At which point do you feel comfortable ignore outliers?

P

Pond James Pond

New member
When calculating MOA for group size people often say count the two furthest shots apart in a given group.

So in a 3 shot group, where two are touching and one is 3" away you still include it.

However, in a 10 shot group where 9 of them are tightly clustered in a huddle of about 1MOA (hypothetically), and a tenth shot lies some 2-3" away, is it not reasonable to class that 10th hole as an invalid shot given the ratio of cluster and flier?

If that holds, at which point do you decide that you group is such that you can ignore one of the hits when calculating your group size?

Or do you not agree with that premise?
 
It depends on if I'm evaluating my shooting ability or the firearms accuracy.

If I'm evaluating me I would include it, but if I was verifying a load for a firearm and had 9 shots clustered together and 1 that was away from the group, I would probably conclude that the flyer was my fault.

Koz
 
How about this; if you have nine tight ones, and a tenth that is oughta the guoup, lay another one in there so that you (hopefully) have a nice tight ten shot group. Call the bad shot your mulligan, and see what your buddies say.:p

Three shot groups are -- just not enough in my book. jd
 
This question points out the disadvantages of using group maximum spread as a measure of ballistic accuracy (technically, it's precision that we care about but I know that most shooters tend to refer to it as accuracy, so I will as well). Measurement of maximum spread has only one advantage: it's easy to do. Other than that it's a terrible way to evaluate accuracy because it's directly correlated with the number of shots in the group (i.e., as you add shots the group can only become larger, never smaller) and uses information from only two shots in the group (the ones furthest apart) no matter how many other shots the group contains.

A much better way to evaluate accuracy is the way the military does it, by calculating circular error probable (CEP), which is the diameter of a circle that, statistically, will include 50% of all shots (that's ALL shots, not just the ones you've actually taken). All data, i.e., all the shots in the group, are used in the calculation and adding more data (more shots) can make the CEP larger or smaller, but will always improve the estimate. Calculation of CEP may be beyond many shooters, but it can be approximated (to about 93%) by placing all the shots taken onto an x,y coordinate system, calculating the distance of each shot from the center, and then taking the average of those distances.

The actual calculation is much easier than that may sound. I use one of those sighting-in targets that's pre-printed with a grid of 1" squares. The aiming point becomes 0,0 and then it's simply a matter of measuring the distance in inches horizontally (x) and vertically (y) of each shot from that point. The center of the group is the average x, average y, and the distance of each shot from the group center is calculated using the good-old Pythagorean theorem. Unless you're targeting ballistic missiles, the average of those distances is close enough to the actual CEP.

Much more information here: www.statshooting.com/papers/measuring-cep-mcmillan2008.pdf
 
Last edited:
Basically, you want to dampen the effect of the occasional outlier(s), or "flyer(s)".

The post that describes the military's approach (calculating circular error probable (CEP)) is probably the most valid, but how many people are actually going to go to the bother, or get it right?

The approach I thought up, without any particular stats justification, is to use 30 rds. on 6 targets, and then just average the size of the 6 groups. I use the mean, but using the median (half above, half below) would further dampen the effect of outliers. And, "outliers" would pertain to luckily small groups, and unluckily large groups - not just an outlying individual round on a single group.

That much shooting means you're going to be getting a hot barrel unless you take a long time to do it all, but for some real life usages that might increase the reality of the test.
 
Pond, James Pond asked:

At which point do you feel comfortable ignore outliers?

--------------------------------------------------------------------------------

When calculating MOA for group size people often say count the two furthest shots apart in a given group.

So in a 3 shot group, where two are touching and one is 3" away you still include it.
Click to expand...


YES you include it!

Then work to do the trouble-shooting to figure out what caused the outlier:
1) In a 3 shot group - Does it always occur.
2) If so, Is it always 3" away at same relative clock position to the other 2?
3) If not then, Was it you; Was it a cold barrel; Was it a clean barrel?
4) etc., etc., etc.

Then , if you can figure it out - Work to eliminate the root cause so it won't happen again.
 
