What contributes to ES and SD beside the obvious?

[.505gibbs]

I was out shooting this weekend trying to find a load for two separate rifles both in .35 Whelen. I had several lots of ammo I had reloaded. Long story short both rifles shot very well with one particular load so I attached the magneto speed to get some velocity numbers. One of the rifles has a practically new Pacnor barrel with a 14 twist and is 23 inches long. The other is a factory Ruger with either a 14 or 16 inch twist and is 22 inches. Probably has about 1100 rounds on it.
The Pacnor gave me a MV of 2936 and an ES of 6. The Ruger was 2894 with an ES of 48. The sample size was rather small, consisting of ten rounds for each rifle. They were fired in two stages of five rounds, switching the chronograph between rifles and allowing them both to cool back to ambient temperature.
For a hunting rifle that likely will never be shot more than 400 yards at game 48 fps for an ES is fine with me. I’m just curious as to what would cause such a large difference between the two?

 

[Dogtown]

I’ll bite: if the one variable between the two is the barrel, then there’s something wonky in that Ruger barrel. Barrels can be really weird that way.

 

[.505gibbs]

I’ll bite: if the one variable between the two is the barrel, then there’s something wonky in that Ruger barrel. Barrels can be really weird that way.

Thanks for taking a stab at it. I know the two are different in the ways the Ruger has a looser chamber and more freebore. Maybe that is what is causing the difference.

 

[DocRDS]

ES is a (somewhat useful) summary number. But more important is to look at your 10 shots. Were they evenly spaced across the interval? Were some of them grouped and you had a 'flier' or outlier (or two)?

People often get caught up in SD and ES but, all your data tells a story.

And you could have just gotten unlucky. WHile that is unlikely, it is not impossible.

I'll say this: A someone who practices statistics for a living, Extreme Spread is only used is shooting. We don't use it in engineering, Medical Trials, or Pure Statistics. Look at ALL your data.

 

[mrcheesemoose]

ES is a (somewhat useful) summary number. But more important is to look at your 10 shots. Were they evenly spaced across the interval? Were some of them grouped and you had a 'flier' or outlier (or two)?

People often get caught up in SD and ES but, all your data tells a story.

And you could have just gotten unlucky. WHile that is unlikely, it is not impossible.

I'll say this: A someone who practices statistics for a living, Extreme Spread is only used is shooting. We don't use it in engineering, Medical Trials, or Pure Statistics. Look at ALL your data.

I’ve seen it used once. That guy was fired the next day

 

[.505gibbs]

Thank you for such an insightful response DocRDS. I should have listed the SD numbers in my initial post. The Ruger was like 17.6 and 19.5 for each of the two 5 round strings with nothing I would call an extreme outlier for any of the shots. I don’t recall the SD for the Pacnor. Don’t have my notes with me currently.
I’m going to try again this weekend. Ran out of time and ammo. Mostly just for fun. For deer hunting I’m happy with both rifles and the load.

 

[ITR_98_111]

I'd give myself a pretty solid 8 out of 10 on reloading consistency, I get better results that the best factory ammo I've shot on average, yet on the wrong day, it varies. Allot goes into the barrel, how far off the lands the bullets are, the bearing material (rifling), etc. I'd shoot them again to verify - but its the reason you have to work up loads per barrel.

Here's some of my ES (5 shot groups); same load, same gun, same day. The maximum ES comparing all 50 shots was 58 with an SD of 16.2.

Same gun, same load - different day.

 

[Dogtown]

"The power of noise"

 

[Doom]

I'd give myself a pretty solid 8 out of 10 on reloading consistency, I get better results that the best factory ammo I've shot on average, yet on the wrong day, it varies. Allot goes into the barrel, how far off the lands the bullets are, the bearing material (rifling), etc. I'd shoot them again to verify - but its the reason you have to work up loads per barrel.

