The Acoustic Guitar Forum

Go Back   The Acoustic Guitar Forum > General Acoustic Guitar and Amplification Discussion > General Acoustic Guitar Discussion

Reply
 
Thread Tools
  #31  
Old 01-17-2014, 09:45 AM
Nailpicker Nailpicker is offline
Registered User
 
Join Date: Dec 2013
Posts: 1,522
Default

I agree. It's already been a long winter here.
Reply With Quote
  #32  
Old 01-17-2014, 09:49 AM
Diamond Dave Diamond Dave is offline
Registered User
 
Join Date: May 2012
Location: Blue Ridge Mountains, VA
Posts: 3,034
Default

Quote:
Originally Posted by picassov7 View Post
Precision, on the other hand, is repeatability.
No, that's accuracy.

These are pretty basic scientific concepts that anybody who took a high school chemistry lab should have been taught.

Accuracy means a bathroom scale is always going to say 10 lbs. when a 10 lb. dumbbell is placed on it.

Precision means the scale is going to say 10.45, 9.98, 10.23, 9.62 and 10.01 lbs. if you weigh the dumbbell 5 times. These are more precise answers (down to 1/100 of a lb.), but not necessarily very accurate, since the readings are nearly 1/2 lb. off in one case.

If the scale says 10.00 lbs. every time, it's precise and accurate.

If the scale says 15, 5, 16, 12 and 4, it's neither precise nor accurate.
__________________
"You can get help from teachers, but you are going to have to learn a lot by yourself, sitting alone in a room."
--Dr. Seuss
Reply With Quote
  #33  
Old 01-17-2014, 09:51 AM
Diamond Dave Diamond Dave is offline
Registered User
 
Join Date: May 2012
Location: Blue Ridge Mountains, VA
Posts: 3,034
Default

Let's throw in some quantum mechanics: the very action of measuring something changes the measurement!
__________________
"You can get help from teachers, but you are going to have to learn a lot by yourself, sitting alone in a room."
--Dr. Seuss
Reply With Quote
  #34  
Old 01-17-2014, 10:00 AM
devellis's Avatar
devellis devellis is offline
Charter Member
 
Join Date: Nov 2007
Location: North Carolina
Posts: 8,399
Default

Quote:
Originally Posted by Diamond Dave View Post
No, that's accuracy.

These are pretty basic scientific concepts that anybody who took a high school chemistry lab should have been taught.

Accuracy means a bathroom scale is always going to say 10 lbs. when a 10 lb. dumbbell is placed on it.

Precision means the scale is going to say 10.45, 9.98, 10.23, 9.62 and 10.01 lbs. if you weigh the dumbbell 5 times. These are more precise answers (down to 1/100 of a lb.), but not necessarily very accurate, since the readings are nearly 1/2 lb. off in one case.

If the scale says 10.00 lbs. every time, it's precise and accurate.

If the scale says 15, 5, 16, 12 and 4, it's neither precise nor accurate.

Part of the problem we're encountering seems to be that different fields use terms differently. I find "consistency" and "bias" to be the less ambiguous terms because "precision" and "accuracy" have informal meanings that don't match their more technical meanings. Actually, I find reliability and validity to be even better, but that's because they're the terms used in my field.

Frankly, I wouldn't trust much that my high school chemistry teacher said, but your mileage may vary.
__________________
Bob DeVellis
Reply With Quote
  #35  
Old 01-17-2014, 10:15 AM
philjs philjs is offline
Registered User
 
Join Date: Sep 2009
Location: Halifax, NS, Canada
Posts: 1,970
Default

Quote:
Originally Posted by HHP View Post
The two diagrams on the left are both neither accurate or precise. They are random and unpredictable.
The diagram mc1 found is better than the one I found (since it has more dots, ie. the sample size is larger) but the above statement is wrong.

Statistically speaking, the mean distances of all of the dots/shots in the two left diagrams, from their cluster center, are very similar (that is the standard deviations of the shots/dots around each cluster center are very similar) BUT the mean distances of all of the dots/shots from the center of the target are not.

The mean center of the cluster of dots/shots in the accurate-labeled diagram IS the target (but the standard deviation is sufficiently large that they are not precise). If the target is the object center then each of the dots/shots in the not accurate-labeled diagram ALL have very large mean distances (though the standard deviation of those dots/shots from each other may not differ from the accurate-labeled diagram).

The clustering of the shots/dots, regardless of the accuracy with regards to the center of the target, show that they are definitely NOT random.

I'm enjoying the discussion, folks!

Phil
__________________
Solo Fingerstyle CDs:
Two Steps Forward, One Step Back (2021)
One Size Does Not Fit All (2018)

I play Crosby, Emerald, Larrivée, Lowden, Rainsong & Tacoma guitars.
Check out my Guitar Website. See guitar photos & info at my Guitars page.
Reply With Quote
  #36  
Old 01-17-2014, 10:16 AM
drplayer's Avatar
drplayer drplayer is offline
Charter Member
 
Join Date: Jul 2011
Posts: 3,292
Default

Well, I generally tune to my Snark, and if it sounds a bit off to me, then I tweak it (which I almost never have to do). Then I just play! I've got enough to concentrate on and worry about just trying to become a better player. Worrying about "how" accurate or "how" precise my tuner "really" is isn't even on my radar...
__________________

Martin 000-28EC
'71 Harmony Buck Owens American
Epiphone Inspired by Gibson J-45
Gold Tone PBR-D Paul Beard Signature Model resonator

"Lean your body forward slightly to support the guitar against your chest, for the poetry of the music should resound in your heart."
-Andrés Segovia
Reply With Quote
  #37  
Old 01-17-2014, 10:18 AM
Diamond Dave Diamond Dave is offline
Registered User
 
Join Date: May 2012
Location: Blue Ridge Mountains, VA
Posts: 3,034
Default

Quote:
Originally Posted by devellis View Post
Frankly, I wouldn't trust much that my high school chemistry teacher said, but your mileage may vary.
Come to think of it, Bob, I kind of feel the same way about my high school chemistry teacher, too. She had a nervous tick that made her constantly say, "Okay, now." We'd keep track with hash marks on our notebooks and she'd say it 30 or 40 times in the course of a 50-minute class.

I received a better explanation in college in an aerial photo interpretation course. The instructor had worked for the military doing intelligence work on surveillance photos taken by U-2s and the like. We learned about accuracy, precision, degrees of confidence, etc.

http://en.wikipedia.org/wiki/Words_o...ve_probability
__________________
"You can get help from teachers, but you are going to have to learn a lot by yourself, sitting alone in a room."
--Dr. Seuss
Reply With Quote
  #38  
Old 01-17-2014, 10:27 AM
HHP HHP is offline
Registered User
 
Join Date: Aug 2007
Location: Indianapolis, IN
Posts: 29,351
Default

Quote:
Originally Posted by philjs View Post
The diagram mc1 found is better than the one I found (since it has more dots, ie. the sample size is larger) but the above statement is wrong.

Statistically speaking, the mean distances of all of the dots/shots in the two left diagrams, from their cluster center, are very similar (that is the standard deviations of the shots/dots around each cluster center are very similar) BUT the mean distances of all of the dots/shots from the center of the target are not.

The mean center of the cluster of dots/shots in the accurate-labeled diagram IS the target (but the standard deviation is sufficiently large that they are not precise). If the target is the object center then each of the dots/shots in the not accurate-labeled diagram ALL have very large mean distances (though the standard deviation of those dots/shots from each other may not differ from the accurate-labeled diagram).

The clustering of the shots/dots, regardless of the accuracy with regards to the center of the target, show that they are definitely NOT random.

I'm enjoying the discussion, folks!

