Help Support Long Range Hunting Forum
Yep, used it for the first time last Monday. Easy to setup less than a minute as opposed to about 45-60 minutes for my old Caldwell, and it did not miss any shots or record any other shots of shooters at the range, just captured my shots. I liked it a lot. I am glad I waited and did not purchase the Lab Radar I was holding off from buying, although Lab Radar has now come out with a compact version. I am happy with what I got.
IHFarmer actually reported the opposite: that Labradar calculates as a sample from a large population, and Garmin calculates the entire small population. Which is it? I rewatched the video, and what I describe is what the guy in the video stated.
Velocity is always a guesstimate if you're shooting at distance, that's why you have to true your ballistics calculator.
I didn't find IHFarmer's post, but in another thread, we've discussed this Garmin SD issue extensively. Garmin calculates (an optimistic) SD for small sample sizes. Less than 20. For 3 shots, the SD from Garmin will always be too optimistic. For 50 shots, it will be correct. As an example, let's say your velocities are 2795, 2800 and 2805 fps. Excel's STDEVP (for a [large] population) would calculate SD to be about 4.1--this is what the Garmin we used always calculated. We'd put the raw velocities for three or five shot strings into Excel and we'd get an overly optimistic SD. What Garmin SHOULD do for small groups is use a function like Excel's STDEV. This is for small group sizes, and produces a SD of 5. When we did this test two months ago, we use my LabRadar and my friend's Garmin at the same time, and for both of our rifles. For both of our rifles and 10 groups of 3 shots each, we got similar velocities between devices, but his Garmin SD's were always better than my LabRadar's SD's. We took all of the velocities, ran them in Excel, and the Garmin used the wrong method--they use SDEVP. LabRadar and Excel do the calculations correctly, and agree for small populations. I've contacted Garmin twice about this, and never heard a word back.
only one meaning for "η ταν η επί τας" It is written in the original language, for a reason. At least it is the only one I was taught in school. It is what the Spartan mother would say to the son about to go to war when she gave him his shield. Carry it back victoriously, or be carried back dead on it.
Yep. You're spot on. By using a "sample", you're admitting that you don't have all the data from a normal population, and your SD should be worse. Some almost 50 years now since my statistics class in college, I still have occasional dreams that I'm going to my statistics final exam, but forgot to go to class or study all quarter. Never had those bad dreams with engineering classes....
I keep seeing the term "wrong" thrown around when it is pointed out that Garmin uses Population SD. Population SD and Sample SD are 2 different statistical metrics, either or both of which can be applied, depending on what is being examined and what is trying to be learned. Calling one of them wrong is not accurate; they are different.
I beg to differ. Garmin calculates it WRONG. First, we're assuming a normal (bell-shaped) distribution. Not a bimodal, triangular, square top distribution, etc. Most of us are shooting smaller samples of data, well under 30, and more likely 5 to 10 shots looking for an average velocity, ES and SD. If one uses the sample SD calculations for smaller samples, (Excel: STDEV) the denominator is sqrt(N-1). For a population, that denominator is sqrt(N). This has a huge effect for the shooter doing small groups. What is worth noting is that as the sample size gets larger and larger towards the "population", the results from STDEV and STDEVP (sample and population) converge with minimal error. Since Garmin doesn't state their method, but uses a population calculation, this is inherently the wrong way to do it. The correct way to do it is to use a Sample SD and let it converge for larger sample sizes or full populations of data. The difference for a set of 30 data points is typically about 0.1 difference in SD between the two methods.
