I assume your original data is capped at 100% (=best score) ? Then for sure your data won't be normally distributed since you are missing the values on the upper side (assuming from your mean that your are in ceiling range anyway)... That is the reason why you get a mean + SD > 100% (since the a normal distribution is assumed around your mean, which does not take your 100%-cap into account).
I'd recommend the solution of Assem with reporting quantiles, but depending on your goal, Mean +/- SD will also be accepted...
Hi Mohamed, basically, if the data distribution is skewed, the mean isn’t the best measure, and in your question is the answer, you can report the median and the interquartile range.