Does anyone here have any experience with Statistical analysis?
As part of my project I take a lot of data, I generally take the mean of the first and last points of each data set and compare the difference. That tells me one thing. I know there's an error associated with taking the mean etc, and I can work that out.
My question is:
If I run the same test X times and the result varies between them, what is the best way to work out the error in this, and combine it with the other errors I have.
There are many ways of doing error analysis. I could use a guide which tells me when to use a certain method, but the google provides me with too much info. I was hoping someone has had to do something similar and can point me in the right direction.
Well, you can calculate the variance (= standard deviation squared) easily enough. First, calculate the mean of all the measurements (called "y bar" in the formula) Then subtract this value from each of your measurements (called y-i) and square what you get. Add all those together and divide by the number of measurements -1, and you get your sample variance. Take the square root of this to get the standard deviation (sigma).
(http://img24.imageshack.us/img24/1754/utennavnl.png) (http://img24.imageshack.us/my.php?image=utennavnl.png)
Around 2/3 of all the measurements lie within + or - one standard deviation from the mean.
You can then go on and find confidence intervals and loads of stuff to get a more precise expression for your error, but the standard deviation normally does the job well enough to describe it.
Cheers, If I want to then combine this result to the error I have already, do I simply add these errors together? Is there a more sophisticated way of combining errors together? I can error bars very quickly becoming rather large is everything is a sum of parts.
http://www.google.co.uk/search?q=combining+errors
Quote from: BlueBall;270533http://www.google.co.uk/search?q=combining+errors
Doh, sometimes I type way too much into google, sometimes a tutorial in googling wouldn't go a miss. Thanks. :blush2: