Show your Tikka ten round groups

Joined
Oct 8, 2019
Messages
2,956
The following works well for me but may not work well for you.

I have never done a 10-shot group in a hunting rifle. If do I shoot for groups I typically limit it to 3 shots as that is more aligned with hunting. I may do multiple 3-shot groups but there'll be ample cool-down time between them. As long as the size is relatively consistent and the shots go where I want them to, things are good. Even then, the most important shot on a hunting rifle is the very first shot.

I cannot control the idiots peppering me with their hot brass while I am shooting. I cannot force those idiots to put up a divider. Having hot brass hitting me and/or going down my shirt does impact my shooting. I cannot control idiots at the range that shoot my target intentionally or not. Nothing like looking at your target and seeing bullet holes appear everywhere. Even with me behind the trigger I am going to make some mistake with each additional shot taken. That is human error and not an issue with the rifle/scope/ammo. I don't use a sled or anything like that; just a simple bipod up front and a rear rest of some sort. Just like I do when hunting. I use the same setup when first setting up new rifle.

I have played around with my T1X (17 HMR shooting 20 gr Hornady). Normally I just go nice and slow working the fundamentals. Other times I shoot at extended ranges for practice dialing for distance and wind.

Here are 15 rounds fired as fast as I could at 100 yards with no adjustments for wind. You can figure out which direction the wind was blowing and which shots were due to the barrel getting hot.
Fast_15.JPG

Here is another batch (at least 23) at 100 yards just taking it nice and slow with zero adjustments for wind. This was right before heading out of town to hunt deer. Even though I was taking it slow, I did still screw the pooch on some of the shots.
Slow.JPG
 
OP
H
Joined
Feb 2, 2020
Messages
1,992
Is there some math involved around the significance of a 10 shot group? It seems like the burden of sample size keeps growing. Used to be 3, then 5, now 10. I seem to remember some software out there that calculates hit probability and I was wondering if a 10 shot group amounts to some sort of significance level or confidence interval.

Where is the point of diminishing returns in the number of rounds piled into a group?

Is there a linear relationship that can be assigned to group size (MOA) and number of shots in the group? If I shoot a 1/4 MOA 3 shot group, can I extrapolate a 1/2 MOA 5 shot (to a lesser confidence) and a 1 MOA 10 shot (to an even lesser level of confidence)?

These Tikka threads have become quite data driven, which I appreciate, and I want to know if the work has been done to vet our actual experimental design. It seems that the interest is there and guys want to do this right, so let’s do it right. Or at least understand what we’re doing beyond just “the internet wants 10 shot groups now”
No, there is no magic with a 10-shot group in terms of statistics. More is better than less, but no magic to 10.
The following works well for me but may not work well for you.

I have never done a 10-shot group in a hunting rifle. If do I shoot for groups I typically limit it to 3 shots as that is more aligned with hunting. I may do multiple 3-shot groups but there'll be ample cool-down time between them. As long as the size is relatively consistent and the shots go where I want them to, things are good. Even then, the most important shot on a hunting rifle is the very first shot.

I cannot control the idiots peppering me with their hot brass while I am shooting. I cannot force those idiots to put up a divider. Having hot brass hitting me and/or going down my shirt does impact my shooting. I cannot control idiots at the range that shoot my target intentionally or not. Nothing like looking at your target and seeing bullet holes appear everywhere. Even with me behind the trigger I am going to make some mistake with each additional shot taken. That is human error and not an issue with the rifle/scope/ammo. I don't use a sled or anything like that; just a simple bipod up front and a rear rest of some sort. Just like I do when hunting. I use the same setup when first setting up new rifle.

I have played around with my T1X (17 HMR shooting 20 gr Hornady). Normally I just go nice and slow working the fundamentals. Other times I shoot at extended ranges for practice dialing for distance and wind.

