One reflection

[scao1]

To rule out that baseline crime rate affects the changes of crime rates, it's better to separate the states into low base line group and high baseline group and then compared the changes of crime rate before and after voting within each group and calculates the significance within each group. We tried our best effort to figure out how to make the plot work that way but unfortunately failed. The plot we have right now may not be a optimal plot to address the research question.