In reply to DaveHK:
I assume the point is that there's the dataset that you need in order to answer the question, and the dataset you have available to you, and that sometimes people don't really recognize the distinction.
I've not watched this video, but as an alternate example - people sometimes try to use the logbook stats to answer what the average lead grade is (the average max lead grade, that is). The average sport grade in the logbooks is flat at 6a+ over many years. But that's the average of routes logged, not of people's max lead grade. If I look at my own data, I've climbed in the 7s every year for about the past 15, but my average grade climbed each year, which is what is feeding the 6a+ average, has fluctuated in the 6b-c range for most of that time. You can analyse the data all you like, but fundamentally it's the wrong data for the question you're trying to answer.
Or as another example - the same poster did a video analysing whether 7a or 7A was harder, right? There's no way for any well rounded climber that the latter is easier. If the data suggests otherwise, it's data that's from a biased sample or otherwise the wrong data to answer the question.