Here are 15 rounds fired as fast as I could at 100 yards with no adjustments for wind. You can figure out which direction the wind was blowing and which shots were due to the barrel getting hot.
View attachment 184911

Here is another batch (at least 23) at 100 yards just taking it nice and slow with zero adjustments for wind. This was right before heading out of town to hunt deer. Even though I was taking it slow, I did still screw the pooch on some of the shots.
View attachment 184912


I think the idea is to have a better or true zero. Which means, statistically, the more shots the better. If you were to put your gun in a machine that is a perfect device for repeatable shooting, the gun might shoot a few 0.50" groups at 100. But, the POI might be slightly different for all of those. If you overlay all those groups together with the same POA, the slightly different POI of the groups might give an overall group size of 0.75" or 1.00". A larger number of shots in a group gives a more accurate representation of the accuracy of the gun, excluding the external forces, like hot brass hitting you while you're shooting.

If you can shoot a 10, or 20, or 30, or even 100 round group at the same POA and eliminate any external forces (other shooters, getting tired, etc) then you have a much better representation of what your zero is. Of course, doing this with 30 rounds on a magnum caliber would be tough to do and have it be truly accurate of what your first few shots would be. Generally, a sample size of 10 is what is considered for something to be statistically valid.
 
Joined
Feb 17, 2017
Messages
1,287
Generally, a sample size of 10 is what is considered for something to be statistically valid.

This is the crux of my question. Where is the power analysis that explains or asserts this? Assigning “statistical validity” requires math around the variance in your individual replicates and your tolerance of uncertainty.

I totally believe that more shots = more statistical significance but it is likely not linear. I am a data scientist but shooting guns is not my field. It’s Form’s.

I 100% am in agreement that a sample size of 10 is better than 5. But you have to consider the error in your individual measurements, meaning your individual shots.

For a more accurate gun, with lower variance shot to shot, adding more rounds to your group tells you less and less. A 10 shot group isn’t 100% more statistically informative than a 5 shot group. Does it add 50%? 10%? 1%? That’s my question.

Is a 10 shot group the standard, or does the accuracy of the rifle dictate how many shots it takes to get a valid sample of: 1. How accurate is it and 2. What is my real zero?

Take an example from the extreme. If gun shoots all bullets into one hole, all the time, every time, guaranteed, why shoot more than just one round to determine group size or zero? Any experiment where sampling error is near zero requires a much lower sample size, and conversely high sampling error can often be overcome by beating it to death with sample size.

So with this in mind, how did we arrive at 10 shots?
 
OP
H
Joined
Feb 2, 2020
Messages
1,992
Take an example from the extreme. If gun shoots all bullets into one hole, all the time, every time, guaranteed, why shoot more than just one round to determine group size or zero? Any experiment where sampling error is near zero requires a much lower sample size, and conversely high sampling error can often be overcome by beating it to death with sample size.

So with this in mind, how did we arrive at 10 shots?

I see what you mean. I know 10 isn't a magical number that makes something statistically relevant; that's over simplifying it but in some cases that's how it works. I think I'd look at it more of as a rifle/shooter system. How many shots can you take that are as free of independent variables as possible, while still getting a large enough sample size to represent the accuracy of the individual rifle? Or how many shots does it take with the rifle/shooter system to know that any shot you take will have a 95% chance of landing in that group without having the shooter start making too many errors?

I have no idea. But 10 sounds good 😅
 
OP
H
Joined
Feb 2, 2020
Messages
1,992
That’s not how statistics work. Nobody is running a legitimate statistical analysis on a data set of 10. More data is definitely better than less, but there is no magical threshold once you hit 10.

I agree. It depends on the number of variables and is much more complicated. But, why not simplify it for people who aren't in a lab?
 

Formidilosus

Super Moderator
Joined
Oct 22, 2014
Messages
8,017
Is there some math involved around the significance of a 10 shot group? It seems like the burden of sample size keeps growing. Used to be 3, then 5, now 10. I seem to remember some software out there that calculates hit probability and I was wondering if a 10 shot group amounts to some sort of significance level or confidence interval.


Yes there is math, to start though it’s actaully opposite. Early 20th century the norm was 20, 30, even 50 round groups. Somewhere around Jack O’Connor’s time it started shifting, first to 10 round groups, then to 5, then finally to 3. This followed marketing and articles that were gear/rifle centric versus experience. The shift from large sample size groups to extremely small sample size was and is 100% about bullshitting users and purchasers as to reality. John Barsness has written about the history multiple times.


