"There are three kinds of lies: lies, damn lies, and statistics." ???
I have 2 guys who use GPS (Carlson BRX7 with SurvPC and Trimble R8 with Access).?ÿ They are laymen or "button pushers" if you will. I want them to understand the *** PRACTICAL USE *** of the numbers they see when looking at their GPS shots. Multiple sessions, amount of time per session, separation of sessions, repeatability, adjustment, removal of blunders, etc. is another discussion.
The two numbers I want them to understand (*** FOR PRACTICAL USE ***) is RMS (hrms, vrms) and standard deviation. These are what they see on their data collectors.
Standard deviation is easy: (1) show/explain a normal distribution - no numbers, no statistics, just a graphical bar plot; (2) show average/mean/middle of normal distribution plot; (3) show +/- 34.1% of the collected solutions from the average/mean/middle.?ÿ Discussion of the significance when 68.2% of your 120 (or whatever) observations are within 0.01', 0.02'...0.1', 0.2', etc.?ÿ They got it - one guy asked what 2 standard deviations meant (95%)...the other guy is not as smart & didn't care (mom dropped him on his head, and we claim he is adopted).
Where I fail is explaining RMS.?ÿ I can't seem to do so without wanting to get into some math, AND THIS IS WHERE I FAIL.?ÿ I need a layman's explanation and application.?ÿ I want a better explanation than "smaller is better" or "don't go above this number". Maybe this is all my explanation should be? Maybe I should just say forget RMS and key on standard deviation? I want a 5-10 minute conversation - maybe a single page printout of some sort.
Thoughts or ideas??ÿ What would you do? Remember, this needs to be for layman.
?ÿ
RMS is a specific kind of average, and is essentially the standard deviation. Under reasonable conditions RMS or std dev tell you how far, "on average", your measurements are failing to perfectly agree with each other. It measures the spread of the darts in the target.
Maybe use an illustration with a few dots and ask where the center is. Then one with more dots approximating a 2-d normal distribution and note you have a much better idea where the center is.
It measures the spread of the darts in the target.
Good
I wouldn't get into the math much at all.
Explain that the receiver is continually calculating a likely (predicted) position based upon all the previous positions (it's looking at all the data it has already gathered), and the RMS is the difference between what it is expecting versus what it is actually measuring at that exact moment.
So that is why it is changing all the time, and why if it remains consistent while you are watching the screen, it is more likely to be an accurate value.
If it is bouncing around a lot, the RTK engine is having trouble predicting the next position, which means the RTK solution is likely less accurate.
(Leaving out the obvious possibility that for older fix/float receivers there is a greater possibility of a bad fix, even with "good" RMS values...)
The darts on the dart board is how I explain accuracy and precision - maybe I can expand the discussion.
The more I think about it, I really do think I should simply tell brothers that RMS is just another statistical function that indicates the expected amount of error when considering the collection of observations (coordinates) for a point and that smaller is better. They should focus on standard deviation because those are numbers they can more easily understand.
Maybe--another measure for how different their measured coordinates differ from the 'ideal' ones.
I always draw a bell curve and mark on 1sd, 2sd and 3sd (68/95/99) which makes it visually obvious.
And for the real simpletons just tell them to double the RMS number on the data collector (95% CI) as the more realistic expectation of quality you require.
A little hands on exercise. That helps get the mental to actual and uses a few different senses. I was taught this before i even knew what standard deviation or any other math as a surveyor. Was a rainy day and the old surveyor had me come into his office. I was studying at a tech school and such. Handed me the plumb bob and had a piece of paper taped to the floor. He made a cross and had me steady the bob but up high enough as i tapped it to the cross I would not always land perfectly on it and we did this about 20 or so times. Been a long time ago. Then he drew the different diameter circles starting small getting larger maybe like 3 anyway at the time I truly didn’t know what he was getting at except I understood that while doing this in the field I needed to be careful as to not be off in our measurements. But that was in the 90’s and i have never forgotten it and today that picture enters my brain as I watch the H and V precisions bounce while taken readings in canopy with rtk. For me if i can put my hands and see touch feel it sticks with me. Those uncertainty errors are something we need to understand for sure. Helps so much in the decision making process. A inverse between two points is linear and angular. The delta north delta east elev. now starts that mental image of the error ellipse.