How US state-specific characteristics relate to political affiliation
The aim of my project is to look at some specific demographics of a few key states in
the United States, in order to potentially draw some conclusions regarding their connection to
the political affiliation that state historically identifies with. For this, I choose to look at the following states:
This is a density graph including
data from the six different states, grouped together as republican (red) and democrat (blue). The data
was downloaded from the FRED website and combined six different time series from 2006-2019 for each state. The colab notebook for the manipulation
of the data can be found here.
As I was combining six different datasets, I firstly merged them using Stata and then exported the csv file onto colab. From this graph, we could draw the conclusion
that democratic states are in general, more educated than republican states.
This graph was made using six different FRED APIs, one for
each of the states. As the FRED API cannot be accessed directly from an API call, I used a CORS reverse proxy. As illustrated by the graph, up until the Financial Crisis of 2008,
the rank of median household income was relatively stable, with no
particular pattern differentiating republican and democratic states. More recently however, specifically after 2014, all three democratic states from our sample
are ranking at the top while the three republican ones are at the bottom. While Republican areas of the country rely
on “traditional” industries that require lower skills, like manufacturing and resource extraction, while Democratic, mostly urban districts contain large concentrations of the nation’s
higher-skill, higher-tech professional and digital services. (source)
For this graph, I used as a proxy for LGBTQ population same-sex couple households.
This proxy might not be the most accurate since a lot of same-sex couples do not live together and there are members of the LGBTQ community that are not in a same-sex relationship.
However, it can still illustrate some useful results. I used data from the Census for same-sex
couple households as well as total number of households and with my own calculations, using Excel,
I generated the data to create the graph. The results are similar throughout the whole period; democratic states have larger LGBTQ populations relative to their total population.
For the linear regression graph, I regressed the percentage of republican vote on the percentage of Fox News listenership
for all 50 states, for more representative results. The graph suggests a strong positive correlation between Fox News and republican voting. Fox News has been described as practicing biased reporting in favour of the
Republican Party, its politicians, and conservative causes while portraying the Democratic Party in a negative light. Research also suggests that Fox News has a causal impact on republican voting. (source)
The chart divides people into three different groups: Nonreligious, Moderately Religious and Very Religious.
The data was edited manually on Vega-Lite using the data from Statista.
The chart suggests that the republican states from our sample tend to be more religious while democratic states have higher percentages of nonreligious populations.
The underlying explanations for the relationship are complex, including differing positions of the parties on moral and values issues such as abortion and same-sex marriage, and geographic patterns of
residency that are simultaneously related to religiousness and partisanship. (source)
Race and Ethnicity
The trellis chart illustrates the race
distribution in the six different states. The data was scraped using Python Pandas and the colab notebook can be found here.
The desired data format in order to create the trellis chart was achieved using Excel.
From the chart we could conclude that the demographic composition of an area does not provide a significant correlation
with its partisanship. Republican voters are in general more likely to be white, however, we see a strongly Democratic state such as Washington being
made up by a mostly white population, while the Republican state of Texas has a less significant percentage of white population. While Texas is not currently considered
a swing state, changes in its demographic composition, mostly due to immigration, has let some to wonder if the state could become more politically competitive down the road.
Conclusion and Limitations
There are a few fundamental differences in the partisan orientation of different demographic
groups, which have also grown wider in recent years, making the USA as polarized as ever. While most of the characteristics presented above provide clear correlations between them and political party
identification, the evidence presented is not enough to draw conclusions about causality. In most cases, the relationship could be two-way causal as parameters influence one another. However, the results shown provide some useful insight
and a motivation for further research on the subject.
Note: The data sources used in each graph are indicated in the subtitle of the chart. For any sources used in the text, links are available.