SD

[BenPerfected]

I have been testing numerous 6BR load combinations. The plan is to shoot in a few 300--600 yd. events in 2015. Taking out the input/skill of the shooter, the quality of the gun and the weather, how much of accuracy is based around SD? How many shots do you take of a given load to establish the SD? How important is an additional 50-75 FPS vs SD?
Lot's to learn,
Ben

 

[ShootDots]

Personally, I do not pay attention to the "S.D." numbers. If I can consistently keep my E.S.'s 12f.p.s. and under, that is the info I need. Accuracy trumps speed all day long, however, if the accuracy level is about the same and the E.S.'s stay low, the faster the better..

 

[Lapua40X]

If I could keep my E.S. at 12FPS I wouldn't worry about S.D.. But seeing as how there always seems to be one in a group of ten that has a mind of its own and my E.S. runs a little higher than that during some test sessions, I look for a S.D. of <12 in my 6BR Norma.

 

[jcampbellsmith]

I ignore ES. I pay some attention to SD. I try and practice and compete as much as I can. I think it's excellent that you are coming up with a plan for the coming season. My one recommendation on planning is to write your plan down and review and update it as you progress.

Regards JCS

 

[Scott Harris]

Having a basic understanding of statistics can be VERY helpful when interpreting chrono numbers. One important thing to keep in mind is, "the larger the sample size, the more useful the info." Likewise, the smaller the sample size, the less useful the info. A sample size of 20 is considered a bare minimum to provide any kind of useful data in many statistical studies. So, reliance on 3 & 5-shot groups is a poor practice IMHO.

ES's and SD's on 3 & 5-shot groups are only a starting point for initial load development. SDs calculated over multiple 20+ shot strings should be used to confirm what you "think" is a good load. I look for loads that will provide an SD of 7 or less across multiple 20+ shot strings measured on different days with ammo prepped at different times. This practice with give you a much better idea of how your load will perform over time.

The magnetospeed provides a great way to collect larger samples of data: simple attach it to your rifle for an entire match and use each string as a sample. Cherry-picking the group size or SD/ES of a single 3 or 5-shot string causes a lot poor load decision-making.

A rough rule of thumb, multiply your SD by 4 to get an estimate for your ES over an entire match. SD is incredibly important for 1K shooting as it greatly influences vertical dispersion. SD becomes progressively less important as the distance to the target shrinks.

ES does not take into consideration the size of your sample. SD does and is therefore the better number to use when estimating the extreme spread you can expect over the course of many matches. As mentioned above, the calculated SD becomes more reliable with larger sample sizes....20+ shots or more preferred.

 

[Alex Wheeler]

Es is what shows on the target. If your es is high, you need to fix your load/loading practices. I ignore sd.

 

[ShootDots]

Es is what shows on the target. If your es is high, you need to fix your load/loading practices. I ignore sd.

++1.. With very low E.S.'s you will see a very small "vertical dispersion" at the target at 600 and beyond. If your E.S.'s are say 20+ you will have difficulty staying "on the waterline" in F-Class competition. Once I THINK I have a good load, I take it out to 1000 yards to see just how "good" the load is.

 

[fredhorace77]

If our chronograph simulates with the POI's on the target above +85% of all the shots, then our chronograph setup is respectable and we can trust the data. If not the Average, ES, and SD values are in to big of error to look upon as factual.
With that said, and we have a good chronograph setup that simulates and patterns the target, then the ES is telling us with strong probability, the actual "vertical dispersion". For me ES from the chronograph and the vertical dispersion on the target answers most all my needs, and the Average tells me my loads speed.
Myself, I pay little attention to SD
Donovan

This very good advice.

 

[Catfur]

Es is what shows on the target. If your es is high, you need to fix your load/loading practices. I ignore sd.

ES is what was shown on the target you shot. It is not in any way predictive of what will be shown on the next target.

SD is predictive, ES is not.

 

[JRS]

Bingo, Catfur

 

[CharlieNC]

Not everyone wants to invest the time to learn basic statistics and the meaning of SD, and that is ok. When you have a set of numbers, ES is calculated from only two of them: the maximum - minimum. On the other hand, SD is calculated using all of the numbers. Thus the SD is a more "robust" parameter and it is therefore more useful and powerful, which is of course the purpose of taking these measurements. For example, how reliable is that average velocity you just obtained? Do you think if you repeat the process that you will get exactly the same average? Well the SD can be used to establish the confidence interval for that average, then you can decide if that range is "tight enough" for your application. This question arises at an interesting time, as the Daily Bulletin has an article comparing the accuracy of the 308 to the 30-06; I doubt the conclusions comparing 99.5% to 99.2% average accuracy are valid if the SD of that data were used for a statistical comparison, meaning there is probably no difference in performance. So what? As shooters we are always seeking to achieve more consistency, higher accuracy, etc and therefore we are always comparing one case to another to decide which is better; and it is easy to draw an erroneous conclusion, especially when the differences are slight, if statistics such as SD are not used to assist in the decision process.

