Jim, when in the process are you cleaning the barrel?

Denton,

I will try one last time to explain this so you can understand.

I believe you base your argument that a 1 inch rifle will shoot .5-1.5 inch groups on the work of Audette (and possibly others). That work says that one can estimate long term gun performance on limited groups. For example, one can expect, based on one 5 shot group, to have long term average performance be between .55-1.45 times the group size. If the group size is 1 inch, you can expect the long term group average to be between .55-1.45 inches. I assume this is the basis of your .5-1.5 inch range for a 1 inch rifle.

The more groups you shoot the more accurate the estimator is and the more the multiplier drops. The more groups that are shot, the more the multiplier approaches the actual group spread performance of the rifle. And it may well be much less than the .55-1.45 multiplier. NOTE: There are many other acceptable spreads in Audette's model than .55-1.45.

Read this carefully- it is very important!!! This tool is used as an ESTIMATOR of future performance based on limited groups when the long term performance is NOT KNOWN. It is not used to compare groups to groups or past performance to current.

Once the gun's performance is known, you use tools such as the mean and standard deviation to identify and troubleshoot performance. The mean, of course, is the average of all the groups shot and the standard deviation is one method of describing how much the groups vary. You can use these to compare groups to groups or past to current performance. You may well find two guns that have the same average but very different standard deviations. For example, you might own two guns that average 1 inch groups but one does it with very consistent groups and the other does it with rather loose groups.

From Denton:

Quote:

---

I keep saying this, and it's true: If your rifle's long term average for many five-shot groups is 1", your five-shot groups will vary from .5" to 1.5", just due to normal, random variation

---

By your own statement, you know your gun's long term performance. You then put the long term average into Audette's estimator as an individual group size. This violates the estimator model's assumptions and voids the model. You have mixed apples and billy goats. The model just plain does not apply in this situation. You cannot take the deviation ranges listed in the estimator and arbitrarily apply them to historical data in place of the actual historical variation.

To summarize for you:
1) The estimator is to be used only to predict future performance based on limited data before the long term performance is known,
2) To identify and troubleshoot performance problems, other tools are used, and
3) Many firearms can have the same average group sizes but very different standard deviations.

Hi Jim.

I'm not going to get all physics/ballistics/etc.

Do yourself a favor, try varying the seating depths first. Make SURE the barrel is cool enough between shots. Make sure barrel is COLD before trying the next 3 or 5 shot groups. That way you are giving each group a fair shake.

I saw where someone brought up an excellent question. IS it indeed ALWAYS your 3rd shot that is the flier???

I have a 7 mag using 150 Sciroccos and IMR4350. Groups great, both 3 and 5 shot strings.

However, when I first began loading and was getting close to the sweet spot, I noticed shots 1 and 3 were touching and 2 was traveling.

I still think adjusting the seating depth is the first step.

I understand perfectly well what you are saying. The problem is that there is a huge flaw in your model, which you have not yet recognized. Absent that flaw, what you say would be correct.



You are apparently accustomed to working with processes where the mean is independent of the standard deviation. That seems to be the root of your misunderstanding. That does not apply in this case. You cannot choose a model where the gun shoots 1" groups, with a standard deviation in group size of 1/16", because it cannot exist in nature.



Group size represents a two dimensional standard deviation. For a given sample size, standard deviation is intimately connected with range (group size). Once you know the distribution is normal, which we have from the Central Limit Theorem, and the number of shots fired per group, and the size of the group, the standard deviation of the group size is already chosen for you. You might think about the Binomial Distribution, where standard deviation is completely determined by the mean and the sample size. Your statement 3 is incorrect.



Of course, as the number of groups fired gets larger, the Central Limit Theorem also changes the limits. That is perfectly obvious, and I have not said otherwise. One problem, however, is that the distribution of group sizes is non-normal, so some of it gets a bit untidy.



Trust me on this one. I teach advanced applied stats for a living.

Denton and TheBoss, You guys are reminding me of why I slept through too many of my statistic classes. Statistitions full of fact and theory signifying ?...... .......DJ

dj, it's actually worse than that!!

A few years ago, I decided I needed to find something that I would enjoy, that gets me outdoors more. I remembered that when I was 16, it was great fun to hunt jackrabbits with a 22, so I bought a 223, and started accumulating guns. Now, I think of things that puzzle me, and design experiments to get answers. I'll spend an hour at the range, and three hours analyzing the results. The whole thing has gone somehow horribly wrong.

I don't blame you for not liking stats class... I've got a real attitude about most of them. They take perfectly simple, elegant notions and wrap them in such cryptic language that they defy understanding.

As to the intense discussions, it is in the nature of statisticians. Did you know that statistics was invented so as to provide employment for those of us that are handy with numbers, but lack the personality to go into accounting?

My .243 (very thin barrel) has always shot groups with one flier. I dont think its recoil, and the trigger is sufficent for big game accuracy, so I dont think I'm jerking it. I always chalked it up to barrel heat, but firing with 30-45 seconds between shots, or firing with several minutes between shots, seems to make no difference.

May have to try some of the suggestions here.