As for why 10? The short answer is given the precision of most rifles, a couple of 10 round groups gives around a 90% confidence of where any round from the rifle will fall. A 30 shot group is above 95% that any round will fall inside the cone of 30.


Where is the point of diminishing returns in the number of rounds piled into a group?


Yes, see above. For a rifle that produces a 30 rounds group of sub 2moa, the fall off over 30 rounds is very steep. For instance, the barrels tested in a Wisman rest at what is one of the worlds best ballistic tests facilities a couple years ago, all shot +/- 2 MOA for average 30 round groups. The first 10 round group produced a circle that covered between 80% to just under 90% of all 100 rounds that were fired. The kicker, was the cone of the first 30 rounds, encompassed just over 90% to 99% of all 100 rounds fired.

Very few legitimate hunting rifles/ammo is better than 2 MOA for 30 rounds.



Is there a linear relationship that can be assigned to group size (MOA) and number of shots in the group? If I shoot a 1/4 MOA 3 shot group, can I extrapolate a 1/2 MOA 5 shot (to a lesser confidence) and a 1 MOA 10 shot (to an even lesser level of confidence)?


No. Well there is, but it useless information.



The answer all of your questions, is in the question itself of- What is the point of shooting groups at all?
 

Formidilosus

Super Moderator
Joined
Oct 22, 2014
Messages
8,017
For a more accurate gun, with lower variance shot to shot, adding more rounds to your group tells you less and less. A 10 shot group isn’t 100% more statistically informative than a 5 shot group. Does it add 50%? 10%? 1%? That’s my question.

Is a 10 shot group the standard, or does the accuracy of the rifle dictate how many shots it takes to get a valid sample of: 1. How accurate is it and 2. What is my real zero?

So with this in mind, how did we arrive at 10 shots?


Yes.


The answer lies in the question as above- why shoot groups at all? What are you trying to achieve?
 

Sled

WKR
Joined
Jun 11, 2018
Messages
2,148
Location
Utah
No, there is no magic with a 10-shot group in terms of statistics. More is better than less, but no magic to 10.

for a hunting gun, like my tikka, my preference is 2 shot groups on steel or paper at varied ranges. i spread those out over time to get a sense of how it really shoots. i like that it gives me different conditions but i get a true cold bore shot and a follow up. add those groups up averaged for distance and i get an idea what distance that gun can hunt (if ballistically capable). that also helps conserve my hunting rounds so i'm not reloading all the time or spending hard earned coin on store ammo.

target guns get multiple rounds on steel at distance under challenging conditions.
 

16Bore

WKR
Joined
Mar 31, 2014
Messages
3,020
Epic ‘slide skullphuck!!!

Never saw so many panic attacks over something a simple as shooting a phuqqin rifle.
Take any 3 of the 10. Is that where you zero the crosshairs? What about the other 3? Or the other 3? Or how about these 4 over here because you “pulled one”. You know, just that “one”. Because you never “pull one” any other time, other than when it “really phuqqin counts” poking a hole in paper. But then the wind stirred that day and you had too much coffee and your vagina was sore. But my target shooting phone app tells me I whiffed one and there was a change in barometric pressure between shots 5 and 6. And I got distracted on shot 7 and had to pee on shot 8 which meant the barrel cooled before shot 9 and I accidently tinkled on my finger and shot 10 slipped.

Shut the hell up and shoot. Not everyone gets a trophy.
 
Joined
Feb 17, 2017
Messages
1,287
Yes there is math, to start though it’s actaully opposite. Early 20th century the norm was 20, 30, even 50 round groups. Somewhere around Jack O’Connor’s time it started shifting, first to 10 round groups, then to 5, then finally to 3. This followed marketing and articles that were gear/rifle centric versus experience. The shift from large sample size groups to extremely small sample size was and is 100% about bullshitting users and purchasers as to reality. John Barsness has written about the history multiple times.


