sample size

  • 400 Views
  • Last Post 21 July 2024
  • Topic Is Solved
Ross Smith posted this 19 July 2024

I broke down and bought the garmin chrono. What a breeze to use. It has pointed out some of my wrong thinking. My question is, what is  good sample size to find a low standard deviation? Is shooting single five shot groups good enough or should the shot string be more? I assume the more shots the better?

Attached Files

  • Liked by
  • Aaron
Order By: Standard | Newest | Votes
Ross Smith posted this 19 July 2024

I found some discussion on this by Larry Gibson and Bill Cook. Pretty well answered my question and gave me more insight on using a chrono in the first place. It pays to search the archives.

Attached Files

John Alexander posted this 20 July 2024

You are right of course more shots are better.  Searchiing the archives is a good idea.

The key question is: if I go through the range of loads in question with single groups for each load, is the one showing the lowest SD telling me the truth? The way to find out is to do it again exactly the same. If it gives you the same load as lowest SD, that should give you some confidence that really is same load. How much confidence is needed is up to you.  Don't be surprised i the results of the second go through rank the best of the first go through as not nearly best.

There is no substitute for lots of shots.

We fool ourselves all the time thinking we have found the best load using single 5 or 10 shot group pairs or a ladder approach.  And if we tried to confirm by doing it again we would find a different best load.

Mother Nature doesn't give an inch on probabilities, whether we like to ignore them or not.

John

John

Attached Files

tomme boy posted this 20 July 2024

Having everything as consistent as possible will prove your load. Don't mix 1x fired brass with 5x fired. Don't switch bullets you made last month with ones you made last week. Don't use different lots of powder. 

 

Mixing these up will get you different results. Consistency in EVERYthing you do will give you the best results. You start switching things around and that is where some of the flyers come from. I even go as far as when seating if I feel a bullet go in harder or softer that round is set aside and used as a fouler. 

Attached Files

  • Liked by
  • muley
  • Ross Smith
Aaron posted this 20 July 2024

Just know that as shooters, we will never see a proper sample size. If you load 5 and chrono all 5, you used the population (N) as your sample size (n). You haven’t “sampled” anything. If I make 10,000 rounds and randomly pull 50 as a sample, i can chrono the 50 and use that data to say with x% of certainty, that the population of 10,000 (N=10,000 and n=50) has the characteristics demonstrated by the sample with x% certainty. If you shot your entire population (N) and you use that data, you can NOT apply that data to a subsequent population. The subsequent population is not of the previous population and must be tested on its own merits.
I know I am drilling deep here but the use of statistical terms and process to evaluate a miniscule number of data points is deceptive. One can say “it’s safe to assume my next 10 cartridges will perform similar to the previous 10” but leave it at that. Applying statistical terminology and values to 10 data points is not useful. Statistical validity comes with LARGE data sets. I used to do Statistical Process Control in corporate. Did in the USMC on the rifle range too. Data became very valuable with huge numbers of data sets on the range when they evaluated our hit potential based on years of data, not just one range session.

With rifle in hand, I confidently go forth into the darkness.

Attached Files

  • Liked by
  • RicinYakima
  • Ross Smith
Larry Gibson posted this 21 July 2024

A ten shot sample is industry standard [SAAMI and other test organizations] with additional 10 shot samples used as confirmation.  I use 3 ten shot sample tests for confirmation of my own loads.  I use ten shot sample tests for requested tests.  A ten shot sample gives a 95% +/- surety the rest of the same lot will fall within a +/- of the test sample averages, and the ES/SD figures.

Understand that SAAMI and other test organizations actually use three different pressure measurements [ MAP, MPLM & MPSM].  Factories load different lots of different cartridges to the respective advertised cartridge velocity with a +/- acceptable variation as long as the measured pressures remain within the three pressure limits set for each cartridge.  That acceptable velocity variation is larger than many think.

I've been chronographing (Oehler chronographs) ammunition (reloads and factory) since '74.  Dr. Oehler has confirmed that a 10 shot chronographed test is a good indicator of what a specific lot will do with additional 10 shot test to confirm.  Again, that is what I do and it has proven to be a sufficient and reliable source of velocity data.  Keep in mind that the average velocity, the ES and the SD figures will most often be slightly different.  That is to be expected, especially if the different tests are not conducted back to back under the same external conditions.  

LMG

Concealment is not cover.........

Attached Files

Tom Acheson posted this 21 July 2024

Speaking of the Garmin…

At a recent match, a friend loaned me his Garmin. I ran a 10-round group with the Garmin. We picked it up to read the summary. Then with the same load, gun, etc. did another 10-round group.

The second group averaged about 100 fps faster. Didn’t make sense.

Theory…. If you don’t put the unit in exactly the same location, it develops different values. Maybe it’s a bit farther to the side or a bit farther along the length of the gun, or cocked slightly different in how it is aimed at the target, etc.

Maybe mark the bench where the 3 feet are located and always return the unit to those 3 marks?

Tom

Attached Files

  • Liked by
  • Bud Hyett
  • Ross Smith
Categories
Search
Popular Tags
This Weeks High Earners
Hot Topics
Close