When you perform an experiment, you obtain certain value which is subject to errors.
Those errors may be of different nature, may be systematic or random. Also, the error may be caused by a human mistake performing the experiment or by the intrinsic model used to perform the experiment.
In any case when you know the accepted or theoretical value of the variable you are measuring, the percent error is computed as:
Absolute error
Percent error = -------------------------------------------- * 100
accepted or theoretical value
And, the absolute error is the difference between the accepted or theoretical value and the experimental value.
If the accepted value is greater than the experimental value, the absolute error is: accepted value - experimental value.
If the accepted value is less than than the experimental value, the absolute error is: experimental value - accepted value.
So, the context of the question and the set of choices to answer is important to know the exact answer.
At, the end may be written as:
|accepted value - experimental value |
Percent error = ---------------------------------------------------- * 100
accepted value.
Which you must use in your context to choose the right choice.