Originally Posted by Mikecr
So re-reading the first section of the article, should I take this solely as 'use of statistics to draw conclusions with limited data'?
Perhaps if it were titled as such..
Because in my view, the article itself 'proved' nothing.
It went unqualified with regard to the title: "Air Temperature Effects On Muzzle Velocity"...
For example, in the opening statement: "It’s a well established scientific fact that air temperature influences muzzle velocity". Well, is this true? Is it something you've qualified, taken with contrast, validated, proven and defined? Seems a foregone conclusion otherwise.
If I setup a chrono and rest at a bench, while it's 0degF out, then quickly pull a gun & ammo out of truck, set it down and fire across the chrono, will velocity be different now than it was 6mos ago while it was 90degF out?
And what if it were both a 30cal AND a 22cal in such a scenario?
I suggest a prediction might not be as simple as it seems.
Does the heater work in my truck? Does it even matter?
What if I left a 30cal gun/ammo AND a 22cal gun/ammo on the bench at 0degF for one additional hour and fired them? And 3hrs?
What if I left the guns out, but kept the ammo in my pocket between shots?
Or if only the ammo was exposed?
This is an example of CONTRAST.
It's very important because despite statistic delusion, only the truth passes ALL tests. So you cannot know the truth until you've challenged it from many many angles and validated that it still does not fail.
If you define ONE validated truth with statistics, you should take care to qualify that as such.
It just seems a stretch to understand what we're to take away from your article, thats all.
Was it a plug for Patagonia?
One more time, this article is all about Data Regression
techniques, using the data most likely at hand, which is Air Temperature, used to derive MV as a result.
Any other issues, like the ones posted are not discussed, simply because that's not the central subject of the piece. That's the terrain of Interior Ballistics.
What the article explains and shows (and proves to some limited extent) is that a "Delta Spline" is the best possible method to infer interpolated/extrapolated values.
Of course, I'm not stating that other factors do not matter. Far from that.
And yes, to answer your question, it's a well established fact that Air Temperature affects MV...for the conditions given in the article.
The military (and some other independent labs) have a good number of tests (which are limited to certain border conditions) showing the correlation as stated.
In short, if you have measured MV/Air Temp pairs (KDPs) you need a technique to extract meaningful predicitions for values outside the tested range (a limited dataset).
If that's the case, well, this article exposes which techniques are best and why, in terms as simple as possible.
Now, you being the end user, is up to you to decide under what conditions the KDPs were obtained, and to repeat that same criteria when using the software.
The method(s) cannot infer nothing about the conditions, just provides the best possible math to derive a best "curve fit" to the sample data.
On the other hand, LoadBase 3.0 gives the end-user the tools to study and qualify the sample data, even to highlight "suspect" KDPs.
Statistics is a very powerful tool, but cannot compensate for erroneous criteria or bad judgement. Magical it's not.