How much "Extreme Spread" is acceptable in competition?

[Prose]

It's interesting that the weight of the fired primers is so consistent no matter the starting weight. It would suggest that the variation in weight is specific to the compound and not the cup.

Also interesting that there is no point where the lines intersect that could be blamed on some sort of outside factor.

 

[mikeeg02]

It's interesting that the weight of the fired primers is so consistent no matter the starting weight. It would suggest that the variation in weight is specific to the compound and not the cup.

This has been tested and told by a lot of people in the past, only to have it thrown in their faces and they were told they were wrong by the "experts". Ill keep sorting my primers........

 

[Prose]

@Mulligan you should use the same cases for 10 more shots with the exact middle weight of primers. I bet you'd have a pretty line that went right between the other two.

 

[Mulligan]

@Mulligan you should use the same cases for 10 more shots with the exact middle weight of primers. I bet you'd have a pretty line that went right between the other two.

I expect you are correct.
I really want to run the same test with my Dasher at 600 on paper.
CW

 

[LGKLAS]

Very interesting read. Thanx all for the comments .

 

[Jeff A]

All, the primer test is finished.
Thanks to all who helped with good advice. I have posted the test results below and I have 8 pages (with lots of pictures) of notes/explanation that are attached.
View attachment 1070701
@BoydAllen, Thanks Boyd.
@dkhunt14 Thanks Matt.
Should I start a new thread with a better title for this????

The report explains what I did and how I did it. Some might find this useful, it did answer a few questions I had, and I appreciate the help framing up the test.

CW

Grammatical edits.....
CW

Excellent report!

 

[Alex Wheeler]

This a breath of fresh air, nice post.

 

[Evan]

Again the initial signal:noise was clearly sufficient to make that conclusion.

Not true at all. The data set is too small to have any sort of valuable signal to noise analysis applied or quantification of error. You can see that heavy primers have an apparent noise of 30 to 40fps, and light primers are 10 to 20fps. The approx average change is 30 fps between light and heavy. The size of the noise and the difference between the two signals are very close. Visually it displays a clear trend that suggests what many suspect to be true: primer weight affects velocity significantly. But the results are from a driven test with a small sample set so any statistical analysis will have large uncertainty.

Mulligan, I'm not attacking your work. I think this is valuable and I appreciate that you took the time to do it and share it with us. It certainly re-enforces the belief that primer weight is a powerful factor in ES.

 

[Dusty Stevens]

Not true at all. The data set is too small to have any sort of valuable signal to noise analysis applied or quantification of error. You can see that heavy primers have an apparent noise of 30 to 40fps, and light primers are 10 to 20fps. The approx average change is 30 fps between light and heavy. The size of the noise and the difference between the two signals are very close. Visually it displays a clear trend that suggests what many suspect to be true: primer weight affects velocity significantly. But the results are from a driven test with a small sample set so any statistical analysis will have large uncertainty.

Mulligan, I'm not attacking your work. I think this is valuable and I appreciate that you took the time to do it and share it with us. It certainly re-enforces the belief that primer weight is a powerful factor in ES.

Theres a man that understands data and how to read between the lines of limited data and extrapolate a good guess.

 

[Toby Bradshaw]

Not true at all. The data set is too small to have any sort of valuable signal to noise analysis applied or quantification of error. You can see that heavy primers have an apparent noise of 30 to 40fps, and light primers are 10 to 20fps. The approx average change is 30 fps between light and heavy. The size of the noise and the difference between the two signals are very close. Visually it displays a clear trend that suggests what many suspect to be true: primer weight affects velocity significantly. But the results are from a driven test with a small sample set so any statistical analysis will have large uncertainty.

Mulligan, I'm not attacking your work. I think this is valuable and I appreciate that you took the time to do it and share it with us. It certainly re-enforces the belief that primer weight is a powerful factor in ES.

Run a t-test (I did) and get back to us. You will find that you are wrong.

 

[6MMsteve]

now I gotta weigh primers I quit

 

[CharlieNC]

Run a t-test (I did) and get back to us. You will find that you are wrong.

Thank you Toby, that was exactly the nature of my initial response in this thread. I recall something about a horse and water....