As for why 10? The short answer is given the precision of most rifles, a couple of 10 round groups gives around a 90% confidence of where any round from the rifle will fall. A 30 shot group is above 95% that any round will fall inside the cone of 30.





Yes, see above. For a rifle that produces a 30 rounds group of sub 2moa, the fall off over 30 rounds is very steep. For instance, the barrels tested in a Wisman rest at what is one of the worlds best ballistic tests facilities a couple years ago, all shot +/- 2 MOA for average 30 round groups. The first 10 round group produced a circle that covered between 80% to just under 90% of all 100 rounds that were fired. The kicker, was the cone of the first 30 rounds, encompassed just over 90% to 99% of all 100 rounds fired.

Very few legitimate hunting rifles/ammo is better than 2 MOA for 30 rounds.






No. Well there is, but it useless information.



The answer all of your questions, is in the question itself of- What is the point of shooting groups at all?

Well said. Thanks Form. That answers what I was getting at. Much appreciated. Sorry to derail the thread; let’s see more groups
 
Joined
Jul 18, 2019
Messages
2,052
This is the crux of my question. Where is the power analysis that explains or asserts this? Assigning “statistical validity” requires math around the variance in your individual replicates and your tolerance of uncertainty.

I totally believe that more shots = more statistical significance but it is likely not linear. I am a data scientist but shooting guns is not my field. It’s Form’s.

I 100% am in agreement that a sample size of 10 is better than 5. But you have to consider the error in your individual measurements, meaning your individual shots.

For a more accurate gun, with lower variance shot to shot, adding more rounds to your group tells you less and less. A 10 shot group isn’t 100% more statistically informative than a 5 shot group. Does it add 50%? 10%? 1%? That’s my question.

Is a 10 shot group the standard, or does the accuracy of the rifle dictate how many shots it takes to get a valid sample of: 1. How accurate is it and 2. What is my real zero?

Take an example from the extreme. If gun shoots all bullets into one hole, all the time, every time, guaranteed, why shoot more than just one round to determine group size or zero? Any experiment where sampling error is near zero requires a much lower sample size, and conversely high sampling error can often be overcome by beating it to death with sample size.

So with this in mind, how did we arrive at 10 shots?

I’ve thought of taking a crack at this one day for the hell of it... taking a pic of the target every shot through my spotting scope and then seeing how the group size grows with shots fired.
 
Joined
Feb 17, 2017
Messages
1,287
Epic ‘slide skullphuck!!!

Never saw so many panic attacks over something a simple as shooting a phuqqin rifle.
Take any 3 of the 10. Is that where you zero the crosshairs? What about the other 3? Or the other 3? Or how about these 4 over here because you “pulled one”. You know, just that “one”. Because you never “pull one” any other time, other than when it “really phuqqin counts” poking a hole in paper. But then the wind stirred that day and you had too much coffee and your vagina was sore. But my target shooting phone app tells me I whiffed one and there was a change in barometric pressure between shots 5 and 6. And I got distracted on shot 7 and had to pee on shot 8 which meant the barrel cooled before shot 9 and I accidently tinkled on my finger and shot 10 slipped.

Shut the hell up and shoot. Not everyone gets a trophy.

Get out 16 the adults are talking. I figured you'd be in here eventually to paste your usual schtick. You should just start pasting this as all your posts:

1. buy a 30-06
2. stop having discussions on online forums
3. shoot more, because all of us can totally shoot all day every day

But in all seriousness, you're not wrong. Or, at least I agree with you. 10 shot groups add a bunch of shooter error. The vast majority of my 10 shot groups could have been stopped at 3 shots and told me everything I need to know from a practical standpoint. Honestly I can't remember a time where I shot a 3 shot group, felt good about the rifle, and then shot a 10 shot group and had that confidence shattered.

