I'm saying that statistically no one gets hurt by McDonald's coffee. If it were a true problem, you would have seen MUCH higher numbers of people being seriously burned. If you can't get out of the noise fluctuations, you have no statistically meaningful data. That's how it works. The real problem that happened with McDonald's is that they had someone on the stand who said essentially just that. 700 people over 10 years got burned, but McDonalds sells over a 10 billion cups of coffee in that time. So you're talking 70 in a billion chance. It goes to show the danger of the product. 70 in a billion is essentially zero. It wasn't a problem. If the coffee was so hot to cause burning or the cup somehow defective, you wouldn't realize a 70 in a billion chance; it would be higher.
Here's the thing though, someone from McDonalds says what I just said. It's factually correct. But the jury hears "OMG, he just said 700 people is zero!!!! WTFOMGBBQ!!!!!one!!one!!!". They freaked out and acted emotionally after that, exactly the opposite of what they are supposed to do. 70 divided by a billion is essentially zero. There will always be people burning themselves somehow or doing something zero, you will rarely realize an actual zero if there is a probability of something happening. Given enough time, all probabilities work out. However, if the coffee was a threat they claimed it to be, you should statistically be above noise; meaning that you would see a statistically significant number of burns in this case. 70 in a billion is not statistically significant.