 

[Mulligan]

It's interesting that the weight of the fired primers is so consistent no matter the starting weight. It would suggest that the variation in weight is specific to the compound and not the cup.

Also interesting that there is no point where the lines intersect that could be blamed on some sort of outside factor.

I also thought it was interesting, that the weights of the spent primers varied so little across the board. I believe the folks at Federal have the business of building primers fairly well wired.
CW

 

[Evan]

Run a t-test (I did) and get back to us. You will find that you are wrong.

You're going to need to respond to what I said cause you're not making sense. Run a test on what? Test that Mulligans data set is too small to statistically analyze or draw scientific conclusions from?

Mulligans set reinforced what some of us believed and is very valuable. I'm agreeing that primer weight effects velocity but we need more data if we are going to use statistics in a rigorous fashion to derive scientifically sound conclusions. With more data points we could quantify the error and conclude what change in primer weight causes what amount of velocity variation.

Right now we can conclude that primer weight matters and that we should sort by weight to reduce variations, but we don't know what the acceptable weight variation within a set of sorted primers should be. Do we need scales that read 0.0001 grains and sort to that level? What velocity variation does that make compared to sorting to 0.001? With more data we could answer that.

 

[Toby Bradshaw]

You're going to need to respond to what I said cause you're not making sense. Run a test on what? Test that Mulligans data set is too small to statistically analyze or draw scientific conclusions from?

Mulligans set reinforced what some of us believed and is very valuable. I'm agreeing that primer weight effects velocity but we need more data if we are going to use statistics in a rigorous fashion to derive scientifically sound conclusions. With more data points we could quantify the error and conclude what change in primer weight causes what amount of velocity variation.

Right now we can conclude that primer weight matters and that we should sort by weight to reduce variations, but we don't know what the acceptable weight variation within a set of sorted primers should be. Do we need scales that read 0.0001 grains and sort to that level? What velocity variation does that make compared to sorting to 0.001? With more data we could answer that.

So far you have made 2 claims -- that Mulligan's dataset is too small to draw conclusions, and that primer weight matters. These can't both be true if based on the data shown.

Instead of typing more text, why not just run the t-test on Mulligan's data and report the results?

[When you do you will find that there is about a 1% chance that the reported muzzle velocities for heavy and light primers are drawn from the same population; i.e., a ~99% chance that the means are different.]

 

[CharlieNC]

So far you have made 2 claims -- that Mulligan's dataset is too small to draw conclusions, and that primer weight matters. These can't both be true if based on the data shown.

Instead of typing more text, why not just run the t-test on Mulligan's data and report the results?

[When you do you will find that there is about a 1% chance that the reported muzzle velocities for heavy and light primers are drawn from the same population; i.e., a ~99% chance that the means are different.]

Toby you're wasting your time suggesting someone use the appropriate tools (in this case the t test) when they have already made up their mind. Just like the initial data did not suit someone's idea of the definition of the scientific method.

 

[Dusty Stevens]

The way i see it ive been at an actual match that dkhunt14 was at. Ive seen him shoot well at 1000yds. If he says he rubs chicken feet on his ammo box before a match you should ask exactly where he gets his chicken feet so you can duplicate it exactly. Im an actual engineer every day so i can sure appreciate actual data but ive also been on the other end where the desk work doesnt agree with the real world.

 

[Brians356]

Toby you're wasting your time suggesting someone use the appropriate tools (in this case the t test) when they have already made up their mind. Just like the initial data did not suit someone's idea of the definition of the scientific method.

"A man convinced against his will ...".
-

 

[Brians356]

So far you have made 2 claims -- that Mulligan's dataset is too small to draw conclusions, and that primer weight matters.

To be fair to Evan, he didn't exactly say too small to draw conclusions (in fact he reached the same broad conclusion), he said "The data set is too small to have any sort of valuable signal to noise analysis applied or quantification of error." Your request for a T-test suggests you disagree with that more specific claim as well. Just trying to keep it real, it's a worthwhile discussion!
-

 

[Evan]

That's the first data set, presented as a scatter plot. Error bars on the average represent 2SD either way, capturing 95% IF it is gaussian. That's a difficult data set to make statistical predictions from. The overlap in error is huge. 1SD, 10fps, is a weight change of 0.04 gns or ~30% of full scale.