Nobody is having a panic attack, we're just nerding out. People will always nerd out despite your crusade to stop it. Is your scroll wheel broken? We'll ping you when we need to know if we should buy a 30-06 or not 😚
 
Joined
Feb 17, 2017
Messages
1,287
JAUB6932.PNG

Tell us about this rifle, to the trained eye it looks like it might have some potential
 
OP
H
Joined
Feb 2, 2020
Messages
1,992
But in all seriousness, you're not wrong. Or, at least I agree with you. 10 shot groups add a bunch of shooter error. The vast majority of my 10 shot groups could have been stopped at 3 shots and told me everything I need to know from a practical standpoint. Honestly I can't remember a time where I shot a 3 shot group, felt good about the rifle, and then shot a 10 shot group and had that confidence shattered.

I think one might be able to account for part of shooter error in 10 round groups by only using the shots that fall within 2 standard deviations of the average POI, which is supposed to account for something like 95% of the samples on the distribution curve. That is of course dependent on a person's ability and I'm not sure that would even work. I can't get too nerdy on it because I got out of taking a required statistics course in college due to my advisor's error 😂 I'll probably just do two or three 10 round groups, throw the shots out that are wayyyy off the rest (hopefully just a couple) and call it good.
 
Joined
Oct 8, 2019
Messages
2,956
Where is the official process for this including all relevant parameters?

What is the required distance for shooting the groups? 100 yards? 200 yards?
What is the required trigger weight?
What is the required cool down time between shots?
Does the group get disqualified if a cease fire is called while shooting the group?
What is the required shooting platform (ex: sled)
Must you have zero stimulants in your bloodstream?
Must you have stimulants in your bloodstream? If so, what type and how much?
What is the required shooting position? Sitting on a bench? Prone?

Without an agreed upon set of parameters, your desired outputs are not going to be very useful as everyone will do it their own way.
 

RCB

WKR
Joined
Apr 1, 2018
Messages
366
Location
CO
Is there some math involved around the significance of a 10 shot group? It seems like the burden of sample size keeps growing. Used to be 3, then 5, now 10. I seem to remember some software out there that calculates hit probability and I was wondering if a 10 shot group amounts to some sort of significance level or confidence interval.

Where is the point of diminishing returns in the number of rounds piled into a group?

Is there a linear relationship that can be assigned to group size (MOA) and number of shots in the group? If I shoot a 1/4 MOA 3 shot group, can I extrapolate a 1/2 MOA 5 shot (to a lesser confidence) and a 1 MOA 10 shot (to an even lesser level of confidence)?

These Tikka threads have become quite data driven, which I appreciate, and I want to know if the work has been done to vet our actual experimental design. It seems that the interest is there and guys want to do this right, so let’s do it right. Or at least understand what we’re doing beyond just “the internet wants 10 shot groups now”
Your question motivated me to do some quick stats. It's all theoretical but I'll share what I got.

First: The standard error of the estimate of the mean of any distribution is inversely proportional with the square root of the sample size. The sqrt of 10 is 3.16 and the sqrt of 3 is 1.63. What this means in practice is that, if you're trying to find your rifle zero, a 10 shot group will localize you to a point about twice as small as a group of 3. (Of course, that assumes fatigue doesn't set in and cause you to make bad shots in the 10-shot group!) So that's nice.

Second: Suppose, as nerdy statistician hunters, we wanted to know our 95% confidence target size. That is the size of target that you can hit 95% of the time. Why 95%? It's arbitrary. But I figure if you can hit a target 19 times out of 20, that's good enough for me. (You might ask - why not 100%? In statistics, you really never get to 100%, there's always *something* that can go wrong.) Now you might ask: given a certain n-shot group size, what is the best estimate for my 95%-confidence target size?

We can at least come up with an educated guess using statistical theory. Suppose horizontal and vertical deviations from your zero are independently normally distributed with the same standard deviation (lots of stats jargon there, can explain a bit if necessary). And suppose you are perfectly zero'd (a strong assumption).
It turns out that the *average* 3 shot-group from such a distribution will be just about 1/2 of your 95% confidence target size. In other words, if your *average* group size is 1 inch, then you ought to be able to hit a 2-inch kill zone 19 times out of 20.
Now, for 10 shot groups, it turns out the factor is about 1.3. In other words. If your average 10-shot group is 1 inch, you can probably hit a target of 1.3 inches 19 times out of 20.
(These were all determined by drawing 100,000 random shot groups from a normal distribution random generator.)
How accurate is all this analysis in the real world? I have no idea. I wouldn't put much money on it. Lots of assumptions there. But hopefully that gives some idea of the relative value of 10 vs. 3?

