Forums
New posts
Search forums
What's new
Articles
Latest reviews
Author list
Classifieds
Log in
Register
What's new
Search
Search
Search titles and first posts only
Search titles only
By:
New posts
Search forums
Menu
Log in
Register
Install the app
Install
Forums
Rifles, Reloading, Optics, Equipment
Reloading
Hornady Podcast
JavaScript is disabled. For a better experience, please enable JavaScript in your browser before proceeding.
You are using an out of date browser. It may not display this or other websites correctly.
You should upgrade or use an
alternative browser
.
Reply to thread
Message
<blockquote data-quote="QuietTexan" data-source="post: 2692672" data-attributes="member: 116181"><p>Not at all. They said make statistically meaningful changes, and build a statistically meaningful population of data points. The guy with the moustache said he does powder charges in 1gn intervals because any smaller of an increment gets lost in the noise of variances in multiple other variables. If someone were to do a charge weight ladder in 0.1gn increments there would be so much noise that unless you shot 30+ shots of each increment there would be no way to quantify the change from each increment. So make big jumps to skip over the noise patterns.</p><p></p><p>This is a graphic from PRB re: seating depth testing, you can see how individual data point lines vary within a range, and in this case he used I think 10 shots per increment to show the actual vertical dispersion range versus the variance in individual data points. In this case it shows someone would be better off to do the bulk "Berger jumps" increments of 0.040" unless they wanted to commit to shooting a meaningful number of shots at each 0.005" increment. That's the balance - to prove the change caused by very fine adjustments takes very large samples. For larger, more coarse adjustments you can use smaller sample sizes because the shifts move more than the larger margin of error from the smaller sample size. The podcast actually mentioned that specifically - small samples do not PROVE anything, what they do is RULE OUT combinations. A very small 5 shot group doesn't confirm that a load is good, but a very bad 2 shot group will rule out a combination because it can't possibly do better than the bad result it already showed you. To prove something (aka show some kind of causal link) takes large sample sizes, so if you aren't going to collect large sample populations, then you have to make changes that are more than the margin of error.</p><p></p><p>[ATTACH=full]420699[/ATTACH]</p><p></p><p></p><p>[URL unfurl="true"]https://precisionrifleblog.com/2020/11/29/statistics-for-shooters/[/URL]</p></blockquote><p></p>
[QUOTE="QuietTexan, post: 2692672, member: 116181"] Not at all. They said make statistically meaningful changes, and build a statistically meaningful population of data points. The guy with the moustache said he does powder charges in 1gn intervals because any smaller of an increment gets lost in the noise of variances in multiple other variables. If someone were to do a charge weight ladder in 0.1gn increments there would be so much noise that unless you shot 30+ shots of each increment there would be no way to quantify the change from each increment. So make big jumps to skip over the noise patterns. This is a graphic from PRB re: seating depth testing, you can see how individual data point lines vary within a range, and in this case he used I think 10 shots per increment to show the actual vertical dispersion range versus the variance in individual data points. In this case it shows someone would be better off to do the bulk "Berger jumps" increments of 0.040" unless they wanted to commit to shooting a meaningful number of shots at each 0.005" increment. That's the balance - to prove the change caused by very fine adjustments takes very large samples. For larger, more coarse adjustments you can use smaller sample sizes because the shifts move more than the larger margin of error from the smaller sample size. The podcast actually mentioned that specifically - small samples do not PROVE anything, what they do is RULE OUT combinations. A very small 5 shot group doesn't confirm that a load is good, but a very bad 2 shot group will rule out a combination because it can't possibly do better than the bad result it already showed you. To prove something (aka show some kind of causal link) takes large sample sizes, so if you aren't going to collect large sample populations, then you have to make changes that are more than the margin of error. [ATTACH type="full" alt="Bullet-Jump-for-Berger-105gr-Hybrid-with-Trend-Lines-645x450.jpg"]420699[/ATTACH] [URL unfurl="true"]https://precisionrifleblog.com/2020/11/29/statistics-for-shooters/[/URL] [/QUOTE]
Insert quotes…
Verification
Post reply
Forums
Rifles, Reloading, Optics, Equipment
Reloading
Hornady Podcast
Top