How do I calculate error?

Ask questions about projects relating to: computer science or pure mathematics (such as probability, statistics, geometry, etc...).
Locked
oiziz
Posts: 6
Joined: Sat Dec 02, 2006 1:47 am

How do I calculate error?

Post by oiziz »

I have done the same experiment 5 times and they all came out with data that was slightly different but still reproducible. I want to calculate error, but I'm not sure how. Does anyone know?

I find a coefficient from the data and result in 5 points. Do I find the average then divide that by the standard deviation to get the percent error?

Also how do I add errors? I average those 5 points to get one point, but I use that single point for another graph and try to do a linear fit. How do I add the error of the linear fit to the average error?

I am a bit confused.
hiramuzammal
Former Expert
Posts: 44
Joined: Sun Oct 16, 2005 2:45 pm

Post by hiramuzammal »

This site explains error analysis in a simple way, although it doesn't have many examples: http://www.astro.virginia.edu/class/maj ... error.html

This site has more examples:
http://www.lhup.edu/~dsimanek/scenario/ ... asures.htm

The first two sites are rather technical, but this site is for younger students: http://www.lhup.edu/~dsimanek/errors.htm
tdaly
Former Expert
Posts: 1415
Joined: Sat Nov 08, 2003 11:27 pm
Occupation: Planetary Scientist
Project Question: N/A
Project Due Date: N/A
Project Status: Not applicable

Post by tdaly »

oiziz,

I'm a bit confused as to what you are trying to do. Correct me if I'm wrong: you did an experiment 5 times and are now analyzing the data from your experiment.

I don't understand what you mean by "I find a coefficient from the data." Did you do a linear regression analysis (e.g. least-squares, etc.) and so by coefficient you mean slope? Or did you do some other kind of analysis.

How do you know you have "error" in your data? Experiments always produce slightly different results because of chance variation. When you do a percent error calculation, you are comparing a predicted value (like from a mathematical model) with an empirical value. It doesn't make sense to find the error in a data set if you don't have anything to compare it to.

Try to answer some of these questions and provide as much information as you can. I will work with you to analyze your data, but I need to know what kind of data you have and what you are trying to do with it. It would be helpful if you told me what your question is, your hypothesis, and explained your procedure.

If you can, try posting your raw data here. Then I can have a copy and you can have a copy to help keep us on the same page.

Looking forward,
All the best,
Terik
davidkallman
Former Expert
Posts: 675
Joined: Thu Feb 03, 2005 3:38 pm

Re: How do I calculate error?

Post by davidkallman »

Hi oiziz!

Once you have calculated error, see previous posts, the question becomes how to display the error limits. One way is to use error bars.

If you input ''how do I display error bars?" to answers.com, you get back ten hits to built-in answers and many web hits.

Among the web hits is:

Science Fair - How to Create a Winning Display:
http://www.scifair.org/howtocreateawinningdisplay.html
(Note: The section on "Error Bars" can be found at the bottom of the page)

No doubt, this page is of general interest beyond this topic!

So plenty to research, hopefully your answer is in there.
Cheers!

Dave
Locked

Return to “Math & Computer Science Sponsored by Hyperion Solutions Corp”