The Student Room Group
Uni students - Complete our survey >>
Uni students - Complete our survey >>
Uni students - Complete our survey >>

question on error bars

Example:

- you have an independent variable (i.e time) and a dependent variable (i.e temperature)

- for every set interval of the independent variable (i.e every 20 seconds), you take 3 measurements of the dependent variable, and then you calculate the mean from the 3 measurements


Do you draw error bars based exactly on the raw data for each point (i.e for the 40 second mark, you took readings of 23, 26, 28 degrees celcius, and you draw the error bars which stretch fom the 28 degree mark to the 23 degree mark)

OR

Do you draw the error bar with the mean in centre, with the length of the bar equal to the range of the 3 readings (i.e 28 - 23 = 5, so a length of 2.5 on either side of where the mean is)
Original post by fuzzybear
Example:

- you have an independent variable (i.e time) and a dependent variable (i.e temperature)

- for every set interval of the independent variable (i.e every 20 seconds), you take 3 measurements of the dependent variable, and then you calculate the mean from the 3 measurements


Do you draw error bars based exactly on the raw data for each point (i.e for the 40 second mark, you took readings of 23, 26, 28 degrees celcius, and you draw the error bars which stretch fom the 28 degree mark to the 23 degree mark)

OR

Do you draw the error bar with the mean in centre, with the length of the bar equal to the range of the 3 readings (i.e 28 - 23 = 5, so a length of 2.5 on either side of where the mean is)


A liitle more context would help here.
Temperature of what?
Measured how? Instrument?
How do you take three temperature measurements at each interval? Simultaneously? Maybe you didn't mean this.

With the lack of detail given the best answer I can give is that normally you would do the 2nd of those two options, but preferably with more than 3 readings.
Reply 2
Avatar for fuzzybear
fuzzybear
OP
12
Original post by Stonebridge
A liitle more context would help here.
Temperature of what?
Measured how? Instrument?
How do you take three temperature measurements at each interval? Simultaneously? Maybe you didn't mean this.

With the lack of detail given the best answer I can give is that normally you would do the 2nd of those two options, but preferably with more than 3 readings.


oh sorry for the confusion, I was referring to repeat readings of a dependent variable, from which we find the average (mean) value which is whats plotted on the graph. This is where I get stuck, how would the error bars be drawn?

would it just be half the range of the repeat readings values stretching above and below the plotted mean point?

or would you draw the error bar such that it stretches to the raw maximum and minimum values from the repeat readings you took
(edited 10 years ago)
The purpose of taking repeated measurements is that the mean value will, statistically, lie nearer to the true value the more measurements you take. Three are better than one but 5 or 6 would be better to get a good mean value.
The spread (range) about the mean then gives you the length of the error bar. (Option 2 as I have already said.)

It's still a bit of guesswork without knowing exactly what you did. No one else can fully know what the uncertainties are in your experiment without seeing what you were doing. Working out uncertainties in an experiment is something only the experimenter can do. It's all about how you take the readings and what confidence you have in them. It's a mistake to imagine that there is only one correct "exam" answer to the question "what are the uncertainties in my measurements?". There are, of course, guidelines to what is a reasonable estimate and what isn't, but within that there is some latitude.

Working out uncertainties isn't a goal in itself. It's done so that you know what confidence you can have in the final result.

Quick Reply

Related discussions

Latest

Trending

Trending

Articles for you