A few weeks ago, I read Damon Cali's blog article demonstrating his innovative analysis method. This article contributed to my motivation to start this blog and discuss statistics applied to shooting. He fired 175 shots on paper, and visualized the 5-shot group sizes as a contour plot to determine the optimal charge and seating depth. As you can read in his article, it is an effective way to identify the smallest group and any trends. I thought it would benefit from a measure of statistical confidence. To rely on the results, you need to know you've fired enough shots. Further, I thought by analyzing each shot individually, rather than the ES of each 5-shot group, it might lead to some deeper insights.
I contacted him for the raw data and performed my own analysis, then wrote a guest blog article, which you can read over at his site, Bison Ballistics. In this article, I use many statistical tools to show how a deeper view of data can lead to unexpected discoveries and a more accurate load. This opportunity presented itself so I took a short detour from my plan to show you exactly how to apply these concepts for yourself. You can look forward to that article very soon!
1 Comment
George Crawford
10/17/2017 12:33:07 pm
I enjoyed reading your analysis. I am a retired engineer with some postgraduate experience in statistics. I have always taken note of decisions made on small sample sizes who confidence intervals are inherently large, even at low probabilities of error like 80 percent. I though his surface plot was a good idea, but I suspect if you plotted the upper and lower 50% confudence limits you would find that his countor was not particulary meaningful and that even a flat plane would fir within the limits of the data. Being retired on limited funds I am all for saving money, but by the same token if you do testing and draw conclusions that arent sound you have not been cost effective either. How do you strike the balance between initial coarse screening tests on limited shots and the inherent uncertainty associated with small samples. Seems to me like most of his surface from the initial round is pretty suspect due to small number of shots, but the later round with more shots is much more certain. Thanks, great article, good thinking on the analysis!
Reply
Leave a Reply. |
Who am I?Adam MacDonald: Canadian FTR shooter, inventor, problem solver. Archives
May 2019
|