Charts with stacked values work well when the focus is on comparing the totals for each column. We look at the absolute value of the column and can measure how many percentages each inner category takes.
But when it comes to measuring the variation of the inner categories over time, things get more complicated.
Our visual perception easily compares look-a-like categories. In the case of a stacked bar chart, we can easily measure only the variation of the lowest category — its baselines are set off from the common axis and have a stable "foundation". But it becomes impossible to measure the variation of the other fractions. In this case, it is better to use separate bar charts.