Here's some of my ES (5 shot groups); same load, same gun, same day. The maximum ES comparing all 50 shots was 58 with an SD of 16.2.
View attachment 8250362

Same gun, same load - different day.
View attachment 8250364

You should be looking at the data differently. You have 50 pieces of data for each day. When you look at ES on a 5 shot basis you are ignoring 60% of your data and looking at 40% of data selected on a biased worst case basis. You are taking the high and low value from each 5 shot group. You actually have enough data make statistically significant observations on your standard deviation which we know how to analyze.

It makes much more sense to calculate the Standard Deviation on all 50 shots and IF you are interested in the extreme spread look at it based on the extreme two values of the 50 shot group. As @DocRDS, extreme spread is not a value that has any statistical significance. It has no definition and has no means or tools to analyze it. It is at most an anecdotal (Based on casual observations or indications rather than rigorous or scientific analysis) value that is used to try to quantify a worst case scenario. In statistical reality the extreme spread of a large normally distributed population will be about 6x the standard deviation (in the 0.3% of the population). In the real world it can occur because of an error in process and exceed that 6x value. In that case those values are called outliers and do not represent the sample set. In the first data set your ES is 3.6x the SD which is not bad. If you had shot more the ES would likely have increased but the SD would likely have stayed in the 13.5 and 20.2 range which is the 95% condfidence interval. I'm not familiar with the cartridge so I don't know if a 16 SD is considered good or bad.

As for same load different day, you are likely shooting at a different temperature (affects velocity and pressure), time of day(affects optical chronographs), sampling errors, chronograph repeatability (including distance to chronograph), maybe others. Unfortunately as I noted above statistics does not provide any tools to analyze the difference and the results become anecdotal. If we were to analyze the standard deviations of the two 50 shot groups we can make some determinations if the different results are indeed significant.

As for the original question, itI would help to know cartridge load specifics (including distance to lands) and how powder is measured. It is possible with only 10 rounds per rifle your ES is due to simple sampling probabilities. In actuality the chambers are different, the barrel bore and lands are different so it is likely that the results are going to be different.

 

[.505gibbs]

Load data for the initial question consisted of Norma 30-06 brass necked up in two steps. Once through a .338-06 die and then the .35 Whelen die. Brass was new and not fireformed. Powder was 60 grains of AR-Comp dispensed through a chargemaster supreme, bullet was a 200 grain Barnes TTSX seated 2.76 cbto. That was .070 off the lands in the Pacnor barrel and .118 in the Ruger. Primers were CCI #34’s

 

[BuildingConceptsllc]

Really should just extrapolate your ES from the SD once you have enough shots to matter. ES of a sample doesn't tell you anything really , as Doom has explained. SD from a decently large sample gives you some pretty reliable parameters

 

[Rio Precision Gunworks]

ES is a (somewhat useful) summary number. But more important is to look at your 10 shots. Were they evenly spaced across the interval? Were some of them grouped and you had a 'flier' or outlier (or two)?

People often get caught up in SD and ES but, all your data tells a story.

And you could have just gotten unlucky. WHile that is unlikely, it is not impossible.

I'll say this: A someone who practices statistics for a living, Extreme Spread is only used is shooting. We don't use it in engineering, Medical Trials, or Pure Statistics. Look at ALL your data.

Yea, using ES is just something people in this industry decided to make up because they either A) don't understand how sd/es actually work or B) think they can shortcut sample size somehow with ES (which means they don't understand how it actually works).

The latest was the most recent incarnation of Lynn who thinks that because you need a large sample size (let's not even talk about using confidence intervals) to narrow down your SD......that you can use ES to somehow shortcut that via the logic of "if I keep my ES small, then everything inside of that is going to be g2g" which is a gross misunderstanding of how sd/es work.


I'm not sure who first perpetuated the whole ES importance in shooting.......likely the same person who believes you can add powder to a case and get the same velocity as a lower charge weight and have a "flat spot."

 

[GTOJOSH]

They have medications all over the radio if you're over 40 to solve es and sd issues.