Ten Shot Groups?

[Picher]

There's no way I could shoot a meaningful 20 shot group at our range unless there was no wind, mirage, water vapor, and all the stars and moon were aligned in a single row. Also, I would not have had any coffee for 5 days, and before that, Decaf for a month.

When I shot my 50 shot target, consisting of 10, 5 shot groups, conditions were perfect, but group center varied somewhat, from group to group, so something was happening. That's why I think you math guys may be wrrrr, wrrr, wrong. But, I wouldn't say that in a room full of math nuts.

 

[someguy2800]

There's no way I could shoot a meaningful 20 shot group at our range unless there was no wind, mirage, water vapor, and all the stars and moon were aligned in a single row. Also, I would not have had any coffee for 5 days, and before that, Decaf for a month.

When I shot my 50 shot target, consisting of 10, 5 shot groups, conditions were perfect, but group center varied somewhat, from group to group, so something was happening. That's why I think you math guys may be wrrrr, wrrr, wrong. But, I wouldn't say that in a room full of math nuts.

Well you just provided your own answer. If your group center varied from group to group you need to overlay all of them on top of each other to get the real average point of impact spread. If you selected one shot from each group randomly and overplayed them you would get a larger group size than your individual groups.

What is the goal? To make ourselves happy by seeing a small cluster of shots on the target or actually learning about the rifle?

 

[denton]

I think the problem though is a representative sample has to be a random selection from the full population. You can't shoot five shots and claim that is a "5 shot sample" of 30 shots or 500 shots. No, it is a total population of 5 shots. If you want to sample, you need to have at least 30 shots to randomly select samples. If you were generating shots anyway, then you could randomly (somehow?) select say 7 from that population and have a true sample. This works in manufacturing because you are manufacturing 1 million units to sell, and you want to sample a tiny percentage that will give you the statistical probability of compliance to a standard. But when you try to apply that to shooting, you find you still have to create those 30 shots anyway, so sampling less than what you have actually shot does not make sense. Now if you had 100 targets of 3-shots then I think you could sample by randomly selecting the proper percentage (I forget how to calculate the exact size the sample should be, but essentially any more or less is less valid then the exact proper number) of say 7-10 targets and overlaying them on POA. The resulting overlay of say 21 shots would then be a valid representation of all 300 shots.

As you point out, there is no getting around doing extensive testing to arrive at a valid statistical model. 3 shots or five shots tells very little on their own. Most guys shooting these I think mentally "overlay" a long series of 3-shot targets as they shoot the same load over the years. But claiming a load is a sub-MOA load after only shooting a couple of 3-shot or even 5-shot targets is nonsense. You MAY be able to say that load is likely more accurate that one shot with two targets of 2" groups but not necessarily. Shoot how ever many shots will ring your chimes, but be aware that you are not basing your decisions on enough information, and comparing your 3-shot results with some else's 10-shot or 30-shot results is meaningless.

It is true that to be valid a sample must fairly represent the population . It is definitely not true that you need 30 samples, unless you are doing tests based on Z scores, which practically nobody does since Gossett invented the T test back in about 1925.

 

[Casefull]

Most of group size doesn't matter near as much as the first cold bore shot. If you want an accuracy measurement that matters in the real world do the following. Take one shot each day for five days . That is the real group that your rifle produces. Most things I shoot at move after the first shot. Next do the same procedure standing without aid of a bench rest or the ground. That would be the shooters group size, as in take the shot in the next two seconds or lose the opportunity.

 

[Walkalong]

@DeadFlies that’s some nice shooting

Speaking of dead flies, this one landed on the wrong target at the wrong time.

 

[Nature Boy]

After my first F Class match a few months ago I realized I really did need to be “MOA all day”, and not just popping 3 shots in a paper plate right before hunting season MOA.

For those unfamiliar with F Class (which included me until 3 months ago) the objective is to put 20 rounds inside a 1 MOA circle at different distances from a prone position.

To that end, I’m shooting more 20 shot groups at 500 yards, both to verify loads and to practice. Today I was verifying seating depth. I shot this at 100 yards to determine the best distance from the lands. My interpretation of the results was to try 0.000.



