At every node, our tree splits into two. As you can see here, 2 is a branch. Each branch splits the sample of survey respondents by a different circumstance. In this case, the first split was in household wealth. This means that our algorithm found the highest variance amongst groups divided by wealth. The branch itself adjusts its size based on the proportion of the sample that has been split into the new group. Please note that the nodes show access rates, and the width of the branches reflects proportion of the sample. Our first split is between the Top 60 Wealth and Bottom 40 wealth, where we see the greatest variance from our circumstances. The Top 60 Wealth has a 100% access rate to electricity, whereas the bottom 40 wealth only has a 82 access rate to electricity.