Always better to just get out and practice more.
 
Joined
Jul 18, 2019
Messages
2,052
Post up the white paper so we can see the standards, methodology, analyses, correlations, exceptions, conclusions, etc.

Lord knows there's a tough audience on here!

I believe what @Formidilosus is saying is that a "group" is a sub-sample of what he's calling the gun's "cone." The confidence in your sub-sample (group) will increase with sample size, i.e., decrease error in you measurement. Probably more appropriate here is the max value and making sure you get it on the paper since that's what defines a group (and of course is always a flier or "my fault" 😂). He can tell me if this is the right interpretation.

Not that I would ever get distracted at work, but a quick example is below. I just did this in a spreadsheet suing a random function for distance from point of aim, between 0" and 2" (if he's suggesting 2 MOA is probably a legit benchmark). With each shot being randomly drawn from the "cone" you can see that the more you shoot the greater the likelihood is that you touch the outer edges of the cone and get an accurate reflection of the gun's capability. You see that 3 shots looks like a solid rifle before all the fliers mess up the beauty!

For me? I shoot 3 and 5 shot groups mainly because I kill nearly everything within 200 yards and haven't had the need to refine much more. Plus my sexy 3 shot groups are fun to post so everyone can see how awesome I am.

@16Bore can poke more fun at me for looking at numbers, but ironically my 30-06 is in the back of my truck so I can do some shooting during lunch today... maybe that'll soften his judgement of my curiosity?! I was born in the 70s and have curve on my brim too if that helps.



1590680094168.png
 
Joined
Jul 18, 2019
Messages
2,052
Your question motivated me to do some quick stats. It's all theoretical but I'll share what I got.

First: The standard error of the estimate of the mean of any distribution is inversely proportional with the square root of the sample size. The sqrt of 10 is 3.16 and the sqrt of 3 is 1.63. What this means in practice is that, if you're trying to find your rifle zero, a 10 shot group will localize you to a point about twice as small as a group of 3. (Of course, that assumes fatigue doesn't set in and cause you to make bad shots in the 10-shot group!) So that's nice.

Second: Suppose, as nerdy statistician hunters, we wanted to know our 95% confidence target size. That is the size of target that you can hit 95% of the time. Why 95%? It's arbitrary. But I figure if you can hit a target 19 times out of 20, that's good enough for me. (You might ask - why not 100%? In statistics, you really never get to 100%, there's always *something* that can go wrong.) Now you might ask: given a certain n-shot group size, what is the best estimate for my 95%-confidence target size?

We can at least come up with an educated guess using statistical theory. Suppose horizontal and vertical deviations from your zero are independently normally distributed with the same standard deviation (lots of stats jargon there, can explain a bit if necessary). And suppose you are perfectly zero'd (a strong assumption).
It turns out that the *average* 3 shot-group from such a distribution will be just about 1/2 of your 95% confidence target size. In other words, if your *average* group size is 1 inch, then you ought to be able to hit a 2-inch kill zone 19 times out of 20.
Now, for 10 shot groups, it turns out the factor is about 1.3. In other words. If your average 10-shot group is 1 inch, you can probably hit a target of 1.3 inches 19 times out of 20.
(These were all determined by drawing 100,000 random shot groups from a normal distribution random generator.)
How accurate is all this analysis in the real world? I have no idea. I wouldn't put much money on it. Lots of assumptions there. But hopefully that gives some idea of the relative value of 10 vs. 3?

Always better to just get out and practice more.
I got distracted at the desk and started thinking along the same lines... tough thing is with groups is it's the max value, not the mean that defines the group... I was thinking one could take group measurements with each consecutive shot to see where the line flattens out. Will be pretty variable among rifles which is maybe why folks just stick with an easy number (like 10).
 
Top