be anomalous?

You would not throw out date points that appeared to be anomalous because you are only entitled to abandon outlier measurements when you have a good reason for believing them to be false. Hope this helps!

░░░░░░░░░░░░▄▐░░░░░░ ░░░░░░▄▄▄░░▄██▄░░░░░ ░░░░░▐▀█▀▌░░░░▀█▄░░░ ░░░░░▐█▄█▌░░░░░░▀█▄░ ░░░░░░▀▄▀░░░▄▄▄▄▄▀▀░ ░░░░▄▄▄██▀▀▀▀░░░░░░░ ░░░█▀▄▄▄█░▀▀░░░░░░░░ ░░░▌░▄▄▄▐▌▀▀▀░░░░░░░ ▄░▐░░░▄▄░█░▀▀░░░░░░░ u have been spooked by the ▀█▌░░░▄░▀█▀░▀░░░░░░░ ░░░░░░░▄▄▐▌▄▄░░░░░░░ ░░░░░░░▀███▀█░▄░░░░░ ░░░░░░▐▌▀▄▀▄▀▐▄░░░░░ spooky skilenton ░░░░░▐▀░░░░░░▐▌░░░░░ ░░░░░░█░░░░░░░░█░░░░ ░░░░░▐▌░░░░░░░░░█░░░ ░░░░░█░░░░░░░░░░▐▌░░ send this to 4 ppl or skelintons will eat you