Measuring Group Sizes & Statistical Methods

[Jumping Frog]

Does anybody else read the gun magazines? Well, I read too many of them, and there was an article last winter I am trying to remember. Maybe you can help with identifying the magazine/issue, or pointing me to an online copy of the article (I have searched without success both my physical world and the online world).

The basic article premise was that the "gold standard" for measuring group sizes has traditionally been shooting a total of five groups of 5 shots and measuring average group size. There are two disadvantages to this method.

First, when you are talking 25 shots total, unless you are using a ransom rest or other mechanical means, at some point the human variableness of the shooter becomes statistical noise in your measurement: you are not measuring the inherent accuracy of the load/firearm, you are also measuring the shooter. That made sense to me in that very few people can be absolutely consistent for 25 shots.

Second, if you are working up a load to compare 3 or 4 different powders with several weights per powder brand, you could easily find yourself needing to shoot 12-16 different series of five 5 shots groups. Needing to shoot 400 rounds to settle on a load is time consuming, expensive, and further creates issues with the "human noise" problem.

The alternative approach: The article said it had found a different means to find an optimum load in a statistically valid way. Here is where my memory gets fuzzy. It was something like shoot a 3 shot group. Measure the distance between shot 1 & 2, 1 & 3, and 2 & 3. Average the distances. If the maximum distance was greater than some percentage of the average, then you had a statistically invalid shot. Shoot the 4th shot and perform similar measurements. If there was still an outlier, shoot the 5th shot and perform the math.

It claimed that ultimately this method allowed you to accomplish comparing loads using 5 shots per load instead of 25 in a manner that was actually more statistically valid.

Unfortunately, I can't locate the article to try out the method. Anyone?

I also figured even if you don't know the specific article, this could be an interesting conversation in it's own right.

 

[sonier]

i duno but how would you deterine what bullet hole was your first, second and etc.?

 

[Jumping Frog]

how would you deterine what bullet hole was your first, second and etc.?

I was labeling them 1,2,3 just to distinguish the different holes. I do not believe the actual order they were shot makes any difference -- they could have just as easily been named A/B/C or Alpha/Beta/Gamma.

 

[Walkalong]

The human element is involved either way. It is more difficult to shoot good 5 shot groups for that very reason. Five shot groups are a better indicator of the rifles capability, if the shooter is up to the task. The other part of the human element is to be able to truthfully call bad shots on ourselves and not blame the rifle. That takes practice as well. Many folks like to blame equipment when it is them, and others make excuses for the equipment, when they should be blaming it on themselves.

I can take my Bench gun, capable of shooting 5 shot groups in the zeros, and shoot bad groups just as easily as good groups. Shooter error. The rifle never changed for the good and bad groups. I did. We never worked up loads with rifles bolted down. We had to do it while factoring in our human error. We had to be able to tell when the rifle was trying to shoot, but we were holding it back, and when it was the rifle, and we needed to change something. Then of course there was when both the rifle and we were shooting well. That is when you see ones and zeros at 100 yards.