PDA

View Full Version : SD? Excellent article.....



Larry Gibson
11-23-2010, 06:35 PM
Many continually question what SD (Standard Deviation) is and what it means, particularly in context to velocity reading with most chronographs. Some question the validity of chronograph use because many times seemingly low SDs don't produce the best groups. For a good explanation of what "SD" really means (or in this case doesn't mean) there is an excellent article in the November issue of Rifle magazine. It is titled; Numbers.. and is written by Dave Scovill in his Spotting Scope column.

It is well worth the read.

Larry Gibson

mpmarty
11-23-2010, 07:38 PM
I generally don't pay ant attention to SD as I'm not a mathematician. What I do look for is shot to shot uniformity and less difference from high to low is always better to me.

BCB
11-23-2010, 07:42 PM
Many continually question what SD (Standard Deviation) is and what it means, particularly in context to velocity reading with most chronographs. Some question the validity of chronograph use because many times seemingly low SDs don't produce the best groups. For a good explanation of what "SD" really means (or in this case doesn't mean) there is an excellent article in the November issue of Rifle magazine. It is titled; Numbers.. and is written by Dave Scovill in his Spotting Scope column.

It is well worth the read.

Larry Gibson

Is there an Internet site where one might read the article if not a subscriber to the magazine?...

Thanks...BCB

Larry Gibson
11-24-2010, 02:04 PM
Try www.riflemagazine.com. I suspect you'd have to subscribe to get a read on the current magazine. After all, they are in the business to sell magazines. I don't subscribe to many magazines anymore as I look at them at the local Border's and if there's something worth reading I buy the magazine. I then save some of the worthwhile articles. This article on SD is a "keeper" for future reference.

Larry Gibson

prs
11-24-2010, 04:08 PM
In a small sample, SD is meaningless. Extreme spread would serve better in very small samples such as ten items. Kicking out the high and low, then averaging the remaining data would serve even better in "larger" small samples. Samples of less then one hundred data points would be quite small, we typically do not bust that many caps in load testing of a single lot. In instances where Normal Distributions and SD yield statistically confident predictions, the sample sizes (n) are often well into the thousands or higher.

prs