So I loaded up 20 and went to 500 yards. After 5 shots I walked down and snapped a pic of the target then walked back and shot the remaining 15.



The first 5 at 500 yards measured 2.5” (0.5 MOA). That’s about as good as I can do. Could have stopped there but I needed to see how a 20 shot string would hold up. As you can see it roughly doubled in size to 4.75” or 0.9 MOA. I’m happy with that and will have no reservations taking that load to the next match.

 

[NWPilgrim]

It is true that to be valid a sample must fairly represent the population . It is definitely not true that you need 30 samples, unless you are doing tests based on Z scores, which practically nobody does since Gossett invented the T test back in about 1925.

Okay then explain to me how you get a smaller sample and not actually shooting the full population? Because what ever number you shoot IS the population. If you want a sample you have to take a random sample from that population of shots on target. But if you already have the 30 shot population why not use it instead of taking a random sample of the same?

You can;t shoot 5 shot in a row and call that a sample because it is not a random sample, it is five consecutive data points.

 

[TN Outlaw]

So, i guess it depends on if your shooting groups to test your skills or the rifle/ammo combo. After so many groups your testing yourself from shooter fatigue, not just your firearm.

Laying out groups that have been fired from several diff range trips and over-laying them for POI does not work either. Each day brings about diff wind cond, humidity, and some days the shooter has good days while others are bad. Of course POI will shift some. Heck, as the day goes on from morn to mid day to aft, conditions change as will POI.

What are these groups testing for? That determines how many rounds should be fired. That and how many the shooter wants to fire.

 

[denton]

Okay then explain to me how you get a smaller sample and not actually shooting the full population? Because what ever number you shoot IS the population. If you want a sample you have to take a random sample from that population of shots on target. But if you already have the 30 shot population why not use it instead of taking a random sample of the same?

You can;t shoot 5 shot in a row and call that a sample because it is not a random sample, it is five consecutive data points.

The population is all the shots that will ever be fired from the rifle, or all the shots using a particular load that will ever be fired from the rifle. You will never know the true population unless you shoot the barrel until it's worn out, and by then the information is no longer useful because you're not shooting that barrel anymore. By sampling, you choose to accept imperfect, but useful and timely information.

A sample is any subset we choose to represent the population, whether that's 2 shots, 5 shots, or 25 shots. Some samples are better than others.

In order to work, the sample must fairly represent the population. Having a randomly drawn sample is one way to do that, but not the only way. Every Nth item samples are considered good practice, for example.

The key question for whether a sample represents a population is whether all the input variables to the process have an opportunity to express themselves. So if you're doing a sample on a calm day at 72 degrees with a clean barrel and a particular load and letting the barrel cool well, five consecutive shots fairly represents those conditions. It's nice if you can get a random sample, but fairly often you can't. If your experiment involves furnace temperatures, it is impractical to heat and cool the furnace randomly. So you typically do not randomize that part of an experiment. You take the best number you can get.

 

[wally]

Today I was verifying seating depth. I shot this at 100 yards to determine the best distance from the lands. My interpretation of the results was to try 0.000

This and neck sizing only (not usually practical for autoloaders) are the "conventional wisdom" for maximum accuracy loads

 

[Peakbagger46]

I’d rather shoot three shot groups for testing and spend the rest of my time/ammo on either distance shooting or practical drills.

 

[Picher]

Well you just provided your own answer. If your group center varied from group to group you need to overlay all of them on top of each other to get the real average point of impact spread. If you selected one shot from each group randomly and overplayed them you would get a larger group size than your individual groups.

What is the goal? To make ourselves happy by seeing a small cluster of shots on the target or actually learning about the rifle?

It's not just the rifle that causes groups to wander a bit. It can be atmospheric conditions, the particular hold/rest, etc. My center didn't vary a whole lot per group, but since the rifle only held 10 shots, I had to reload and reset the rifle and my hold. However the 10 groups of 5 shots on the same piece of paper was quite telling and certainly worth trying.