 

[BenPerfected]

Not everyone wants to invest the time to learn basic statistics and the meaning of SD, and that is ok. When you have a set of numbers, ES is calculated from only two of them: the maximum - minimum. On the other hand, SD is calculated using all of the numbers. Thus the SD is a more "robust" parameter and it is therefore more useful and powerful, which is of course the purpose of taking these measurements. For example, how reliable is that average velocity you just obtained? Do you think if you repeat the process that you will get exactly the same average? Well the SD can be used to establish the confidence interval for that average, then you can decide if that range is "tight enough" for your application. This question arises at an interesting time, as the Daily Bulletin has an article comparing the accuracy of the 308 to the 30-06; I doubt the conclusions comparing 99.5% to 99.2% average accuracy are valid if the SD of that data were used for a statistical comparison, meaning there is probably no difference in performance. So what? As shooters we are always seeking to achieve more consistency, higher accuracy, etc and therefore we are always comparing one case to another to decide which is better; and it is easy to draw an erroneous conclusion, especially when the differences are slight, if statistics such as SD are not used to assist in the decision process.

Dead on for me! This helps me better understand SD and how to use statistics in shooting. I dropped statistics and added a marketing class as I better understand BS. Going forward, when a load starts to look promising, a min of 20 shots will be used to make statistical comparisons.
Thanks for the continuing feedback.
Ben

 

[CharlieNC]

Donovan cases such as this where there might be excessive variability at play, whether known or unknown, is exactly where the application of statistics is valuable. For example if you want to decide whether load A is less variable than load B, the the simple "F-test" is used where F = {SD(max)/SD(min)}^2. For 20 shots each, if the F-ratio is greater than 2.0, then there is a 95% probability that the SD are not the same and you can conclude that one loading process gives more consistent results than the other. If chrono variability were suspect, then it would be prudent to alternately shoot load A vs load B such that the effect of the chrono is spread comparably across the two. Of course if the chrono effect is high vs the effect of loading, then it would take many more samples to distinguish between the two. Statistics is all about distinguishing the signal from the noise, and if done properly will greatly reduce the number of samples (meaning time, cost, etc) to reach valid conclusions.

 

[BenPerfected]

Ben -

That's all good and fine, as long as your chronograph data is accurate (as I alluded to above).
Else wise 20-shots for SD is a total waste of barrel life, components, and time.

Good Luck
Donovan

The Magneto Speed Chronograph is being used so hopefully the data will be reliable. Thanks for the Good Luck, I know I will be needing a lot of that at the range.

 

[Scott Harris]

Ben -

That's all good and fine, as long as your chronograph data is accurate (as I alluded to above).
Else wise 20-shots for SD is a total waste of barrel life, components, and time.

Good Luck
Donovan

Agreed. Bad chronos are worse than no chronos.

 

[dixieppc]

Not everyone wants to invest the time to learn basic statistics and the meaning of SD, and that is ok. When you have a set of numbers, ES is calculated from only two of them: the maximum - minimum. On the other hand, SD is calculated using all of the numbers. Thus the SD is a more "robust" parameter and it is therefore more useful and powerful, which is of course the purpose of taking these measurements. For example, how reliable is that average velocity you just obtained? Do you think if you repeat the process that you will get exactly the same average? Well the SD can be used to establish the confidence interval for that average, then you can decide if that range is "tight enough" for your application. This question arises at an interesting time, as the Daily Bulletin has an article comparing the accuracy of the 308 to the 30-06; I doubt the conclusions comparing 99.5% to 99.2% average accuracy are valid if the SD of that data were used for a statistical comparison, meaning there is probably no difference in performance. So what? As shooters we are always seeking to achieve more consistency, higher accuracy, etc and therefore we are always comparing one case to another to decide which is better; and it is easy to draw an erroneous conclusion, especially when the differences are slight, if statistics such as SD are not used to assist in the decision process.

Charles,
Thanks for the statistics lesson. I'll be the first to admit that statistics is not my best suit. My lowest grade in college was a required core class in statistics. Of course it might have had something to do with alcohol and girls. Anyway I have been using standard deviation as my numerical zero range test for a load before moving out in distance and looking at group size. All of my initial load tests are run in my back yard through a chronograph at 10 yards. I have found that I can have a high extreme spread due to a single shot going off the reservation but still have an acceptable standard deviation on a 10-20 shot string. Of course, removing that single shot's velocity from the calculations will bring the extreme spread down significantly but not change the standard deviation all that much. Therefore, I put more stock in standard deviation than I do extreme spread.

Am I thinking wrong on this?

As a side note and off the subject, I have been cutting down on those single shots going off the reservation by paying more attention to bullet seating tension with the addition of a K&M force measurement option added to my arbor press.

 

[CharlieNC]

Your thinking is correct Dixie, and it is always prudent to look at all the data rather than just going off of the statistics themselves. You have experienced an "outlier" which is not within the normal distribution of the other data; its just like calling a flier when you know you pulled it.

 

[Alex Wheeler]

You guys shoot for statistics, I'll shoot for small groups. ES is whats needed for small groups. You can get good sd with a high or low shot, but you wont get good groups that way.

 

[noload]

CharlieNC, Thanks for refreshing my memory. I think I will pay more attention to SD.

 

[Catfur]

ES is whats needed for small groups.

Magical thinking. ES IS NOT PREDICTIVE. SD is. ES tells you something about the group you shot, past tense, nothing about what the next group will look like. SD contains the information you want. If you compare the outliers (if there are any) to the SD, then that can, at best, tell you you have a problem (I.E. that your distribution is not normal, which it should be).