A table uses the inclusive classes 15-24, 25-34, 35-44, 45-54. These must be converted to 14.5-24.5, 24.5-34.5, ... before the median formula is used, yet the same correction is not needed to compute the mean by the class-mark method. Why?

Making the classes continuous shifts each stated limit by half the gap, here 0.5, so 15-24 becomes 14.5-24.5. The class MARK does not move: $\frac{15+24}{2}=19.5$ and $\frac{14.5+24.5}{2}=19.5$ are the same, so the mean by the class-mark or step-deviation method is completely unaffected. The median formula, however, begins at l itself, so using 25 in place of 24.5 pushes the median 0.5 too high. Frequencies are never altered by the correction, and these inclusive classes do in fact have equal size (10 after correction).