The Student Room Group

AS Physics (OCR) Coursework - Line of worst fit

Hello,

I've been told that I have to draw a line of "worst fit" in order to find the percentage uncertainty in the gradient...I understand the process for finding the uncertainty but I don't understand how to draw the line of worst fit?? Would anyone be able to help me out here please?

Please note this isn't anything specific from the real coursework - it's just a practice :smile:

Thanks!
Original post by MollyMcFly1
Hello,

I've been told that I have to draw a line of "worst fit" in order to find the percentage uncertainty in the gradient...I understand the process for finding the uncertainty but I don't understand how to draw the line of worst fit?? Would anyone be able to help me out here please?

Please note this isn't anything specific from the real coursework - it's just a practice :smile:

Thanks!


You need error bars on the data points.
The line of worst fit should be the line with the greatest or least slope, away from your line of best fit, that still goes through all (or nearly all) the error bars.

This graph is exaggerated to illustrate the point.



The dotted lines could be lines of worst fit. One greater gradient and one less.
Reply 2
Original post by Stonebridge
You need error bars on the data points.
The line of worst fit should be the line with the greatest or least slope, away from your line of best fit, that still goes through all (or nearly all) the error bars.

This graph is exaggerated to illustrate the point.



The dotted lines could be lines of worst fit. One greater gradient and one less.


But we've been told not to use error bars? :-/
Original post by MollyMcFly1
But we've been told not to use error bars? :-/


In which case I would
a) ask why you've been asked not to use error bars. Seems very odd to say the least.

b) to answer your question - it makes it much more difficult. You just have to make an intelligent guess. Imagine the example above without the error bars! See what I mean?
Sorry, but without error bars you just have to try to draw a line you think could possibly fit the points but which is at an angle to your best line. There is no "right" answer. You just have to judge it by eye. No two graphs are the same and each one has to be judged individually. It's needs practice on a number of graphs (with teacher feedback) to get the hang of it.
The two lines (best and worst) are really your personal statement of what you consider to be the best line and one which, though not the best, could reasonably be considered to go through most of the points.

Sorry I can't be more helpful.
(edited 10 years ago)
Reply 4
Original post by Stonebridge
In which case I would
a) ask why you've been asked not to use error bars. Seems very odd to say the least.

b) to answer your question - it makes it much more difficult. You just have to make an intelligent guess. Imagine the example above without the error bars! See what I mean?
Sorry, but without error bars you just have to try to draw a line you think could possibly fit the points but which is at an angle to your best line. There is no "right" answer. You just have to judge it by eye. No two graphs are the same and each one has to be judged individually. It's needs practice on a number of graphs (with teacher feedback) to get the hang of it.
The two lines (best and worst) are really your personal statement of what you consider to be the best line and one which, though not the best, could reasonably be considered to go through most of the points.

Sorry I can't be more helpful.


Ok, thank you - I know, it's very strange...it says in the OCR guide to practical skills that error bars aren't necessary so I'm assuming they're just looking for a rough estimate? I read somewhere that if you rotate your ruler from the centre of the LoBF until the points on either side become uneven then you have your line of worst fit...
Original post by MollyMcFly1
Ok, thank you - I know, it's very strange...it says in the OCR guide to practical skills that error bars aren't necessary so I'm assuming they're just looking for a rough estimate? I read somewhere that if you rotate your ruler from the centre of the LoBF until the points on either side become uneven then you have your line of worst fit...


I'm sure that would work.
The problem is, without error bars you have no idea how far away from the plotted points can be considered "within experimental error".
Anyway. Good luck.
Reply 6
As far as I know, and from personal experience, they already give you the best and worst lines of fit in the actual thing.

Posted from TSR Mobile
Reply 7
Starting this year, they will give you both best and worst lines on the graph at As level.
Previously you just had to guess - which was unsatisfactory to say the least.
Reply 8
Original post by teachercol
Starting this year, they will give you both best and worst lines on the graph at As level.
Previously you just had to guess - which was unsatisfactory to say the least.


Aaah brilliant - thanks for letting me know! That makes a lot more sense than just having to slam a ruler down somewhere random haha!
Original post by MollyMcFly1
Ok, thank you - I know, it's very strange...it says in the OCR guide to practical skills that error bars aren't necessary so I'm assuming they're just looking for a rough estimate? I read somewhere that if you rotate your ruler from the centre of the LoBF until the points on either side become uneven then you have your line

Can you explain the method of rotating the ruler for the line of worst fit please

Quick Reply