Phil
Using the rifle/target analogy, a rifle that produces a 6" group centered around a bullseye is considered far less accurate and precise than a rifle that produces a 1" group that is 7" off center. In the first, you have no idea where each shot lands and therefore cannot predict where the next shot will go outside of a 6" radius. At 100yds, you would say that rifle is accurate to 6 minutes of angle.

The second example would produce 1 minute of angle accuracy and you can predictably know where the next shot will land so you can adjust/calibrate reliably.
Reply With Quote
  #39  
Old 01-17-2014, 11:43 AM
devellis's Avatar
devellis devellis is offline
Charter Member
 
Join Date: Nov 2007
Location: North Carolina
Posts: 8,399
Default

Quote:
Originally Posted by HHP View Post
Using the rifle/target analogy, a rifle that produces a 6" group centered around a bullseye is considered far less accurate and precise than a rifle that produces a 1" group that is 7" off center. In the first, you have no idea where each shot lands and therefore cannot predict where the next shot will go outside of a 6" radius. At 100yds, you would say that rifle is accurate to 6 minutes of angle.

The second example would produce 1 minute of angle accuracy and you can predictably know where the next shot will land so you can adjust/calibrate reliably.
This is a good example of the "different usages" point I made earlier. What you say makes perfect sense but in the context of measurement theory, the terms are used differently. The word "accurate" really isn't the best descriptor for the situation where the mean of the distribution coincides with the true score (the center of the target in this analogy) because it's so counterintuitive that widely dispersed outcomes could have anything to do with accuracy. That's why I prefer "lack of bias" or "validity" as descriptors (although that gets complicated, too, because validity is limited by reliability).

An example that might work is shooting at a target with a blunderbuss. If the sights are properly aligned, the outcomes will still be widely dispersed because the gun just isn't very consistent. But if the sights are misaligned (does a blunderbuss even have sights?), the results will be both widely dispersed and off-center. The way the terms were used originally, the former would be described as accurate but imprecise (that is, centered around the target but scattered) while the latter would be neither accurate nor precise (scattered and off-center). Now, in normal parlance, it seems like a stretch to call the first instance "accurate" and probably most of us wouldn't use that word in that situation. It's a poor word to describe the extent to which the central tendency of the observations (irrespective of scatter) corresponds to the "target," but it is used to mean that in some situations. Again, I prefer to describe that as "lack of bias" (i.e., a lack of a tendency to be off systematically, irrespective of any random error). It's really just a case of a specialized usage and a general usage of a word conflicting. Kind of like the misunderstanding that might occur when we tell a non-guitarist that we're going to the store to get a new g-string.
__________________
Bob DeVellis
Reply With Quote
  #40  
Old 01-17-2014, 11:51 AM
Diamond Dave Diamond Dave is offline
Registered User
 
Join Date: May 2012
Location: Blue Ridge Mountains, VA
Posts: 3,034
Default

Quote:
Originally Posted by devellis View Post
a new g-string.
A new g-string, two nuts, a strap and a saddle. Just another Friday night.
__________________
"You can get help from teachers, but you are going to have to learn a lot by yourself, sitting alone in a room."
--Dr. Seuss
Reply With Quote
  #41  
Old 01-17-2014, 12:23 PM
ecguitar44 ecguitar44 is offline
Registered User
 
Join Date: Nov 2012
Posts: 3,104
Default

Quote:
Originally Posted by HHP View Post
Calibration is your desired reference point. In the case of a tuner, you might use a reference tone like a tuning fork. In a rifle, it would be the target center.

When you sight a rifle, you first achieve accuracy (shots consistently impacting the same point) and then calibrate ( adjust sights to move point of impact to desired point). If you can't get the shots to hit the same point of impact (accurate), you have no way to know how to adjust or calibrate.

Like rifles have adjustments for windage and elevation, tuners have calibration screws or dials.
You have some basic misunderstandings of the term "calibration" as it refers to measurement systems and measurement theory.

