Login | Register

Welcome, Guest. Please login or register.

July 12, 2020, 11:05:19 pm

Author Topic: Percentage Error vs Percentage Uncertainty  (Read 270 times)

0 Members and 1 Guest are viewing this topic.

Rakuu

  • Adventurer
  • *
  • Posts: 12
  • Respect: +1
Percentage Error vs Percentage Uncertainty
« on: May 30, 2020, 09:15:56 pm »
+2
Hey guys,

Just wanna double check what exactly is the difference between the two since I always get them mixed up. I've done all the calculations for my data but struggling a bit to discuss them.
This is how I see them
  • %error = how wrong my data is - more about accuracy?
  • % uncertainty= how confident I am with my data? - more about precision? low value of this = better

According from my data (if it helps):
  • Percentage uncertainty ranges from 0 to 6% (0 to 2% most of the time and just one 6%)
  • Percentage error ranges from 5 to 40% (half of them being over 20%)

Low percentage error + Low percentage uncertainty = means it's good right?
However, one of my results is 0% percentage uncertainty but also the highest percentage error (40%) => Can I conclude that my data isn't that reliable even though it's showing the relationship I wanted in the graph?

Thank you.  :)



Bri MT

  • VIC MVP - 2018
  • Administrator
  • ATAR Notes Legend
  • *****
  • Posts: 3762
  • invest in wellbeing so it can invest in you
  • Respect: +2786
Re: Percentage Error vs Percentage Uncertainty
« Reply #1 on: May 31, 2020, 09:13:23 am »
+1
Hey,

You're on the right track here with the difference between uncertainty and error as used in QCE physics.

Yep, you want high precision and high accuracy - this is often not the case for IA2 data so try not to stress about this to much!

E.g. in year 12 I did a projectile motion experiment using light gates to measure initial velocity. iirc I had 4 decimal places on my m/s data and when I plotted this I had tiny bars for my uncertainty but the trend was not reflective of what we know about projectile motion. My results had high precision but low accuracy.

What you want to do is describe this and explain the results you got. Greater error than uncertainty means that you can't simply attribute this variation in your data from the expected relationship to measurement. Consider extraneous variables that may have impacted the results and look to your observations to see if you can use this to explain outliers.

Remember that reliability is about getting the same results again - I don't think you should conclude that based on one data point.

How are you getting 0% uncertainty, is this rounding?

Hope this helps!

Rakuu

  • Adventurer
  • *
  • Posts: 12
  • Respect: +1
Re: Percentage Error vs Percentage Uncertainty
« Reply #2 on: May 31, 2020, 12:19:21 pm »
+1
Hey,

You're on the right track here with the difference between uncertainty and error as used in QCE physics.

Yep, you want high precision and high accuracy - this is often not the case for IA2 data so try not to stress about this to much!

E.g. in year 12 I did a projectile motion experiment using light gates to measure initial velocity. iirc I had 4 decimal places on my m/s data and when I plotted this I had tiny bars for my uncertainty but the trend was not reflective of what we know about projectile motion. My results had high precision but low accuracy.

What you want to do is describe this and explain the results you got. Greater error than uncertainty means that you can't simply attribute this variation in your data from the expected relationship to measurement. Consider extraneous variables that may have impacted the results and look to your observations to see if you can use this to explain outliers.

Remember that reliability is about getting the same results again - I don't think you should conclude that based on one data point.

How are you getting 0% uncertainty, is this rounding?

Hope this helps!

Thanks so much for replying! One  of my results have 0% uncertainty since we were measuring the mass (3 trials) and they happen to happen to be exactly the same (in which I highly think there were probably a few errors happening although it only seems to apply on that specific result).

My graph on the other hand is showing the trend line I was hoping for but other information for my graph includes:

  • my maximum and minimum equation have a very little difference 
  •   linear trend line equation is higher than the maximum

Will it be helpful to mention any of these or it really wouldn’t make a difference?

Bri MT

  • VIC MVP - 2018
  • Administrator
  • ATAR Notes Legend
  • *****
  • Posts: 3762
  • invest in wellbeing so it can invest in you
  • Respect: +2786
Re: Percentage Error vs Percentage Uncertainty
« Reply #3 on: May 31, 2020, 02:19:49 pm »
+1
Thanks so much for replying! One  of my results have 0% uncertainty since we were measuring the mass (3 trials) and they happen to happen to be exactly the same (in which I highly think there were probably a few errors happening although it only seems to apply on that specific result).

My graph on the other hand is showing the trend line I was hoping for but other information for my graph includes:

  • my maximum and minimum equation have a very little difference 
  •   linear trend line equation is higher than the maximum

Will it be helpful to mention any of these or it really wouldn’t make a difference?

No worries!

That seems like a strange way to get an uncertainty value... if it's what your teacher told you to do I guess stick with it but generally you would use the uncertainty reported by the scales' manufacturer or implied by the scales.

I would mention those :)