Adding the idea 1 files

[marinmoore]

No description provided.

[marinmoore]

Adding the mini project idea files

[alaina-rongione]

Idea 1:

I think this is a very strong project and I am interested to see the results of your project. You know what exactly you are going to research, but start thinking about the how. Showing correlations is great, but also thing of some EDA and models that will help you make a conclusion. Also, I think it would be very interesting to see mental health statistics as well. Something that you could actually explore is the comparison of mental and physical health. Additionally, consider other biases, as well. Gender and race are also areas of discrimination, especially in medicine. I'm excited to see what you find and keep up the good work

[alaina-rongione]

Idea 2:

This is another very strong topic, and I like how detailed you are with your proposal. There is a lot of exploratory opportunity with this project and I look forward to seeing the findings. I do think you should be a bit more specific in what the factors are that you are looking at. I would assume there is a lot of data out there for all the factors that cause low birth weight and preterm birth. Focus on 1-3 different factors (possibly the most popular or major ones) and explore those. Having too many factors can cause confusion with the project, and have you dedicate so much time to the project that you may not have.

[ernbilen]

Interesting question. Causation/correlation is a big challenge to address here. I like how you are addressing it by using state level policy changes which gives an exogenous shock on discrimination. Of course it is important to define discrimination clearly with its multi layers. Is it discrimination in neighborhood, workplace, or news signals coming from policy changes? I think your differences in differences would be capturing the news signals, but there is the possibility that those news go to "neutrals" who wouldn't discriminate normally, but may be swayed by the passing of bills. So these identification channels need some thought and maybe some extra work.
I wouldn't put too emphasis on aggregate graphs on health and bills. It's just too broad as you said. But you have data on state level so you can do so much better. Looking at MO graph, you may be onto something! Are the bars showing statistically significant differences across years? You can put error bars and show.
I wonder if there are bills that reversed discrimination, which should improve health outcomes if the channel works both ways.
Very well discussed implications and ethics.