Cause and effect. And prediction.

[garandman]

Let's wax a bit philosophical about cause and effect in the firearms world. And its usefulness to predict outcomes for oursevles and others.

We all test loads, and bullet weights, and primer type / seating depth, and neck tension, and annealing and sorting brass and a dozen other factors.

The possibility for a false positive, or assigning an effect / outcome to a particular cause seems real. For instance, (and only as single example) we might test primer seating depth on our best load, and come to a conclusion that seating depth does / does not matter. How many times do we need to repeat that test? Should it be on different days / various different conditions ( ambient temp / humidity / altitude / baro pressure, etc)

And what is the possibility that our results aren't due to some other cause than we are testing - brass stress / failure, or an unkown bad wind read, hot / cold air pockets in the bullet path, some bad powder kernels, or just good old shooter error?

Then, what is the prediction value of what happenned in my gun being universally / conditionally / rarely true in other ppls guns? There a fair amount of dogmatism about our experience being of predictive value for other shooters in their guns in the firearms world.

How do you go about testing e.g. primer seating depth (or anything else) in your gun / reloads? How many times / varying atmosphereic conditions do you test in? What are your "falsification" methods? And what do you see as the predictive value for your test results in other guns / other shooters?


ETA: I'll answer for myself - I'll test my premise (e,g, CCI match primers are better than Federal match primers, JUST as an example) on three different days, in a variety of temps / humidities ) to see if I've *really* got something, rather than a false positive.

 

[butchlambert]

Let's wax a bit philosophical about cause and effect in the firearms world. And its usefulness to predict outcomes for oursevles and others.

We all test loads, and bullet weights, and primer type / seating depth, and neck tension, and annealing and sorting brass and a dozen other factors.

The possibility for a false positive, or assigning an effect / outcome to a particular cause seems real. For instance, we might test primer seating depth on our best load, and come to a conclusion that seating depth does / does not matter. How many times do we need to repeat that test? Should it be on different days / various different conditions ( ambient temp / humidity / altitude / baro pressure, etc)

And what is the possibility that our results aren't due to some other cause than we are testing - brass stress / failure, or an unkown bad wind read, hot / cold air pockets in the bullet path, some bad powder kernels, or just good old shooter error?

Then, what is the prediction value of what happenned in my gun being universally / conditionally / rarely true in other ppls guns? There a fair amount of dogmatism about our experience being of predictive value for other shooters in their guns in the firearms world.

How do you go about testing e.g. primer seating depth (or anything else) in your gun / reloads? How many times / varying atmosphereic conditions do you test in? What are your "falsification" methods? And what do you see as the predictive value for your test results in other guns / other shooters?

I don't check primer seating depth as I uniform the depths. I will not go so far as measuring primer height.

 

[DiffEQ]

Have you ever designed an engineering test or scientific experiment?
Identify the variables, control, or normalize the unknowns so that you can isolate the test to the variable under consideration.
Something such as primer seating depth at 100 yards is inconsequential. Stretch it out to 1000 where the accuracy requirements are <1/2 MOA, and you may see the difference.

There are easier ways to test this using sensors and a non-firing case, which then eliminate all the other environmental variables which are crowbars in comparison to the primer seating depth's toothpick, but who wants to spend the tens of thousands on equipment to do such a test?

Edit to add:
One simple way to see if it makes a difference:

1. Find a load that is tuned to your specific rifle, so you have a baseline accuracy expectation.
2. Then, narrow the candidates for the test:
   1. Sort brass for identical case fill.
   2. Sort that pile for identical primer cup dimensions.
   3. Select primers which all have identical geometry, cup diameter, and depth.
   4. Seat the primer to a specific force level
   5. Test to see if it affects accuracy
3. Then, using identical brass like selected above, mix it up with different primer manufacturers or seating force/depth.
4. See if there is an effect on accuracy.

2nd edit:
I'd hazard to guess you'd not find much difference. Consider that to be this experimentor's bias. What is important is that you are consistent in your primer selection and seating.

 

[garandman]

ETA: I'll answer for myself - I'll test my premise (e,g, CCI match primers are better than Federal match primers, JUST as an example) on three different days, in a variety of temps / humidities ) to see if I've *really* got something, rather than a false positive.

 

[drop_point]

There are ways to test with such a sample size with statistical significance in which we can, with confidence, accept or reject our null hypothesis. Once you understand statistical analysis, you'll have a better understanding of testing methodology. Science, bro.

 

[garandman]

There are ways to test with such a sample size with statistical significance in which we can, with confidence, accept or reject our null hypothesis. Once you understand statistical analysis, you'll have a better understanding of testing methodology. Science, bro.

I agree with that concept. I'm looking for specifc ways you do that. Pick a topic (e.g. sorting brass, or whatever you have tested) and lay 'em on us. Examples of how you came to confidence in your testing or identified a false positive will help.

