01-01-2022, 03:00 PM
(01-01-2022, 02:49 PM)mavsluvr Wrote: I am also struck by the idea that tiny sample sizes are statistically significant. One of the cornerstones of statistics is that when you use sample sizes that are too small to be representative, it can and does lead you into wildly improbably conclusions. Partnow emphasizes something we probably all viscerally know -- that misleading data (such as that supplied by mini-sample sizes) provides a significantly worse basis for reaching conclusion than no data.
So if your wife cheated on you once you wouldn't find that "statistically significant" and would just move on with life as normal? How many times does she have to cheat on you for it to be enough data for you to pay attention to?
The point: Even a single data point is INCREDIBLY significant and is telling you something. Obviously MORE data is helpful, but oftentimes in life your data sets are small and you have to discern with wisdom how to interpret what is going on. The NFL deals with this ALL the time. Y'all are having fun mocking me for using a ten game set, but damn in the NFL that is a TON of data.