Chrony sample size

[deepcore]

What's the minimum sample size (number of rounds fired thru) do you consider brings significant data to the table when using your chronograph?

[Brian Pfleuger]

"Statistically valid" would be around 30, I think....

Real world, I use 3 or 5. That tells me all I need to know. The numbers are either very close and consistent or they're not. The odds of getting 3 close velocities in a row with a load that would not be consistent over-all are small (enough for me).

[alloy]

Sometimes 3 or 4 rounds is enough info for me to move to the next step in charge weight or whatever. I mean, if I shoot some 230 grain .45acp working up a new powder from a start load, and they chrono at 620 fps that's about all I need to know about that.
Closer to the final load I usually sample about 10 rounds.
Guess it depends on what I'm doing, and how accurate I am trying to be, tried some Powerbond projectiles at cast data once....620fps.

They use jacketed data.

[jdillon]

The larger the sample size, the better the statistical inference. My Chrony will remember up to ten shots in a string then will overwrite beginning at the first. Not sure it is really critical on pistols, but with rifles a larger sample size will provide a better estimate of the mean and standard deviation. Smaller samples tend to skew the data. At least that is what I can recollect from more than 20 credit hours in statistics but that was a few years ago

[saands]

It depends on what I am trying to accomplish.

If I am working up a load from a wildcat and all I have is my QuickLoad predictions, then I will shoot one across it and see if I am in the ballpark. Of course there is variability and we are trying to minimize that, but if I want to see if I am "in the ballpark" then I don't need a lot of precision.

Now ... if I am going to use the data to develop a ballistics table for one of my long range loads, I need to eliminate all the errors that I can, so I will run at least 10 and maybe 15 over it. And then I will reload another batch and run 10-15 of THOSE over it as well. That way I can take into account my own batch to batch variation.

The standard error is a measure of the difference between your sample mean and the actual mean of your whole population. The standard error goes down with the square root of the sample size ... so your error will be 1/10 as large for a sample of 100 as it will for a sample of just 1. Likewise, a sample of 9 will give you 1/3 as much error as a sample of 1. As you can see, adding more samples goes a LOOONG way in the beginning, but is pretty worthless after you already have several samples.

Saands

[rclark]

For quick velocity, accuracy, POA check 10-15 is what I use. For what goes into my books, I like to shoot ~30 to get the SD and ES. Talking revolver cartridges here.