When data violates assumption of ANOVA, then we apply transformation. But my question is that in square root transformation some value are zero then why we add only 0.5 in all observations, why not 0.1, 0.2 or 1.5 etc. Is any particular logic behind this?