Question on Calculating Extreme Spread

powerspc

It Sounded Like a Good Idea at the Time
Full Member
Minuteman
Supporter
  • Mar 15, 2018
    1,274
    5,075
    You shoot a five shot group same charge, brass, primer and bullet seating depth and the velocity records as such:

    1: 2640
    2: 2636
    3: 2642
    4: 2595
    5: 2635
    E.S. with #4: 47, E.S. without #4: 7

    I would consider shot #4 as an outlier as compared to the group; so you shoot another 5 shot group same charge, brass, primer and bullet seating depth and see this:

    1: 2638
    2: 2605
    3: 2641
    4: 2635
    5: 2640
    E.S. with #2: 36, E.S. without #2: 6

    I would consider shot #2 as an outlier in this group and as compared to the previous group, question:

    Is it usual, customary, “permissible” to throw out the outliers when calculating E.S., or does that defeat the whole purpose of calculating it in the first place?
     
    If you seperated the cases that were the outliers they might tell a story. The reason to not throw out the number is you have an inconsistency in the loading process, components or assembly. This is a good thing since you’re 80% there.
     
    If it is your first shot that yields the discrepancy, I'd entertain discarding it, even a magneto can biff the first rd. But, here, I would up my sample size, to say 15 or 20rds. Of coarse you can't just hammer 20 quick rds, get a rhythm, and go. Then evaluate.
    Also, you need to shoulder your gun the same each firing, if you fire a chrono rd free recoil, your velocity is dropping.
     
    In general, data points should not be discarded out of hand. Rather than guess, there is a simple statistical method to assist in evaluating whether a data point might be a statistical "outlier" and then perhaps discarded.

    The maximum and minimum are very sensitive to outliers. This is for the simple reason that if any value is added to a data set that is less than the minimum, then the minimum changes and it is this new value. In a similar way, if any value that exceeds the maximum is included in a data set, then the maximum will change.

    Rank the data from minimum to maximum values.

    2595
    2635
    2636
    2640
    2642

    The first quartile Q1 - this represents a quarter of the way through the list of all the data

    Q1 = 2635

    The third quartile Q3 - this represents three quarters of the way through the list of all the data

    Q3 = 2640

    The interquartile range (IQR) rule is useful in detecting the presence of outliers. Outliers are individual values that fall outside of the overall pattern of the rest of the data. This definition is somewhat vague and subjective, so it is helpful to have a rule to help in considering if a data point truly is an outlier.

    The interquartile range shows how the data is spread about the median (in this case 2636).
    It is less susceptible than the range / extreme spread to outliers.

    Calculate the interquartile range for the data = Q3 - Q1

    IQR = 2640 - 2635 = 5

    Next, multiply by 1.5
    IQR X 1.5 = 7.5

    Subtract 1.5 from Q1.
    Q1 - 1.5 = 2633.5 any number less than this is suspect to be an outlier i.e. 2595

    Add 1.5 to Q3.
    Q3 + 1.5 = 2641.5 any number greater than this is suspect to be an outlier i.e. 2642



    It is important to remember that this only is a rule of thumb and generally holds . Any potential outlier obtained by this method should be examined in the context of the entire set of data.
     
    Last edited:
    or does that defeat the whole purpose of calculating it in the first place?
    Yes it does.
    Why not do this first, take 10 rds, get set up, manhandle the gun the same each shot, do not even aim, just close your bolt and fire, get 10 shots off in under a minute.
    Right now, you are at 20% outliers as you call them, not acceptable, take your technique out of the equation first. I do not feel this will be all that easy to solve, seat depth or neck tension differences are not going to give you a 40fps spread.
     
    In general, data points should not be discarded out of hand. Rather than guess, there is a simple statistical method to assist in evaluating whether a data point might be a statistical "outlier" and then perhaps discarded.

    The maximum and minimum are very sensitive to outliers. This is for the simple reason that if any value is added to a data set that is less than the minimum, then the minimum changes and it is this new value. In a similar way, if any value that exceeds the maximum is included in a data set, then the maximum will change.

    Rank the data from minimum to maximum values.

    2595
    2635
    2636
    2640
    2642

    The first quartile Q1 - this represents a quarter of the way through the list of all the data

    Q1 = 2635

    The third quartile Q3 - this represents three quarters of the way through the list of all the data

    Q3 = 2640

    The interquartile range (IQR) rule is useful in detecting the presence of outliers. Outliers are individual values that fall outside of the overall pattern of the rest of the data. This definition is somewhat vague and subjective, so it is helpful to have a rule to help in considering if a data point truly is an outlier.

    The interquartile range shows how the data is spread about the median (in this case 2636).
    It is less susceptible than the range / extreme spread to outliers.

    Calculate the interquartile range for the data = Q3 - Q1

    IQR = 2640 - 2635 = 5

    Next, multiply by 1.5
    IQR X 1.5 = 7.5

    Subtract 1.5 from Q1.
    Q1 - 1.5 = 2633.5 any number less than this is suspect to be an outlier i.e. 2595

    Add 1.5 to Q3.
    Q3 + 1.5 = 2641.5 any number greater than this is suspect to be an outlier i.e. 2642



    It is important to remember that this only is a rule of thumb and generally holds . Any potential outlier obtained by this method should be examined in the context of the entire set of data.

    Thanks for walking us through this F-15, err, I mean strikeeagle1
     
    If it is your first shot that yields the discrepancy, I'd entertain discarding it, even a magneto can biff the first rd. But, here, I would up my sample size, to say 15 or 20rds. Of coarse you can't just hammer 20 quick rds, get a rhythm, and go. Then evaluate.
    Also, you need to shoulder your gun the same each firing, if you fire a chrono rd free recoil, your velocity is dropping.

    Appreciate all the feedback and comments; I think I need to go back to the basics on my technique and make sure I'm not relying on the rear bag to absorb the recoil as opposed to my shoulder. That alone might very well explain the discrepancy in the numbers. as I'm fairly confident in my reloading routine. Thanks again!
     
    • Like
    Reactions: Milo 2.5
    I shoot with my magneto on all the time. My cold bore shot is 100fps slower than the rest of my shots. I never discard outliers nor do I accept "flyers" or pulled shots. Its sucks to accept a shot that blows ES out but it happens all the time. I'm sure the guys loading on the fx 120i to a single kernel can get rid of most of this however I did see a guy on youtube who loads to a single kernel on a satorious does all the meticulous brass prep etc and he still had an ES of 19 over 10 shots!!!!