Well, let's approach this is a way journalists no longer do--with academic rigor.
We are attempting to create a serious of sums based on differing data sets of differing values of precision.
Now, one approach could be to base our sampling where you use a mean of zero, this is handy inthat we can use that to help bias null entries in our data (which we can correct for, uniformly, at the end--we hope). While that builds a simpler data set, we still wind up with a lot of "fuzz," inaccuracy built in to it. And, by applying a correction at the end, that inaccuracy is increased. Which is less than desirable. (Refer back to the very flawed CBS news story that used NFA ownership to establish "gun ownership" on a per-State basis [and still gets used today, despite being utterly flawed].)
The other way to gin up the numbers is to use a n incremental system, which accepts a certain amount of inherent inaccuracy, that, should, over time, tend to reduce those inaccuracies. So, you use a constant source for data, and continue to add that data back upon itself. Perhaps that is just "raw" NICS data. (We just have to remember that data has built-in limitations, and wants factoring.) Mathematically, you typically apply a factor, like 3*SD, or SD², which "pushes" the extreme values out from "under" the area of the central, medial, distribution. The maths are elegant, and well-proven, which, generally, results in good data sets.
Ok, this has an issue in that its "full" accuracy is only achieved over long time spans--decades--so it's not immediately useful (which is highly undesirable for journos who think of a a week as a long time span).
That's why that latter method is much used for medical research, though, the value of the datset only improves with time.
Now, we also cannot escape the fact that we gun owners are, by and large, individualists with strong senses of how privacy and liberty are very much inter-linked. Which makes us contrarian almost by reflex. So, we are terrible subjects to get reliable data from (on any topic). Which is a double-edged sword. Politically, there is great strength in numbers* even as such precises enumerations could be used "against" us. So, it's complicated.
________________________________________________
*There is an excellent, if complicated, political debate that, having an imprecise number of arms bearing Americans is much like concealed carry. Caution ought be used as you never know who is ir isn't armed out there. Wher ethis becomes complicated is in our current era of political polarization, which may not care to be cautious at all.