Calibration is NOT your desired reference point. Calibration is the process of adjusting your system by comparing to a known value.

In this way, "calibration" is the process of adjusting your sites based on how close you are to your target. Or, turning your tuning peg so that the note matches what is being shown on the tuner.
__________________
侘 寂 -- wabi-sabi -- acceptance of transience and imperfection by finding beauty in that which is imperfect, impermanent, and incomplete
Reply With Quote
  #42  
Old 01-17-2014, 12:24 PM
ecguitar44 ecguitar44 is offline
Registered User
 
Join Date: Nov 2012
Posts: 3,104
Default

This is the most succinct way to differentiate "accuracy" and "precision" as commonly used in measurement theory:



Calibration is the process of adjusting the center of the distribution to be closer to the reference point.

Simple as pie.
__________________
侘 寂 -- wabi-sabi -- acceptance of transience and imperfection by finding beauty in that which is imperfect, impermanent, and incomplete

Last edited by ecguitar44; 01-17-2014 at 12:34 PM.
Reply With Quote
  #43  
Old 01-17-2014, 12:31 PM
HeyJoe42 HeyJoe42 is offline
Registered User
 
Join Date: Jul 2012
Location: SF Bay Area
Posts: 117
Default

It doesn't need to be so complicated...

Accuracy...deviation from the true value...accepted as bias as well.

Precision...variation in repeated measturements...accepted as spread as well.

In principle, it is much easier to take a precise instrument and adjust for accuracy. This is a basic method of process development. It is more difficult to take in imprecise instrument and adjust it to give useful, accurate measurements...unless you are willing to take an average of many readings...which is what most inexpensive (and inprecise) instruments do with varying degrees of success.

Do not confuse displayed digits or significant figures or measurement resolution...those are all aspects of the measurement instrument...but are not fundamental concepts of error analysis.

Hey gang...go back to tuning and playing your guitars!

Cheers.
__________________
2012 Taylor GA3-12
2012 Epiphone Les Paul Ukulele
2008 Cort Earth 70e
1975 Fender Stratocaster
2007 Fender Standard Stratocaster
1967 Epiphone Madrid EC-30

"Make every song you sing your favorite tune."
Reply With Quote
  #44  
Old 01-17-2014, 12:45 PM
ljguitar's Avatar
ljguitar ljguitar is offline
Charter Member
 
Join Date: Mar 2005
Location: wyoming
Posts: 42,556
Default

Quote:
Originally Posted by HeyJoe42 View Post
It doesn't need to be so complicated...Hey gang...go back to tuning and playing your guitars!
Hey Joe!

You do realize some people take some topics to extremes, right?

And the adjuring to go back to playing is completely wasted when there is a good 'discussion' (argument/debate/dispute) going on.


__________________

Baby #1.1
Baby #1.2
Baby #02
Baby #03
Baby #04
Baby #05

Larry's songs...

…Just because you've argued someone into silence doesn't mean you have convinced them…
Reply With Quote
  #45  
Old 01-17-2014, 12:49 PM
ecguitar44 ecguitar44 is offline
Registered User
 
Join Date: Nov 2012
Posts: 3,104
Default

Quote:
Originally Posted by ljguitar View Post
Hey Joe!

You do realize some people take some topics to extremes, right?

And the adjuring to go back to playing is completely wasted when there is a good 'discussion' (argument/debate/dispute) going on.


Exactly! That's the whole point of forums like this!
__________________
侘 寂 -- wabi-sabi -- acceptance of transience and imperfection by finding beauty in that which is imperfect, impermanent, and incomplete
Reply With Quote
Reply

  The Acoustic Guitar Forum > General Acoustic Guitar and Amplification Discussion > General Acoustic Guitar Discussion

Thread Tools





All times are GMT -6. The time now is 11:14 AM.


Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2024, vBulletin Solutions Inc.
Copyright ©2000 - 2022, The Acoustic Guitar Forum
vB Ad Management by =RedTyger=