What type of error bars should I use in my graphs?

[emu112]

In excel, there are 5 types; fixed value, percentage, standard deviation, standard error, and custom. Which one do I use??

[IB`NOT`ez]

As a GENERAL rule of thumb, use (Max-Min)/2 errors if you have less than 5 trials, or you noticed an anomaly. Standard deviation error bars work better for trials numbering 10 or above. For Physics and Chemistry, it's safer to proceed with (Max-Min)/2 error bars as there's fewer factors that can cause a wide spread/range of data. 

Percentage uncertainties of averaged values are also good to process, just to show the extent of systematic errors, and allow further discussion on the apparatus used.

A amazing thing to have in your IA is a "Table of Uncertainties" (an example image from a level 7 Physics IA is embedded on this post). It ticks so many of the Criterias – Analysis, Exploration (since it's part of your Methodology & Variables), Evaluation (you'll be referencing values and processings done here in your conclusion), and Personal Engagement e.g. "The velocity's uncertainty is small but I chose not to ignore it due to its role as an Independent Variable", "Slip-off could occur on the sand... safer to assume x uncertainty", "Furthermore, the ruler has a gap..." – these demonstrate personal engagement and input with your exploration, as well as demonstrating a high degree of precision with your analytical skills. It also adds to your communication when you start processing your data, as you've outlined in the table the uncertainties you'll be propagating in the later sections of your IA. 

(Source: extensive discussion with my Biology & Chemistry teachers when doing my IAs, as well as my school's Physics teacher who was an IB Physics IA moderator)

Good luck! 

Edited April 27, 2017 by IB`ez