Geometric mean concentration is used for antibody titers because the titer data usually does not fit a linear scale. When the data is skewed and not normally distributed, it is better to calculate the geometric mean instead of the arithmetic mean, which may not give a good representation of the results.