Adding a small noise to the original data might be helpful in visualizing data, especially for categorical data. The scatterplot of two categorical variables just m x n points, from which we do not know how many sample points are actually in these graphical symbols. If we jitter() the data before drawing a scatterplot, the effect can be better.
Refer to the example introduced in the start page of R package ''animation''.