Ah! The problem of more reporting leading to less information!
Let's do a thought experiment:
Our National Home Economics Standards Board [NHESB] (less formally known as the Home Eck Board) sets a requirement that all home bakers in the USA must report the amount of gluten they use each week in their Betty Crocker recipes, and how many dozens of cookies they produce each reporting period. After a few months this settles into a pattern where some bakers consistently report the amounts in grams while others consistently use apothecary grains. This is workable because the bright boys at the NHESB recognize a high gluten:dozen ratio infers that the gluten is being reported in grams and they do indeed know how to use a gram to grains conversion table. It is possible to determine over time whether American home cooked cookies are becoming more or less gluten free. No problem.
But then the NHESB changes the reporting requirements. Now the gluten needs to be reported by volume rather than weight. But again the mode of measurement is not specified. So some home bakers report in ounces, but some are using dry ounces while others are reporting in liquid ounces. And still others are using the troy ounce numbers from their kitchen scales to determine the ounce volume of the standard dissolved flour solution they pour into the cookie dough. And then there are households where Mom bakes a lot of cookies, who report in pints--- but are these wet pints like so many per liter, or dry pints like so many per bushel?
Everyone is reporting their usage accurately, but without any standard measure, there is no way to assess whether American kids are getting more or less gluten in their cookies. It is a mess. And when you start talking about cookies that are warm (or even hot!) from the oven, it is, in deed, a hot mess.
And don't get me going about the NHESB's treatment of daily sodium ingestion levels....
Coming back to the original problem:
A sure way to fuck up the ability to do reasonable statistical studies over time is to force arbitrary changes in the measuring process without fully documenting either the old way or the new way. To say it differently, we are often better served by reporting that is self-consistent over time than by increasingly detailed reporting that sacrifices consistency.