Statistics for Handloaders

[chkunz]

There is a great article in the 46th edition of the Lyman Reloading Handbook titled "Statistics for Handloaders". This article provides a simplified version of how to use statistical analysis to evaluate data from the reloading process that would be helpful to a lot of us handloaders. For example, if you change a component in a load you can use some elementary statistical analysis to tell you if there is a probability that the change improved the accuracy of the load. There are very good instructions and charts and tables and example problems worked out that make it easy to follow and use for your analysis of your data. We all have our methods that we use for analysis of our handloads but this will provide an additional tool that you may find beneficial.

If you have a copy of this reference and have the skills to post a copy of this article I would appreciate it very much and I think others would also.



Take care, Clyde

 

[K22]

The Lyman Reloading Handbook is one of the best if not the best especially for beginners. Don't have this edition but I'd be interested in how they use statistical analysis to predict accuracy of load changes.

 

[ebb]

I usually rely on the target to give me that information. I find it is the most reliable source.

 

[CharlieNC]

I usually rely on the target to give me that information. I find it is the most reliable source.

When testing loads, techniques, etc it is very easy to be misled (that is wrong) by the target by simply jumping on the smallest group size. Statistics can be used to determine if there is a real difference.

 

[NorCalMikie]

it is very easy to be misled (that is wrong) by the target by simply jumping on the smallest group size.

What else is there? If you can't hit it, why bother? Too fast? Too slow? If they're all going in the same hole,you're done.

 

[CharlieNC]

What else is there? If you can't hit it, why bother? Too fast? Too slow? If they're all going in the same hole,you're done.

Since they don't go into the same hole every time, you need to know when a change gets you closer to the one hole performance vs just seeing normal variability on the target. When flipping a coin you will get a heads 50% of the time. You can get a number of heads in a row, but its still a 50% proposition and doesn't mean you have a miracle coin.

 

[memilanuk]

What else is there? If you can't hit it, why bother? Too fast? Too slow? If they're all going in the same hole,you're done.

Fliers don't always go 'out' of the group...

 

[NorCalMikie]

When flipping a coin you will get a heads 50% of the time

Not with MY 2 headed quarter.

Fliers don't always go 'out' of the group..

We talking a group as in a bunch of bullets just hanging around together or a 5 shot "GROUP" that can be covered with a dime?

 

[dsculley]

How many times have you heard of someone locating a load that shot bugholes during development only to find that it was not repeatable? Maybe it has happened to you. When used right statistics can be a powerful tool,

 

[memilanuk]

Not with MY 2 headed quarter.

We talking a group as in a bunch of bullets just hanging around together or a 5 shot "GROUP" that can be covered with a dime?

We're talking when you shoot a group, and you *know* you didn't break the shots *that* good, but hey, they still went in the middle. Yep, lets go with that load. Its 'magic'... riiiiight.

While I am a pretty firm believer in statistics in general, I do take some exception to the 'shoot every load until you have a statistically significant sample' camp. Mainly in that there has to be a balance between collecting enough data to be reasonably confident, and actually using the damn gun for what it was intended for. Most people want to be 99.999% sure, with only three shot groups. *That* rarely works out, in my experience.

 

[NorCalMikie]

How many times have you heard of someone locating a load that shot bugholes during development only to find that it was not repeatable? Maybe it has happened to you.

If that load wasn't repeatable, you didn't do enough load workup. And all a 3 shot group will do is get you on paper, if you're lucky. And I don't go with "one" 5 shot group either.

 

[KevinThomas]

Interesting thread, especially given the fact that the author of the piece in question here plainly agrees with the last few posts in this thread. I just re-read the piece, and he was pretty adamant about NOT discounting flyers, and that sort of thing. I've long joked that the most hated words the average handloader will ever hear is "statistically valid sample size." The loose translation of that is, "far more shots that the average shooter is willing to expend."

 

[chkunz]

Interesting thread, especially given the fact that the author of the piece in question here plainly agrees with the last few posts in this thread. I just re-read the piece, and he was pretty adamant about NOT discounting flyers, and that sort of thing. I've long joked that the most hated words the average handloader will ever hear is "statistically valid sample size." The loose translation of that is, "far more shots that the average shooter is willing to expend."

Kevin, do you have the capability to post a copy of the article? Some folks are critical and have not even read it. For those folks that have an open mind toward the use of statistics for data analysis it might be helpful. Also, every shot we shoot is data so we are not limited to load development at the bench for our analysis. Best wishes, Clyde.

 

[KevinThomas]

Kevin, do you have the capability to post a copy of the article? Some folks are critical and have not even read it. For those folks that have an open mind toward the use of statistics for data analysis it might be helpful. Also, every shot we shoot is data so we are not limited to load development at the bench for our analysis. Best wishes, Clyde.

Unfortunately, I do not.

 

[chkunz]

Unfortunately, I do not.

Also, I thought this would be worth posting in the daily "Bulletin". But maybe not since it got negative response from several folks.

 

[KevinThomas]

Personally, I think it would. Lyman may not be opposed to seeing it posted, so long as proper attribution was given. It was some good info, whether some choose to acknowledge it or not.

 

[jepp2]

I have it, but will wait to post it until I secure permission from Lyman. It involves 8 pages, and that much material needs permission.

 

[CharlieNC]

Interesting thread, especially given the fact that the author of the piece in question here plainly agrees with the last few posts in this thread. I just re-read the piece, and he was pretty adamant about NOT discounting flyers, and that sort of thing. I've long joked that the most hated words the average handloader will ever hear is "statistically valid sample size." The loose translation of that is, "far more shots that the average shooter is willing to expend."

And statistically valid sample size is an abused and misused term! For example a ladder test is often run using one shot per load to obtain valid results, ie the node as the flat spot of an overall curve. The sample size conundrum arises when one desires a high level of confidence about a single answer, taken in isolation. Target group size, like velocity ES, requires more samples to achieve confidence than using the SD of the same data. But why rehash the same....

 

[jepp2]

If you have a copy of this reference and have the skills to post a copy of this article I would appreciate it very much and I think others would also.

Good Afternoon Jim,


As long as we are credited as the source than yes you have our permission to share those pages.


Best Regards,



Customer Service

Lyman Products Corp.

475 Smith St., Middletown, CT 06457

Phone: 800-225-9626

So for the benefit of anyone wanting to see this information, I have a pdf file of the 7 pages. Too big to post here, but PM me with your email address, and I will gladly send it to you. I will include the first 4 pages in this reply, so you can see a preview.

 

[jepp2]

And here are the other 4 pages