Argument with my brother and his PhD math buddy.

[DougMH]

My brother sent me an email trying to prove that SD is irrelevant. It sort of made sense and he said he'd run it by his PhD math buddy at Naval Postgraduate School who agreed.

So, I was thinking, if that's so, why do almost all Chronys display both SD and ES (as well as MV).

So, I did a bit of Excel work myself and come up with the attached pic.

I realize an ES of 10 if you're shooting rounds at 2670 is fantastic and my graphic example at the bottom of the spreadsheet numbers could be a bit extreme, but it does make a valid point I think.

In the A column, there an ES of 10, but all the rounds fired, 10 and 10 are at the extremes of MV resulting in a SD over 5. In the B column, the MVs are more evenly distributed across the ES of MV resulting in an SD under 3.

It seem to me the shot patterns are accurately depicted.

 

[Erik Cortina]

I shoot F-Class. An ES of 10 will not throw you outside the 10 ring.
I agree that SD is irrelevant because if you have a shot string such as 2670, 2670, 2670, 2670, 2670, 2670, 2670, 2670, 2670, 2700, your SD will be low, but you will still lose a point due to high ES.

 

[Lapua40X]

If the mean (sum of the population divided by the number in the population) is 725 fps, and the standard deviation is calculated at 3, then the range of velocities might be in the neighborhood ot 720 - 730 fps. The SD is helpful here because consistency in velocity is what we seek in load development, with a dreamworld SD of "0" - where the population reflects a collection of identical velocity readings from the chronograph. So the SD is helpful for us in that it helps us identify velocity consistency to fight against those vertical prints on the target.

 

[Lapua40X]

I shoot F-Class. An ES of 10 will not throw you outside the 10 ring.
I agree that SD is irrelevant because if you have a shot string such as 2670, 2670, 2670, 2670, 2670, 2670, 2670, 2670, 2670, 2700, your SD will be low, but you will still lose a point due to high ES.

I agree that an ES of 10 may not throw you outside the 10 ring, but a statistical population with an ES of 10 offers a SD of 3. Whereas your example of an ES of 30 will register an SD of 9, based on the 2700 fps outlier in the population. IMO, and SD of 9 would not be satisfactory.

 

[Turbulent Turtle]

One needs to understand the concept of SD and its application here in order to see why it's "meaningless." Now, I haven't done statistical analyses in a long time, so my recollection may be faulty in some things so bear with me.

Standard deviation is calculated in one of two ways depending on its purpose. If you are doing an analysis of a finite population, you use the number of entries found to calculate the dispersion from the mean value, using the square root of the variance, (divide by n). If you are doing a sample of a larger population, you would be using n-1. I am not sure which formula is used by the Chrony, I would need to check.

Either way, one standard deviation from the mean encompasses about 68% of the population. Two standard deviations from the mean represent 95% of the population and three standard deviations represent about 99.7% of the population.

How does this apply to us? Unless we are recording every shot all the time, we should be using the n-1 method because we are in fact sampling the ammunition. So in a box of 100 rounds, with and SD of 5 and a mean of 2700, you can expect to have 68 rounds that will be between 2695 and 2705. You can expect to have 95 rounds between 2690 and 2710. You can expect to have 99 rounds to be between 2685 and 2715. Does that mean that you necessarily have one bad round (over 3 SDs)? Not at all, but statistically speaking the chance is there and the fun thing is you don't know if or when it will occur. The fun thing is you never know if there is one "bad one" and which one it is, but if one shows up, the score will be impacted.

I realize lots of people obsess over the SD values; I happen to be one who doesn't care or even know what my ammo's SD and ES values are. I just know what my ammo does on the target after I assemble it the very best way I know how.

 

[Turbulent Turtle]

I found the SD formula used by the Chrony at their website. It's the total population analysis one; dividing my n. For predicting what's in you box as I explained above, the formula should use n-1. If you want to try to predict based on the sample, you should transfer your observations to Excel and use their formula STDEV.S not STDEV.P.

http://shootingchrony.com/manual_ASC&AMC.htm#formula

 

[GSPV]

SD is information. It's valuable in some situations, not in others.

For example, if you had an extreme spread of 70 FPS and knew that you had a very low SD, it might clue you that there is something wrong with that one round.

 

[XTR]

The one thing that shooters never take into acct is whether or not the tested population is large enough to be statistically valid, and in most cases it is not. As Denys refers to, what we really want in our world is that three standard deviations, or at least two, falls inside the 10 ring at 1000 yards. Look at enough sample data and you will find that outlier, it happens.

Now take your population and your results and run a Student t-test on it and see if your chrony SD answer is worth the pixels it takes to type it. Take your data and run a Wilcoxon Rank Sum or a Sign test and see if you have enough data points. My guess is that unless you are looking at a minimum of 14 to 20 results you've got nada.

I'm with Denys here, I just use the Chrono to know about what my MV is, I decide if the load is good shooting it at 300 and 600 yards and looking at group sizes (usually 10 shot groups)

 

[Lapua40X]

... I just use the Chrono to know about what my MV is, I decide if the load is good shooting it at 300 and 600 yards and looking at group sizes (usually 10 shot groups)

Well said. The Chrono should never be viewed as the only tool in the tool box. When the thermometer says it's 30 degrees and we prepare for 30 degrees without considering the wind chill we're gonna be cold.

 

[mysticplayer]

That was a great explanation by bayou shooter. Thanks for making it easy to understand.

I started shooting over my chronie years back until I discovered that the error in the clock was larger then the numbers I was trying to calculate. So all the math really didn't mean anything.

I then used it only to get a ball park velocity and let the target tell me what was going on. I test at 250yds or 300m and it was very easy to see how the load changed with variations in what I did. Then confirmed as far as 700 yds. If things were still solid, then the rest of the journey to 1000yd would be fine.

I now have a magnetospeed based on claims of more consistent readings and smaller device error. I have no way to confirm either way. Recent testing of new load receipe showed very consistent velocities which agreed with nice small flat groups at 250yds. The velocity was what I was expecting so I will leave it at that.

Jerry

 

[BoydAllen]

Typical samples are too small, and targets show ES. End of story