I'm only shooting for myself, not competition. I "normally" measure the tightest rounds in the group, center to center, but will include the fliers also; Like one of my better 308 loads gets "three into 7/8 with one out to 1 1/8". I record all the factors when shooting a group and have some "bad eye days" where my marksmanship is somewhere out in left field. And since I'm just recording results to find a good load, I rarely rely on one or two groups to determine a "good" load...

Finding a "good" load takes me a long time, but I enjoy every minute of loading, shooting, recording, comparing results with other loads and starting over next week...
 
Just to clarify, I asked myself this question today as I tested some loads for my .223. All my test groups were 5 shots at different targets.

On one particular charge weight I actually had 4 shots in what looked like a 1.5MOA cluster. I was about to shoot the 5th and instructed myself not to screw up, but concentrate. Murphy turned up and the 5th shot flew low by about 3 inches, so that possible 1.5MOA result was dead in the water.

Probably my fault as I focussed too much on the importance of that last shot.

I will be including it in my calculations, but it got me thinking about when it is reasonable to say "surely that one must be an exception if all the others are together...!"

Would it in groups of 10 and over, or only 20 and over etc? I don't know.

Hence my question.
 
To take you out of the equation as much as possible you need to use a lead sled (still the old Mark 1 eyeball variations to deal with)

If you get say a 1 inch 10 shot group at 100 yds with lead sled and it opens up beyond that when you shoot off a rest, then its you.

Per Bart B, when I am really trying to see what a load and I do I go with a 10 shot group.

I do shoot 5 shot groups looking for those nice sub .5 MOA and occasionally get it but I know I don't shoot that well nor the hand loads are that good.

PS: I also do not get excited about accuracy. I pick a point of aim on the target and assess the group. If I get a group I want closer to the aim point I adjust it over but seldom worry about spot on, if its a good load and rifle you can always do that but I am testing a lot and not worth it.

in other words I shoot for consistency y as I am not in completion.

I believe some competitions awards to both, best accuracy and best group.

If one guy has the best of both then he get to choose which one he wants and the next guy with the other best score gets that.
 
A tenth shot lying some 2-3" away is most likely the shooter, but can be anything including one tiny defect in the bullet. A tenth shot lying some 2-3" away only matters of it's consistent.
There is no calculating MOA for group size. There's just measuring the group size. An MOA is an absolute. One MOA being considered 1" at 100 yards.
 
Are you interested in determining how accurate YOU are as a shooter, or you interested in determining the RIFLES potential with that load?

If there is a shot that is a called flyer, one that you knew as soon as you pulled the trigger was a shooter induced error even before you see the target, I wouldn't consider the rifle or load at fault.

I don't have any faith in a single group. It doesn't matter if it is a 3, 5, or 10 shot group you can either get very lucky, or very unlucky.

If I'm interested in determining the rifles potential I like to shoot a series of 3 shot groups over several range trips. At least 10 groups of 3 and try to determine an average. More than 3 shots at a time and you are measuring the concentration levels of the shooter more than the rifles potential.

If I have a fluke that I know is my error I wouldn't include that group in the average if determining the rifle or load potential. If I'm interested in determining my potential to shoot accurately then all shots and all groups would be counted.
 
There is no calculating MOA for group size. There's just measuring the group size. An MOA is an absolute. One MOA being considered 1" at 100 yards.
Click to expand...

Yes, you can use MOA instead of group size. A .65" group at 100 yards would be considered .65 MOA, a 1 1/2" group at 100 would be considered 1.5 MOA, a 1" group at 300 yards would be considered .33 MOA.
 
>At which point do you feel comfortable ignore outliers? <

Almost never - it's how you're actually shooting.

The one flyer I can remember that I ignored was when I was shooting a handgun and got bit by a horsefly at exactly the wrong moment.
 
With most AR15s, the first shot (manually loaded), and more often the last shot from a magazine are slight to significant fliers.

When I shoot AR groups, I have 2 targets. The first shot from a magazine gets shot on one target and the group on a 2nd target. If I shoot the last round from a magazine, I do not shoot it into a group either.

If I load all rounds off a sled, I can get the smallest group that AR will produce in most cases.
 
