GROUP DIFFERENCES

  • 242 Views
  • Last Post 21 September 2018
joeb33050 posted this 18 September 2018

 

GROUP DIFFERENCES

 

TO SEE IF THE GROUPS SHOT WITH TWO LOADS HAVE DIFFERENT ACCURACY;

 

SHOOT THE SAME NUMBER OF GROUPS WITH EACH LOAD;

 

MEASURE EACH GROUP;

 

FIGURE THE AVERAGE GROUP SIZE FOR EACH LOAD

 

DIVIDE LARGER AVERAGE/SMALLER AVERAGE, GET “TEST RATIO”,

 

LOOK UP THE TABLE ENTRY.

 

IF TEST ENTRY > TABLE ENTRY, WE ARE ~ 95% SURE THAT THE LOAD ACCURACIES ARE DIFFERENT.

 

IF TEST ENTRY < TABLE ENTRY, WE ARE NOT~ 95% SURE THAT THE LOAD ACCURACIES ARE DIFFERENT.

 

EXAMPLE: SHOTS / GROUP = 5, FOR 15 GROUPS SHOT WITH EACH LOAD.

 

THE TABLE ENTRY IS 1.11.

 

 DIVIDE LARGER AVERAGE/SMALLER AVERAGE, GET “TEST RATIO”.

 

IF TEST RATIO > TABLE ENTRY, EX:1.18, WE ARE ~ 95% SURE THAT THE LOAD ACCURACIES ARE DIFFERENT.

 

 

 

IF TEST RATIO < TABLE ENTRY, EX: .98, WE ARE NOT ~ 95% SURE THAT THE LOAD ACCURACIES ARE DIFFERENT.

 

 

 

Attached Files

Order By: Standard | Newest | Votes
Scearcy posted this 18 September 2018

Bravo Joe!!!!

I am going to reproduce this table on a card and tape it to my loading room wall.

Jim

Attached Files

John Alexander posted this 18 September 2018

 

Ditto about pasting the table in my reloading room. Thanks Joe.

This is not a criticism of Joe's contribution but unfortunately it is often impractical to achieve a 95% confidence level in many kinds of decisions including deciding what load is best by shooting groups. Very few of us are going to shoot 20 5-shot groups of each load to decide which load is better when the difference in average group size is 10%. If we do shoot that many groups for one decision we won't have time to test for many other loads and maybe miss something better.  That would be counterproductive so we have to make decisions with less than 95% confidence levels.  That is OK we do it all the time for other decisions.

We seldom have the luxury of having a 95% confidence level for life decisions about things much more important that which load is best (to change jobs, to continue our education, get married, invest in which stock etc.).  The confidence level must be much less for these decisions (50% of marriages fail, an investor would soon be rich if he guessed which way a stock was going to move even 70% of the time and few are.)

It seems to me that what we need to offer us practical support for decision making is a table like Joe's above but based on a confidence level of 60 or maybe 80% it would require a number of groups that we might actually fire and being right most of the time wouldn't be bad. Something to think about.

John

Attached Files

RicinYakima posted this 18 September 2018

I'm taping this up to remind myself that it is the numbers that count.

Attached Files

RicinYakima posted this 19 September 2018

That gets laminated and posted just below. Thank you. Ric

Attached Files

John Alexander posted this 20 September 2018

Terrific Joe.  Thanks.

Now it would be possible to make all those little decisions about which load is best with a consistent level of confidence.  I could be learning to love statistics after all.

John

woops just looked again and numbers for 95% confidence level as in first table? What am I missing?

Attached Files

joeb33050 posted this 20 September 2018

the first time around I used the Z statistic, because the number of groups >30 and I dislike the t statistic.

For up to 30 groups, the second dance, I used t. There's not a lot of difference, as you noted.

Here's 5 and 10 shot groups:

 

I put this stuff up because there's no critic/editor/finder of errors here. I'm looking for criticism, to avoid or remove mistakes.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Attached Files

Scearcy posted this 20 September 2018

Joe should the 95% and 85% rows be the same for the 5 shot groups or is that a typo?

Great stuff.

Jim

 

Attached Files

John Alexander posted this 21 September 2018

Thanks Joe.

Gives some of us that like numbers valuable information not easily available.

John

Attached Files

joeb33050 posted this 21 September 2018

Joe should the 95% and 85% rows be the same for the 5 shot groups or is that a typo?

Great stuff.

Jim

My mistake, fixed now. Thanks;

joe b.

 

 

Attached Files

Close