SD: overrated...?

[IdahoSharpshooter]

I have read and reread all five of Brian's books, and the Berger manual.

I understand the concept of SD, but I am skeptical of its importance as it relates to accuracy and the bell curve to real world use.

The bell curve logic is that the more groups that are fired, the more the average group size will grow significantly.

I do not see how firing more groups, once you have found an accurate load, will cause accuracy to deteriorate.

I also do not see why, the more groups that are fired, why SD will double or triple.

thank you for explaining this dichotomy to me.

Rich

 

[JohnKielly]

Perhaps you should be reading a book on statistics & understanding the implications of 1,2 & three standard deviations.

The implication is that a low level of statistical precision (& a 10 shot group fits that category) you may well maintain all your shots within the first standard deviation, but eventually, and it might be sooner than later, you'll get that shot that's a third standard deviation, and if you're like most of us, you call it a flyer or offer some other excuse.

 

[CharlieNC]

For the people in your city, what is the average height and the range of heights? How good is the answer if you measure 2 people, vs 5, vs 50, vs 1000, vs all? Of course measuring all gives the right answer, but how many is good enough? Practically we know shooting a group size of 2 is typically smaller than 5, which is smaller than 20. For shooting throw in the aspects that accuracy and precision are different, and the answer becomes much more extensive to assess; think about scoring a Fclass target for which both elements are required.

 

[BradY]

One quote that I have seen and really sticks in my mind is:

ES tells you what has just happened, SD is an indication on what to expect.

 

[Rtheurer]

Yes South Dakota is Over Rated......

 

[Larryh128]

Yes South Dakota is Over Rated......

Made me laugh.

 

[IdahoSharpshooter]

Perhaps I am missing the point, or need to clarify.

Anyone here ever hear of the Houston Warehouse? Indoors, so basically a zero condition range. If I have, for instance, a 6PPC that I can count on to shoot ten-shot groups under 1/4 MOA at the short range distances. 100/200/300yds, in the warehouse. Why would my accuracy level change for the worse? What would cause my ES, and therefor SD to increase?

thank you again,

Rich

 

[IdahoSharpshooter]

CharlieNC,

apples and oranges. When we discuss ES and SD we are talking one rifle with one specific load that consistently produces the same basic MV, and ES and SD.
Example: a 6 Dasher that shoots the 108 Berger 3000fps, and sub-1/4 moa at 100yds. If the initial 10-shot group chrono's 3000fps, and the ES is (say) 15fps. ES works out to be 8.
What will cause the MV ES/SD to double, or even triple? Only a velocity spread increase can cause the ES/SD to change enough to change the validity of the results.

 

[Toby Bradshaw]

Perhaps I am missing the point, or need to clarify.

Anyone here ever hear of the Houston Warehouse? Indoors, so basically a zero condition range. If I have, for instance, a 6PPC that I can count on to shoot ten-shot groups under 1/4 MOA at the short range distances. 100/200/300yds, in the warehouse. Why would my accuracy level change for the worse? What would cause my ES, and therefor SD to increase?

thank you again,

Rich

Increasing ES does not necessarily increase SD. As sample size increases, SD will converge on a single value. Increasing sample size will always increase ES if ES follows a typical distribution (Gaussian, Poisson, etc.).

 

[SheepDog]

