I still believe in the health department stats on covid, while they may not be 100% perfect and one can assume there's always more covid.
I don't because of my extensive experience in dealing with government metrics and reporting.
I trust that they know what they are doing, a lot of work goes into those stats by the pros.
I don't because I know better. When one sets up stats ( be it minitab, excel, NWA or whatever package) the guy running the numbers is never the same guy who builds the database to pull from ( just too much work)
Most databases contain 90+% of junk or unusable data to begin with because of the weak or missing controls on the input side. (I spend a lot of time analyzing this)
Then comes relevance- often the data doesn't directly fit the question being analyzed so it has to be "massaged" to make it fit ( as I pointed out yesterday in another post in the table of assumptions)
That's not counting the final step in how to build the comparison for publishing (selecting the question the data allegedly represents)
So, if you had a perfect database with 100% accurate and relevant data to begin with ( and you have a better chance of finding bigfoot and an alien riding nessie in the loch than finding a database with 50% relevant and accurate data in the first place)
By the time you cull and massage it- your literal accuracy is at best 30% ( and that's before the question tweaking and publishing starts)
That's why so many political polls are inaccurate- its not the statistical math- its all the data quality and massaging that throws it off.
That's part of the plan because the most "secure courier" of false information is the courier who has NO IDEA he is carrying anything so he gives off no signs.
People attack the statistics guy who defends himself and his methods but he is NOT the one providing the cooked data or manipulating it on the back side.