I had a fun lesson today in Statistics. We just started talking about experimental design and actually how the statistics works (fuck off linear models, we're moving to important things now) and I really wanted to leave a good impression for when we start talking about confidence intervals and p-values. So I came up with an activity to do.
The big question of the day that I wanted to answer was, "How do we know that an event is due to factors OTHER than random chance?" The best example I could think of was a question I had on a statistics test. "How many times would you have to a six on a six-sided die out of 120 rolls in order to be 95% confident that the die is loaded?" Now, since I'm a good gamer, I don't have any loaded dice. But I can force a card.
 |
| Pick a card, any card.... | |
So I forced the card and then later in the lesson I revealed the card. How can we be certain that I wasn't just plain lucky? It spawned a fun discussion and realization that, no matter what the situation and what the sample, there's a small random chance that the unlikely happens. And thus, we never "prove" or "disprove" anything with statistics. We can only be 99% confident.
Statistics is fun.
No comments:
Post a Comment