denton
Member
If you're striving for consistency in your reloading, you might be concerned about standard deviation since that is a measure of variation. The following ideas might make your work more productive:
1. Standard deviation is a measure of dispersion, or lack of consistency. The bigger the standard deviation, the more dispersed, or variable, your data are.
2. It's harder to get a precise standard deviation than it is to get a precise mean (average). With about 3 dozen samples, you've usually got a good estimate of the mean. It takes a lot more to get a precise estimate of standard deviation. The standard deviation of muzzle velocity from a group of 5 shots is very imprecise.
3. What shooters call "extreme spread", statisticians call "range". Range can be converted to standard deviation if you know sample size. For 5 shots, divide range by 2.3 to get standard deviation. The number you get by this method will usually not exactly match the conventional SD calculation, because both numbers are imprecise estimates, but one is about as good as the other.
4. Random variation does not add like most people think it does. If there are two or more sources of variation, and if one is much larger than the rest, the large one will almost totally determine the total variation. This has some important consequences for shooters. One is that you can fruitlessly fiddle forever with the lesser sources of variation. In order to improve your process, you must find the large contributors to variation, and fix those first. If you are shooting 6" groups offhand with a 1/4" bench rest rifle, and switch to a rifle that does 2" groups off a bench rest, your groups will NOT grow by 1 3/4", because the largest source of variation is your shooting skill, not the firearm.
5. Standard deviation is defined for all real-world distributions, not just the normal distribution. On the other hand, no real-world collection of data was ever normally distributed. But many are close enough to be useful.
6. The rule that 68% of data falls within plus and minus 1 standard deviation of the mean, 95% within two, and 99.7% within three is correct for the normal distribution. It is also a good approximation even if the data are not normally distributed. Normality is overrated.
7. Commercial ammunition I've tested has about a 35 FPS SD in muzzle velocity. (SAAMI assumes 2% of MV.) It's not hard to get handload SD down into the low 20s, and with a little effort down into the single digits. But in my reliably 5/8" 5-shot group 223, it doesn't seem to make much, if any, difference.
1. Standard deviation is a measure of dispersion, or lack of consistency. The bigger the standard deviation, the more dispersed, or variable, your data are.
2. It's harder to get a precise standard deviation than it is to get a precise mean (average). With about 3 dozen samples, you've usually got a good estimate of the mean. It takes a lot more to get a precise estimate of standard deviation. The standard deviation of muzzle velocity from a group of 5 shots is very imprecise.
3. What shooters call "extreme spread", statisticians call "range". Range can be converted to standard deviation if you know sample size. For 5 shots, divide range by 2.3 to get standard deviation. The number you get by this method will usually not exactly match the conventional SD calculation, because both numbers are imprecise estimates, but one is about as good as the other.
4. Random variation does not add like most people think it does. If there are two or more sources of variation, and if one is much larger than the rest, the large one will almost totally determine the total variation. This has some important consequences for shooters. One is that you can fruitlessly fiddle forever with the lesser sources of variation. In order to improve your process, you must find the large contributors to variation, and fix those first. If you are shooting 6" groups offhand with a 1/4" bench rest rifle, and switch to a rifle that does 2" groups off a bench rest, your groups will NOT grow by 1 3/4", because the largest source of variation is your shooting skill, not the firearm.
5. Standard deviation is defined for all real-world distributions, not just the normal distribution. On the other hand, no real-world collection of data was ever normally distributed. But many are close enough to be useful.
6. The rule that 68% of data falls within plus and minus 1 standard deviation of the mean, 95% within two, and 99.7% within three is correct for the normal distribution. It is also a good approximation even if the data are not normally distributed. Normality is overrated.
7. Commercial ammunition I've tested has about a 35 FPS SD in muzzle velocity. (SAAMI assumes 2% of MV.) It's not hard to get handload SD down into the low 20s, and with a little effort down into the single digits. But in my reliably 5/8" 5-shot group 223, it doesn't seem to make much, if any, difference.