Thanx.

 

[BoydAllen]

On your primer test, over the years I have seen a lot of what I thought bogus tests published. The tester held everything else constant except the primer, ignoring the idea that changing the primer might change the tune, requiring a retune to determine if the best accuracy obtainable with a given primer in that rifle was better, worse, or the same. This sort of adjustment is most easily done loading at the range.

 

[miningshawn]

I have learned in my field (mining) to never trust one result. We instead build a sample based on a trend. Sometimes a trend of 3 samples works, sometimes more (many more!) are needed to establish a baseline. No answer seems to be truly linear.

As a result - I cannot trust the results of a single Audette's ladder test. I understand the basis, and I understand the premise, but I cannot personally trust the result. Is round 5 a flyer or is it actual? Did I have a different seating force on round 2? What variables did I - the handloader - introduce?

To ensure that the results are sufficiently correct - I shoot two ladder tests. If the results differ to greatly from my expected parameter - I shoot three ladder tests and average the results. I establish a trend, average the results, and plot the curve. This reduces the inconsistency that I produced such as: point of aim, primer seating force, neck tension, seating depth, powder measurement, and so forth.

I personally look for nodes rather than perfection, knowing that I will introduce variables. If a node has a window of opportunity that stretches over 0.3 grains, is tolerant of temperature, is not dependent on absolutely perfect seating depth, and produces results of 0.5MOA or less then I have a winner. The rifle and cartridge system has now exhibited a standard of potential and all other problems at the target now rest with me - the shooter.

Were I to begin shooting benchrest the only changes I would introduce would be to control the parameters by whatever fashion I could - case volume, sorting by weight, custom dies, primer seating force, sorting by bearing length or diameter, and so forth. These are not steps that I perform because my chosen shooting disciplines do not "require" that level of accuracy. If I were to introduce them, then I would again revert to performing a series of tests rather than trusting a single observation.

I'm interested - but there is little opportunity for BR in my area, and I don't have the time to dedicate to it.

 

[drop_point]

I agree with that concept. I'm looking for specifc ways you do that. Pick a topic (e.g. sorting brass, or whatever you have tested) and lay 'em on us. Examples of how you came to confidence in your testing or identified a false positive will help.

Thanx.

It is important to only change one variable at a time. I'd start with something you know to be working...perhaps a load that is already determined to be in the node for both charge and seating depth. Its also important to recognize that the results are applicable only to the test barrel. I am a bit rust on my statistics, so I hope I'm coming through clearly.

First you must determine what you want to measure. Are you measuring Precision? Velocity? Score? Then determine a sample size to adequately represent the entirety of the population you want to represent, perhaps useful rounds of barrel life.

Then determine a sample size to represent that entire population with a confidence level you find adequate (there are formulas for this you can find on statistical analysis web sites). Your control group should be your existing ammunition. Your test group should be what variable you intend to test. It is OK to test on separate days.

I will pull some numbers out of the blue here (I'm not going to plug in the formulas right now to come up with "correct" numbers.

Let's pretend you want to determine if volume sorted brass will improve scores at 1000 yards vs randomly selected brass. The brass should be prepped in the same way minus volume sorting. The random brass will be control while the sorted brass will be the test subject. So you formulate the null hypothesis that volume-sorting brass improves F-Class scores at 1,000 yards.

Let's say your barrel life is 2000 rounds and you may use the formula to determine 30 rounds will give you the confidence you wish. Prepare 30 rounds of both the test group and the control group. Perhaps test ten of each per day. Perhaps shoot one round of control into one target and one round of control into a second target, alternating groups. After three days, you now have actual data to analyze and determine whether you can reject or fail to reject your null hypothesis. You can calculate mean, standard deviation, and median based on score. There are infinite ways to analyze the stats, but you could use a simple paired t-test to determine the likelihood that a given shot's score is related to the dichotomous variable; sorted vs unsorted.

I hope this helps, but really the best thing is to do some statistics research.



ETA: https://forum.accurateshooter.com/attachments/qp0405vandenbrandedoe-pdf.1283151/

this was in another thread, it may help guide you.

 

[butchlambert]

I think you fellers are bored. I think watching college football and then go out and learn to read your flags.

 

[pirate ammo]

another headache, durn.