Are you interested in determining how accurate YOU are as a shooter, or you interested in determining the RIFLES potential with that load?
Click to expand...

Well, as a relatively new shooter with about 1000rds on an AR (a healthy chunk of which was just trying to set-up my scope and the like) we can safely say that as a shooter, I am a constantly changing parameter. Hopefully for the better.

So really I am trying to evaluate a set of loads for consistency in my rifle. I am well aware that my set-up is not ideally. But at least it is consistently "not ideal"!
:D

With most AR15s, the first shot (manually loaded), and more often the last shot from a magazine are slight to significant fliers.
Click to expand...

Using the round-robin method, every shot was both the first and last shot in the mag! So again, not ideal, but consistently applied.

The one flyer I can remember that I ignored was when I was shooting a handgun and got bit by a horsefly at exactly the wrong moment.
Click to expand...

I'll have to remember that one for any particularly embarrassing shots... ;)
 
Thats a good question that I will attempt to answer from my own experience.

First never count the flyer, its shooter error assuming you are hand loading and your ammo is consistent. Even with new factory ammo its probably still shooter error. Gun show reloads in my experience are never consistent, I have never shot Black Hills reloads they seem like they might be good.

If you shoot frequently then you know as soon as you pull the trigger if the shot was good from a shooter aspect. You can just tell by how the rifle felt when it recoiled.

If your flyers are in the same general area and your good shots are grouped separately then you can work on whats causing the flyers from a shooter perspective. It means when you screw up you do it consistantly the same way.

If your flyers or group is all over the place. Different point of impact every shot then you have a rifle problem. Sights, stock or something else.

If you have a decent group but just not 1 MOA or less at 100 yards chronograph the loads and see what the MAD, ES and SD percentages are. If not under .5% then load development needs to be looked at.

Also make sure your rifle is set up for the style of shooting you are doing. Scope hight, scope position, and cheek rest position is different whether you are shooting prone, off hand, off a rest or on a bench. I struggled with this for over 15 years until I had a very experienced long range shooter watch me shoot and schooled me on the importance of rifle setup and how to do it. I believe its probably the most important contributor to accuracy second only to a properly assembled rifle.

I hope this helps a little. YouTube some videos on properly setting up rifles.
 
Nice pointers, Madgunner.

Thanks.

Set-up. Given my access to kit and more experienced shooters, a well-schooled long range shooter like the one you described may have a lot to comment on! :eek:

But my groups have been tightening with this gun, on factory ammo, so I must be doing something right. The question is, have a reached a self-imposed ceiling in my improvement because of aspects of my technique I am unaware of.

Hard to say.

Looking forward to seeing how the groups work out, later today!
Fliers or no fliers.
 
I thing it depends on whether the flyer was shooter induced. I shoot an air rifle capable to shooting 60 shots into one ragged hole. From a rest with match grade pellets. I am not a shooter capable of shooting 60 shots into one ragged hole (yet), but I am able to call my shots so I know what I was doing when the air rifle discharged.

For testing centerfire ammunition handloaded by myself, I try to use a bench and a rest to eliminate as much of the shooter error as possible. But that is not always the case, so I try to shoot 5 or 10 shot groups to get a good feel for how the load will perform with me at the controls.

Odds are if there are on or two obvious flyers in a 10 shot group into the 8 or 7 ring it's an operator error on my part, especially if I called it.

And this gets down to a shift I made in my handloading for accuracy for my semi auto rifles. I no longer load to get the absolute most accuracy out of my rifles, the quest for one ragged hole can be a fools errand (or at least a very expensive task). I load to make ammunition that will stay in the 10 ring if I do my part as the shooter. Even government issued match grade Mk262 ammunition fails to stay sub MOA in a 50 shot group, generally between 1.3 and 1.6 MOA from an accuracy device, but that's well within the 2 MOA standard for the 10 ring.

So to sum up all of that, if it's called flyer because you know you screwed up the shot, I say don't count it in group size. If it isn't a called flyer, count it. But overall group size is only a measure of potential precision, not accuracy.

Jimro
 
You must log in or register to reply here.
Back
Top