 

[someguy2800]

It's not just the rifle that causes groups to wander a bit. It can be atmospheric conditions, the particular hold/rest, etc. My center didn't vary a whole lot per group, but since the rifle only held 10 shots, I had to reload and reset the rifle and my hold. However the 10 groups of 5 shots on the same piece of paper was quite telling and certainly worth trying.

Of course, we are not shooting in a laboratory, changing conditions is just part of shooting. My point is that just because a rifle can put 5 shots into a certain size group several times does not mean that is its true accuracy. For example lets say you shot a 3 or 5 shot group with a rifle, and then placed a sticker exactly over that group so that it just covers the edge of every hole. Could you place all of the next 5 shots onto that dot without touching the edge? How about the next 20 shots?

Anyway getting back the original post yes I like to shoot larger samples since it teaches me more about what I can expect from the rifle than small samples do. I have moved my shooting practice away from shooting group sizes on paper and more towards hitting actual targets at distance so this is valuable info for me. There is no practical situation where the wind is going to wait for me, so when the wind blows I dope the best I can (which is not much) and pull the trigger.

 

[TN Outlaw]

Of course, we are not shooting in a laboratory, changing conditions is just part of shooting. My point is that just because a rifle can put 5 shots into a certain size group several times does not mean that is its true accuracy. For example lets say you shot a 3 or 5 shot group with a rifle, and then placed a sticker exactly over that group so that it just covers the edge of every hole. Could you place all of the next 5 shots onto that dot without touching the edge? How about the next 20 shots?

Anyway getting back the original post yes I like to shoot larger samples since it teaches me more about what I can expect from the rifle than small samples do. I have moved my shooting practice away from shooting group sizes on paper and more towards hitting actual targets at distance so this is valuable info for me. There is no practical situation where the wind is going to wait for me, so when the wind blows I dope the best I can (which is not much) and pull the trigger.

That seems like testing how well you can hit with that particular rifle/load combo more so than pure accuracy testing. Which works very well for what you use the rifle for. I dont blame you for practicing that way but wanting to know purely what the rifle is capable of, any differing factors degrade the data. Heck, raising up to do a mag change and then cheeking up to a diff point on the stock can and will change POI. Its hard to shoot another group to the exact same section of the target because of the shooters mistakes, wind, bbl temp being diff, humidity, etc. Not just the guns ability.

If all conditions are the exact same somehow, i will agree the more groups shot and even the more rounds per group will give more good data to draw from and tell you more about what the rifle is capable of.

The only problem i see is looking at all these different groups from different days and overlaying them as one. Having a group shot at 37º with a 5-9mph W wind and another shot at 49º with a 14mph SE wind, then one group shot in 86º with a 7mph SW wind and comparing POI for all the groups as one, does nothing for telling us how tight our rifle will shoot. They are ALL seperate data points. If you want to know POI for a particular wind/humidity, it should work great. Not for a pure accuracy test though.

Now, measuring all these groups and comparing for avg group size, that seems more feasible to me.

 

[Picher]

Another problem with shooting .22LR groups is the relatively imperfect ammo. Even the best match ammo has enough variation as to cause POI and grouping variations from batch to batch. I've tested a lot of ammo as part of my Rimfire Benchrest endeavors over several years and it's an imperfect "world". Once in a while you find a batch of a particular brand/type that shoots great. When you do, you should buy even more of it than you really can afford.

 

[d2wing]

It depends on the situation. My primary goal is seeing where the first shot out of my deer rifle is going. Target rifles are different. If my rifle is already sighted in I shoot a 3 round group for confirmation. But if I am sighting in a new scope, testing ammo or a new rifle I shot 5 round groups usually. I may shoot ten rounds. For a variety of reasons some rifles change POI when they heat up. So more than three round can be counter productive for a hunting rifle. If I am shooting a target rifle I like to shoot ten round groups to see what happens. It depends on what the goal is and what rifle you are shooting for me. I do not believe in excluding flyers or cherry picking groups. I like to know where the bullet is going with certainty. There are no excuses in hunting or competition.I see no point in being dishonest with myself or on line.