Food Access

In the course of our analysis, we found that food access may not be a suitable instrument variable because the factor itself is less associated with voting propensity, and at the same time, a qualifying instrument variable should not have any relationship with other variables that affect voting propensity. 

However, since we don't know the vehicle ownership rate of the analyzed population, we can't actually tell the true economic level of the person based on the distance of food access alone. 

On the other hand, we have limited data at our disposal, because we only have the data for one-half, one mile, ten miles, and twenty miles. One mile is the distance from most grocery stores, while ten miles is the radius of a city. 

GDP Per Capita

    In the exploration of the relationship between economic output and political identity, the analysis of county-level Gross Domestic Product (GDP) and voting patterns unveils intriguing insights into the dynamics of the American political landscape. 

The Brookings Institution's study on the 2016 and 2020 U.S. elections highlights a clear economic demarcation between counties favoring Democratic candidates and those supporting Republican candidates:

The institution scrutinized the results of the 2016 and 2020 U.S. elections and discerned a clear economic demarcation between counties voting for Democratic candidates (Hillary Clinton in 2016 and Joe Biden in 2020) and those favoring the Republican candidate (Donald Trump in both years). Counties voting Democrat represented a larger fraction of the U.S. GDP: 64% in 2016 and 71% in 2020. In contrast, counties supporting Trump accounted for 36% and 29% of the GDP in 2016 and 2020, respectively. This economic divide extends beyond sheer GDP numbers, with stark differences in the types of economies these counties represent.

Democratic-leaning counties are characterized by more diverse, educated, and white-collar professional economies, while Republican-leaning counties exhibit different economic characteristics, relying more on traditional industries. These economic distinctions exacerbate the division between people's preferences in elections since a lot of people will prioritize the interests of their respective constituencies. Furthermore, the relationship between GDP per capita and political identity is complex, as evidenced by a density graph representing county-level data.

Based on the dataset that we used and the graph that we produced: 

The X-axis represents the GDP per capita, with a particular focus on the range of approximately $26,000 to $27,000. The Y-axis measures the density of counties, differentiated by their political leaning: strongly Democratic (3.0), leaning Democratic, tossup, leaning Republican, and strongly Republican (2.4).

The density graph illustrates two significant peaks at the GDP per capita range of approximately $30000. Both Democratic and Republican strongholds tend to concentrate around the same GDP per capita range ($26,000 to $27,000), implying that the economic output per person does not significantly differ between these two political extremes. However, the density of Democratic counties at this GDP per capita range is slightly higher (3.0) than that of Republican counties (2.4).

However, the graph also reveals a significant difference in the economic profiles of counties with lean and tossup political affiliations. These counties, rather than exhibiting a strong peak, follow a more normal distribution on the graph. Although there is a slight skew to the right, indicating a modest overrepresentation of wealthier counties, the distribution's standard deviation is notably lower. This lower standard deviation indicates a relatively even spread of economic output amongst these counties, pointing to an economic diversity that is less pronounced in strong Democratic or Republican counties.

Interestingly, counties categorized as lean or tossup display higher GDP per capita than their counterparts with a strong party alignment. This phenomenon suggests that economic wealth is not a decisive factor in strong party affiliation at the county level. This counters intuitive assumptions that economic wealth might correlate with a specific political inclination, underscoring the multifaceted relationship between economic output and political identity.

In sum, the density graph contributes an invaluable dimension to our understanding of the interplay between GDP per capita and political alignment. It elucidates the economic similarities and differences between counties with different political affiliations, providing a nuanced picture of the economic landscape of American politics. The graph offers a stark visual representation of the economic disparity within the American political landscape, highlighting the critical role of economic factors in shaping political identities.

The graph reveals significant peaks in the GDP per capita range, indicating similar economic output for both Democratic and Republican strongholds. However, counties with lean or tossup political affiliations display a more evenly spread distribution of economic output, suggesting that economic wealth is a significant factor in strong party affiliation at the county level. This challenges conventional assumptions and underscores the multifaceted nature of the relationship between economic output and political identity.

Housing Price

       Another issue we decided to look at is wealth. Unfortunately, it is difficult to calculate people’s wealth, so we attempted to use an Instrumental Variable Starter House Prices for personal wealth, which is Starter House Prices, as a proxy for personal wealth. If Starter House Prices were a suitable instrumental variable, we would anticipate observing consistent trends in GDP growth corresponding to changes in Starter Housing Prices.

Upon examining the relationship between Starter House Prices and political affiliations, we found that Republicans exhibited slightly higher variance in house prices compared to Democrats. Specifically, when focusing on counties where Republicans won by a margin of 65% or more, the standard deviation of log-transformed Starter Housing Prices was 0.67, while strong Democrats had a standard deviation of 0.40.

This discrepancy could be attributed to Republicans' preferences for rural/suburban areas.  However, the mean house prices were similar between Democrats and Republicans. These findings suggest that Starter Housing Prices may have a limited influence on overall wealth.

In evaluating Starter House Prices as an instrumental variable for wealth, it becomes apparent that it falls short of meeting the necessary assumptions. When using instrumental variables, two critical assumptions are made: first, that the instrumental variable is relevant to the analysis, and second, that the variable is solely associated with wealth and not influenced by other factors.  However, these assumptions can be easily disproved in the case of house prices. Building houses incur substantial costs, estimated at around $300,000 according to Forbes, which directly impacts housing prices. Additionally, house prices can be influenced by their commercial value, as landlords may purchase properties as revenue-generating assets. Consequently, housing prices are not a reliable metric for measuring wealth as their effects can be explained by other factors.