I have read about the Houston Warehouse... and the trouble some of the shooters had under as ideal conditions as they could get.
There is no such thing as zero condition situation. Even if you take all the wind and temperature variables out you still have barometric pressure and humidity as variables that are outside of your gun. You still have all the variables in your gun to complicate the issue. Just because the first 10 rounds you fire go to a 1/4 inch group does not mean that the 11th round will have identical interior and exterior conditions. If you have taken the time to carefully measure, weigh and equalize each part of your load, you know that bullets vary in bearing length, ogive shape and length. You know that there are deviations in powder kernel size, shape and probably in coating effectiveness. There are variations in primers too. Normally you get a decent average of variations in each load but these variations can "stack up" in rounds and you have no control over that. On top of that we all know that chambers and barrels change each time you shoot.
I am not a long range competitor but I test my loads over a period of months just to make sure they are consistent enough for my hunting purposes. Once in a while I find a load that seems to be very good with 20 rounds fired into 4 targets with five shots in each and all 20 fired into a backing target as an aggregate. Then a couple of months later it goes south on me under different conditions. I have to reset my process and find out what causes the group to change. That is what makes loading for consistent accuracy such a challenge. It might be that the variations in my components has stacked up in those rounds and the next time I shoot it will go back to a nice small group. It might also be that I am on the edge of a harmonics node and under the wrong conditions it is slower or faster enough to change the point of impact with the lowest common denominator affecting the group size.
The more rounds you use to get a standard deviation and extreme velocity spread the more likely you are to understand the range of variations your load can have.

 

[BoydAllen]

I think that you are confused. While accuracy at long range has ES (on a particular target) as one, albeit important component, by itself, it is no guarantee of accuracy. You can be out of tune and have excellent uniformity of velocity. Furthermore, when you bring up the warehouse, you may be forgetting that ES and SD are not nearly as important in the short range benchrest game. Nobody worries about either. They tune by looking at groups pretty much exclusively. The last thing that I should mention is that the real reason that we are bothered about SD is because chronograph manufacturers included it in their list of thing that can be read from their instruments, probably in anticipation of their use by ammunition manufacturers, and the larger samples that would be taken by those users. What I have learned from all of these on line discussions is that there are vast numbers of shooters who, even if they had a statistics class, through no fault of their own, from complete disuse, have forgotten that brief bit of knowledge that allowed them to pass the class. The very idea of discussing the SD of a five shot sample is ludicrous, yet it is done all the time. Your targets only "see" ES, and they do not see that at warehouse distance. I have spoken with someone who spent significant time in the warehouse, and he told me that he never saw the level of accuracy that was reported as common.

 

[Jay Christopherson]

I think that you are confused. While accuracy at long range has ES (on a particular target) as one, albeit important component, by itself, it is no guarantee of accuracy. You can be out of tune and have excellent uniformity of velocity. Furthermore, when you bring up the warehouse, you may be forgetting that ES and SD are not nearly as important in the short range benchrest game. Nobody worries about either. They tune by looking at groups pretty much exclusively. The last thing that I should mention is that the real reason that we are bothered about SD is because chronograph manufacturers included it in their list of thing that can be read from their instruments, probably in anticipation of their use by ammunition manufacturers, and the larger samples that would be taken by those users. What I have learned from all of these on line discussions is that there are vast numbers of shooters who, even if they had a statistics class, through no fault of their own, from complete disuse, have forgotten that brief bit of knowledge that allowed them to pass the class. The very idea of discussing the SD of a five shot sample is ludicrous, yet it is done all the time. Your targets only "see" ES, and they do not see that at warehouse distance. I have spoken with someone who spent significant time in the warehouse, and he told me that he never saw the level of accuracy that was reported as common.

This, however, is why I've so enjoyed my Labradar and frankly, why my load development methods and expectations have been entirely reset. With the Labradar (and lots of accommodating match directors) I can chronograph nearly every single round that goes through my barrel and now SD becomes a significant measure. On my #1 barrel, I have 800+ rounds of recorded data through it, including full match strings. What I will say primarily, is that there are a lot of people talking about single digit that and single digit this that would be entirely shocked if they were to chrono a full 60-shot match. I've also nearly discarded ES as a significant statistic - I look at it because it's there, but I don't load to it anymore. I've had plenty of match strings with what would previously have been entirely unacceptable ES, show a nice clean target. I've also had very low ES strings show ridiculous vertical. SD tells me how consistent my loading practices are over a long period of time and that's what I want to know.

