Thread: 3 shot groups are useless!

Originally Posted by Kiwi Greg

So what I'm getting at here is my light rifle & arguably my heavy rifle (which normally shoots better) are genuine 0.1 moa 5 shot rifles but only in the right conditions without an idiot driving them

I've not felt the need to post further on this thread, what with Gimp holding the corner so well on his own. However, and perhaps surprisingly, I find that it is Greg's post that I keep coming back to as the one that is going to finally prompt me to hit the keyboard again, as being in need of some clarification in my view.

The general theme of the thread is about what constitutes statistical significance, and what statements on performance can fairly be made about a rifle from a given set of target data.

The conclusion that Greg’s two 6PPC benchrest rifles are both 0.1moa performers CANNOT be drawn from the targets he has posted. I hope that that is clear, because I didn’t think that the way it was presented, and in the context of the other posts, that it necessarily was?

Whereas other members have posted sets of targets and asked the question “what is the capability of this rifle when these targets are taken in isolation”, I wanted to point out that in Greg’s post the targets are not to be taken in isolation. He is drawing on a large amount of prior knowledge of his rifles that we are not being made privy to in making the statement that both these rifles are “genuine 0.1moa 5-shot rifles”.

He is in the very unusual position of having access to something of a performance benchmark in these two rifles and where groups that exceed a previously established baseline sizing can then reasonably be attributed to causes other than those relating to the rifles alone.

It becomes a demonstration that when we are approaching an assessment of a rifle from the opposite direction — where we have little knowledge of the rifle’s actual capability and are hoping to interpret a set of targets to draw our conclusions — that amongst other factors, the influence of the shooter and environment will be contributors to the group size and mean POI. While we of course accept this, such separation is rarely able to be demonstrated as it has been here.

Originally Posted by Puffin

I've not felt the need to post further on this thread, what with Gimp holding the corner so well on his own. However, and perhaps surprisingly, I find that it is Greg's post that I keep coming back to as the one that is going to finally prompt me to hit the keyboard again, as being in need of some clarification in my view.

The general theme of the thread is about what constitutes statistical significance, and what statements on performance can fairly be made about a rifle from a given set of target data.

The conclusion that Greg’s two 6PPC benchrest rifles are both 0.1moa performers CANNOT be drawn from the targets he has posted. I hope that that is clear, because I didn’t think that the way it was presented, and in the context of the other posts, that it necessarily was?

Whereas other members have posted sets of targets and asked the question “what is the capability of this rifle when these targets are taken in isolation”, I wanted to point out that in Greg’s post the targets are not to be taken in isolation. He is drawing on a large amount of prior knowledge of his rifles that we are not being made privy to in making the statement that both these rifles are “genuine 0.1moa 5-shot rifles”.

He is in the very unusual position of having access to something of a performance benchmark in these two rifles and where groups that exceed a previously established baseline sizing can then reasonably be attributed to causes other than those relating to the rifles alone.

It becomes a demonstration that when we are approaching an assessment of a rifle from the opposite direction — where we have little knowledge of the rifle’s actual capability and are hoping to interpret a set of targets to draw our conclusions — that amongst other factors, the influence of the shooter and environment will be contributors to the group size and mean POI. While we of course accept this, such separation is rarely able to be demonstrated as it has been here.

It is interesting and I have also been thinking about it.

The question that arises is "what is meant by genuine 0.1moa 5-shot rifles ?"

The definition of that is rather important.

Is it -

- Rifles that will occasionally shoot a 0.1moa 5 shot group, as part of the natural variation in group sizes (but with an average group size larger than that)

OR

- Rifles that produce, on average, 0.1moa 5 shot groups (meaning that the natural variation results in both larger and smaller groups from time to time)

OR

- Rifles that can be characterised as having sufficient precision to produce a high hit probability on a 0.1MOA target (disregarding the other aspects of hit probability)


Of course, the rifle itself is not the entire picture, the shooter error is part of it.

what would be interesting would be to see the difference in average group size (or mean radius of all shots in those composite groups) between Greg vs, in his words - someone not an idiot driving them. (not having a go Greg !)


It is interesting to observe that an average 5 shot group size for a benchrest rifle being shot by an expert is over .25MOA. Makes the idea of .25MOA hunting rifles seem vanishingly unlikely in reality.

It would also be interesting to plot all shots and understand the overall actual precision - as the average of multiple groups is less informative than one single large group

Originally Posted by gimp

It is interesting and I have also been thinking about it.

The question that arises is "what is meant by genuine 0.1moa 5-shot rifles ?"

The definition of that is rather important.

Is it -

- Rifles that will occasionally shoot a 0.1moa 5 shot group, as part of the natural variation in group sizes (but with an average group size larger than that)

OR

- Rifles that produce, on average, 0.1moa 5 shot groups (meaning that the natural variation results in both larger and smaller groups from time to time)

OR

- Rifles that can be characterised as having sufficient precision to produce a high hit probability on a 0.1MOA target (disregarding the other aspects of hit probability)


Of course, the rifle itself is not the entire picture, the shooter error is part of it.

what would be interesting would be to see the difference in average group size (or mean radius of all shots in those composite groups) between Greg vs, in his words - someone not an idiot driving them. (not having a go Greg !)


It is interesting to observe that an average 5 shot group size for a benchrest rifle being shot by an expert is over .25MOA. Makes the idea of .25MOA hunting rifles seem vanishingly unlikely in reality.

It would also be interesting to plot all shots and understand the overall actual precision - as the average of multiple groups is less informative than one single large group

Haha - even .75 is nearly vanishing rare Id say.

But even experts get sucked in. I just watched Desert Precisions latest video, normally Kenny us pretty sound. But here are two consecutive 10 shot groups with a small but significant POI shift between them. Oh dear! And the irony, he's looking at 22 Ammo SDs with only 10 shot sample sizes

Originally Posted by Tentman

Haha - even .75 is nearly vanishing rare Id say.

The single groups posted recently with some larger sample sizes (10 rd) on this forum indicate that you're right - considering that you should expect about 30% variation in group size up/down from the average size of a 10rd group... and the "actual" precision is the high end, and better represented with 20-30rd of data

Been some really good groups that people have posted, but all around the .7-1.5moa for 10rd.