I agree that data can't be selective.
But I question when people include old data. Is 2015 & earlier TV viewership #'s relevant to projecting 2023 and later? Sure there might be situations where 5 years of data hurts a school like Florida State or USC which have been down recently, but have a strong national brand over the last 30 years.
I guess I would rather look deeper at the exceptions vs. the broad brush of 10 years of viewership history for all schools. If TV viewership logistics were the same today as 2012, then maybe 10 years would be reasonable. But I think how people watch TV and competing viewership options (Netflix, Gaming, etc.) would seem to support a shorter window.
But in the end, I not sure the networks care about viewership vs. market size where schools are located and brand name.
I do think with how much things have changed over the last decade in how people get their TV its hard to nail down the actual numbers on some of this. I also think the trend on CFB and on some teams has changed a lot.
TV and streaming has changed the game so much, that I think some of this should be shown in trends not straight up stats. Especially when talking about data from 10+ years ago.
This is why I think what would be better is when they use these kinds of data, they a need to compare, FS1 to FS1, and ABC to ABC, not ABC to FS2 like most do. I also think they need to add a trend stat. So if ISUs numbers are on an upward Trend that should be highlighted, and by how much. Similarly if Wazzu is on a downward trend.
I also dont think excluding 2020 is really a necessary variable, considering the entire country went through the same thing, if one conference or school did a better job during that time, it should reflect on that program as a positive as well. But in the grand scheme of things when discussing viewership numbers, it really should not matter, it still about the average viewers during those games. And everyone TV views should have went up, regardless if they had 5 games or 10.
Too many of these stats not only dont show the entire picture, but then are manipulated to show the narrative the author desires, because of biases.