We began by visualizing a variety of the statistics in our data sets in Observable using choropleths and a dropdown selection for which variable to display. This made it easy to compare a range of variables on a county level across the country. Much of this exploration can be found in the following Observable notebook: https://observablehq.com/d/d0722932b0d15450. We initially intended on telling a story about the factors that impact quality of life, but switched to looking at mental health as a proxy that could be investigated more directly with the data we had. Intuitively, it made sense that poverty rate, insufficient sleep, and median household income showed strong correlation with the mental health metric, but interestingly, some of the other factors we thought would've had significant impacts, such as education, excessive drinking, physical inactivity, and unemployment, showed less influence than we would've thought. In fact, excessive drinking even showed some negative correlation, implying that people who drank more tended to struggle less with mental health issues. However, after further analysis and thought, we were able to tie this, rather, to the people of a given county's sociability scores.
We begin with a choropleth which shows both mental health data for the whole county which transitions to filter out every state but North Dakota and West Virginia. We did this to justify our focus points of West Virginia and North Dakota to the viewer as they then have time to see how the two states compare to the larger US. Next, we chose one county from each state to use as a case study. Each county was selected as it had either the best (in the case of North Dakota) or worst (in the case of West Virginia) mental health in the state. Using a radar chart to compare these two counties across sociability, obesity, unemployment, poverty, and insufficient sleep rates shows that the most major variance across these areas is sociability and poverty.
We continued investigating both of these statistics specifically. We chose to show the difference in the number of social interactions in a simple, yet aesthetically pleasing way (displaying figures for each daily interaction) to communicate how much more social Williams County was than McDowell. To visualize the job quality gap, we created bar graphs comparing the two counties over the following statistics: median household income, average travel time to work, average work week, and overall job satisfaction. We again chose relatively simple visualizations which would clearly display the data we wished to show, using additional customization for both intuitive and aesthetic appeal. We were most proud of the high level of customization we gave all of our visualizations, giving attention to every aesthetic aspect of each piece. Working with the choropleth proved difficult, and we were not able to zoom in on specific regions as we had hoped, however, we were still very pleased with how all the visualizations came together and told the story in our data.