This is all geared towards slow fire shooting that involves several matches, though. I know plenty of top string-fire shooters that won't even look at a chronograph and couldn't care less about ES or SD. The group is king (and rightfully so in that case).

 

[dskogman]

It's simple probability and statistics. The accuracy of the gun doesn't change but our measurement of the accuracy does. SD and ES need to combined with sample size and margin of error to be useful. If you have a small sample (say 3 shots) the margin of error is high and your SD and ES could be off 100% or maybe only 10%. The more shots you measure the lower the margin of error and the more accurate the SD/ES will get. Based on Brian's experience this usually will tend to make them larger.

 

[CharlieNC]

Once again I would refer you to a post I did in the Ballistics Forum several months ago specifically regarding chrono statistics. This included a discussion regarding sample size (number of shots), and an excel spreadsheet to carry out the appropriate analyses on chrono data for a simple case. To say a rifle and load possess "constant" accuracy and chrono performance is a bold statement to make in a statistical sense. It is straightforward to evaluate chrono data, but there is no agreed definition to characterize target performance which incorporates both accuracy and precision.

 

[jwink]

SD exists because there is no way that a firearm held by or controlled by a human can be expected to perform exactly time and time again with no margin of error even under the most exacting of conditions. The more something is done, the more likely that conditions will change, even inside a warehouse. Thus groups grow as shots fired grows.

 

[CharlieNC]

Damn, when you put it that way, maybe not "everyone" in Butte is short, red headed and have bad tattoos lol.

Tom

Funny my tattoo has grown larger over the years.

 

[rardoin]

This, however, is why I've so enjoyed my Labradar and frankly, why my load development methods and expectations have been entirely reset. With the Labradar (and lots of accommodating match directors) I can chronograph nearly every single round that goes through my barrel and now SD becomes a significant measure. ......... What I will say primarily, is that there are a lot of people talking about single digit that and single digit this that would be entirely shocked if they were to chrono a full 60-shot match. ....... SD tells me how consistent my loading practices are over a long period of time and that's what I want to know.

+++++
It really opened my eyes also on a 3 string F-open match...recorded all 60 record shots plus sighters....

 

[IdahoSharpshooter]

CharlieNC,

I believe I said "consistent" accuracy and precision. Your last sentence makes sense.
That said, I still hold that only the number of shots for record should be considered, in terms of the significance of ES and SD data. Accuracy trumps everything.
What I read in terms of group enlargement suggests to me, that only conditions, at extended range, really matter.

I contend that, if you build a rifle that will consistently group 1/2 moa, at 1000 yards; only external conditions, wind, temperature, barometric pressure, and mirage can cause a significant increase in the placement and size of bullet holes on said target. If I shoot a clean target, 100-10X at 1000 yards, only a change in those external influences can degrade the proven accuracy to a significant degree. IE, I should be able to duplicate the groups within one SD, and maintain a 6/10ths moa level of accuracy, even if I cannot always replicate the 1/2 moa level.

I guess I am asking, what can cause a load to increase from 1 SD to two or three SD?

Bear with me, I normally just shoot live varmints at extended ranges here in SW Idaho instead of paper.

 

[tmwinds]

SD exists because there is no way that a firearm held by or controlled by a human can be expected to perform exactly time and time again with no margin of error even under the most exacting of conditions. The more something is done, the more likely that conditions will change, even inside a warehouse. Thus groups grow as shots fired grows.

Your group will never get better after the first shot. It can only get worse.

 

[CharlieNC]

Idaho if you want to use group size to define accuracy (it is actually a measure of precision, accuracy meaning how far from the bulls eye) then go shoot groups 20 times (across a period of time). The average and standard deviation of the 20 group sizes can be used to infer what you would expect to shoot in the future if nothing changes; the result is the normal bell curve distribution, not a constant. This approach should also be used to determine if you have made an outstanding improvement, or if something has gone to crap, vs just getting the occasional groups that are just at the tails of the bell curve.