 

[BoydAllen]

You fellows would be horrified by how I tune with a new powder. My PPC is a reliable rifle. It you feed it the right stuff and read the flags correctly, output will accurately reflect input, be it flag reading or load tuning.
On a morning when the wind is easy to deal with, shooting from a sturdy bench, with flags out, I will seat my bullets so that the rifling makes light marks, using about .002 neck tension (.262 neck) and start a one shot per load test in .3 gr. increments with a safe load that is well below what I expect to use. I shoot all shots on a single target, taking care to wait for the same condition, noting the position of each bullet hole as I shoot. Generally this will result in some holes being spaced out and others clustering. I run my test up to the point that I get bolt lift issues that indicate that I would not want to go that high for normal shooting, note all the details of the previous load, including ambient conditions (temp. and humidity), and then decide what my three shot test(s) will be based on the charge weights that give clusters rather than strung out bullet holes. I load up three of the best looking cluster, either the middle or average load depending on the number, and carefully shoot a group. If that works, I go to five, and if that looks good, I experiment with seating depth making small moves at a maximum (in the rifling) of .002. Generally I will be finished with a good load in less than two hours, having fired less than 40 rounds. All of this is done at 100 yards. You cannot imagine how amusing I find all of the internet accounts of long, drawn out load testing, that seem to often result in incomprehensible results, because the shooter apparently had serious rifle, bench/rest, and or shooting technique issues, not to mention lack of proper bedding, wind flags, and proper scope adjustment. If you are going to test one thing, everything else has to be right, and consistent.

 

[garandman]

I think you fellers are bored. I think watching college football and then go out and learn to read your flags.

On different days, but, yeah... absolutely...

 

[AckleymanII]

Boyd, I use the exact same technique on 7mm Rem mag and 7 STW's, with the only exception is to use .5g increment increase in powder charges. Last year, I had a new 257 Weatherby with .030 freebore, 9T. Shooting 115g Bergers, it took 17 shots total to find a load in the 2's, 2570 fps on a 25" bbl.

Yesterday, after cleaning an older SS Ruger Mark II All Weather in 270 winchester down to bare metal, I used the same technique to work up a load with 130g and 140g Nosler ballistic tips. I worked up loads shooting very small groups(sub half inch) with both bullets using H4831, Win brass, WLR primers. Key to the ballistic tips was to touch the lands.

Kiss the lands
Find pressure
Tweek the seating depth if needed
Change primers if needed

In the 270 above, changing from a cci 250 to a Win large rifle produced amazing results with all three shots in the 3's with the 140g and 4's with the 130's with a rifle that has already killed 150+ deer.

 

[garandman]

@BoydAllen ... Thx. I am certain my load testing protocol would be viewed as totally wrong by others. I start about mid range on the amount of powder , with 5 shot groups for that charge weight. I go up in 0.2 grain increments until I find a node.

I will test that load 3 times on 3 different range trips. (My falsification process) If it continues to group well then I start varying seating depths / jump to lands. That has always gotten me to between 0.2 and 0.45 MOA. ... The notable exception being AR15 loads which I have never been able to do better than 0.7 MOA.

I never sort bullets or brass or anneal or worry about primer pockets. What I'm getting is not probably bench rest grade accuracy, And my next step is prolly going to be to start fiddling with those things.

Another time I tried to test between C C I benchrest small rifle primer's and Remington BR small rifle primers. My test showed that CCI worked better with this load. As others noted I probably should have recalibrated the full load to actually make it a primer against primer test. I used all the same lot of brass, But honestly I probably did not do enough work to keep all the other variables the same so I would regard my results as suspect.

Point being actually testing 1 component against another component for accuracy involves a lot of work to make sure it is an apples to Apple's comparison.

 

[pirate ammo]

I load to enjoy myself, the same with shooting. I've loaded for awhile but for the shooting part really only the last year. I've shot a lot of pigs the last several years 3 inch groups at 100 yards were great. Now I'm trying to do better but when it becomes an obsession it's time to quit. I done a 1 inch group at 300 yards last Monday
and I'm going to build me a frame for it and hang it on the wall. For me it's always what the hell lets try this.

 

[KMart]

Somebody needs to go and anneal some brass. They are costing themselves money.

 

[mikecr]

The first efforts would go to understanding something well enough to isolate it for testing.
With that and objective thinking, a limited conclusion should be within reach.
'I never cared about it, so it doesn't matter' goes nowhere (even if they set out to prove that).
If they don't care about it, it's VERY likely they don't understand it to begin.

 

[drop_point]

Somebody needs to go and anneal some brass. They are costing themselves money.

I can't imagine why you WOULDN'T anneal your brass every firing for the life of the brass and the consistency of process.

 

[bluealtered]

O, boy, ... I learned many years ago at great expense in racing that you can not compensate for every single thing you are going to encounter at every meet no matter who you are or how much money you spend.

The only thing you can do is build a solid foundation with good equipment and pay attention to others and what is going on around you, and yes that means thinking for yourself. If you fall for every trick of the week you may luck into a win once in while, but more likely you will just end up with a large pile of expensive stuff laying around in the corner of your shop that no body wants anymore. Think first, spend second.