# A tibble: 2,636 × 6
`PAC Name (Affiliate)` Country of Origin/Pa…¹ Total Dems Repubs year
<chr> <chr> <chr> <chr> <chr> <dbl>
1 7-Eleven Japan/Ito-Yokado $8500 $1500 $7000 2000
2 ABB Group Switzerland/Asea Brow… $460… $170… $28500 2000
3 Accenture UK/Accenture plc $759… $230… $52984 2000
4 ACE INA UK/ACE Group $385… $125… $26000 2000
5 Acuson Corp (Siemens AG) Germany/Siemens AG $2000 $2000 $0 2000
6 Adtranz (DaimlerChrysler) Germany/DaimlerChrysl… $105… $100… $500 2000
7 AE Staley Manufacturing (Tat… UK/Tate & Lyle $240… $100… $14000 2000
8 AEGON USA (AEGON NV) Netherlands/Aegon NV $582… $105… $47750 2000
9 AIM Management Group UK/AMVESCAP $250… $100… $15000 2000
10 Air Liquide America France/L'Air Liquide … $0 $0 $0 2000
# ℹ 2,626 more rows
# ℹ abbreviated name: ¹`Country of Origin/Parent Company`PAC Analysis
Introduction
This project aims to examine the impact of foreign-connected Political Action Committee (PAC) donations on US politics from 2000 to 2024. Specifically, it will analyze the possible correlation between the contributions made by PACs affiliated with foreign companies or entities to Democratic and Republican seats in Congress to the presidential election results. The ultimate goal is to determine whether there is a relationship between Foreign PAC Donations and US political results, and if so, to understand the extent to which such donations influence political outcomes and partisan dynamics in the United States.
Data Methodology
OpenSecrets
For OpenSecrets Dataset (Raw: 2636 Rows; No NAs; Predictors: PAC Donation amount):
The dataset includes information on PAC donations from foreign-connected entities to US political parties from 2000 to 2024 (Initially in two-year intervals but was later combined).
It consists of columns such as PAC Name (Affiliate), Country of Origin/Parent Company, Total donation amount, and amounts donated to Democratic and Republican parties. Each row represents a different PAC donation entry, detailing the contribution amounts and affiliations.
Data types in the dataset include strings for PAC names and countries of origin and integers or floats for donation amounts.
After some data wrangling and processing, the ultiamte PAC dataset that will be analyze upon looks like following:
# A tibble: 286 × 5
country dems_sum repubs_sum total_sum year_cycle
<chr> <dbl> <dbl> <dbl> <dbl>
1 Australia 61850 174000 235850 2000
2 Australia 70019 42333 112352 2002
3 Australia 137969 113498 252705 2004
4 Australia 21088 8500 29588 2006
5 Australia 25158 13750 38908 2008
6 Australia 32376 17000 49376 2010
7 Australia 53353 24500 77853 2012
8 Australia 31655 44100 75755 2014
9 Australia 39285 92200 131485 2016
10 Australia 35601 116506 152107 2018
# ℹ 276 more rowsCongressional Distribution by Parties
For Wikipedia Dataset (Raw: 118 Rows; No NAs; Target Variable: Seats by Parties):
The dataset provides a record of the party divisions within the United States Congress since 1789 through various congressional sessions.
It includes columns such as Congress, Years, Senate composition (Total, Anti-Administration, Pro-Administration, Others, Vacancies), House of Representatives composition (Total, Anti-Administration, Pro-Administration, Others, Vacancies), President, and Trifecta status. Each row corresponds to a specific congressional session, detailing the composition of the Senate and House of Representatives, any presidential affiliations, and whether a trifecta scenario was present.
Data types in the dataset include integers for numerical values such as the total number of seats and vacancies, strings for textual information such as the President’s name, and categorical variables for party affiliations.
Similarly, the ultiamte Congress Parties dataset that will be analyze upon looks like following:
# A tibble: 13 × 5
year democratic_senate_1 republican_senate_1 democratic_house_1
<dbl> <dbl> <dbl> <dbl>
1 2000 45 55 211
2 2002 50 50 212
3 2004 48 51 205
4 2006 44 55 202
5 2008 49 49 233
6 2010 56 40 257
7 2012 51 47 193
8 2014 53 45 201
9 2016 44 54 188
10 2018 46 52 194
11 2020 45 53 235
12 2022 48 50 222
13 2024 48 49 213
# ℹ 1 more variable: republican_house_1 <dbl>Analysis
Overview of PAC Influence on US Congress Representation
Correlation Between House of Representative and Donation Percentages: Republican: 0.745 Democratic: 0.736
Correlation Between Senate and Donation Percentages: Republican: 0.612 Democratic 0.604
From the plots, it seems that the party representation in Congress almost parallels the trend of foreign-connected PAC donations.
In fact, looking closely, the trend of donations seem the predict the representation as the changes in donations happen earlier than the changes in representation.
Additionally, it seems that Republicans usually receive more PAC donations than Democrats, but the percentage of donations of the Democrats seem to be growing closer to that of Republicans in recent years.
PAC by Countries/Regions
Election Prediction Using PAC Donations
Visualization
This graph illustrates the total sum of PAC donations received by Democratic and Republican candidates over different election cycles from 2000 to 2024. The donations are represented by blue lines for Democrats (dems_sum) and red lines for Republicans (repubs_sum). The background shading indicates the president’s party during each period, with blue representing Democratic presidencies and red representing Republican presidencies.
Observation: - From 2000 to around 2008, Republican PAC donations were consistently higher than Democratic donations, reflecting stronger Republican fundraising during this period.
A significant surge in donations for both parties occurred around 2008-2012, with Republicans experiencing a peak in donations around 2012.
Post-2012, Republican donations showed a sharp decline, while Democratic donations remained relatively stable, with slight fluctuations.
By 2020, Democratic donations briefly exceeded Republican donations but then both saw a decline leading into 2024.
By 2024, although both party have significantly less donations, the Republican party have a slightly better edge in terms of momentum, which sole based on the historical pattern, Donald J Trump is expected to win the 2024 Presidential Election over current President Joe Biden.
The fluctuation in donations aligns with the changes in the president’s party, suggesting that political context significantly impacts fundraising efforts.
This graph tracks the yearly changes in PAC donations for Democrats (diffdems) and Republicans (diffrepubs) over the same period. The changes are calculated based on the differences in donations from one cycle to the next. The background shading again represents the president’s party, with blue for Democratic presidencies and red for Republican presidencies.
Observation:
Winning Party Momentum: There is a noticeable spike in the change of donations for the winning party in election years:
2008 (Obama): Significant increase in Democratic donations around 2008-2010; momentum prevails Republican Party.
2016 (Trump): Marked increase in Republican donations around 2016; momentum prevails Democratic Party.
2020 (Biden): Significant increase in Democratic donations around 2020; momentum prevails Republican Party.
Both parties show fluctuations in the change of donations, with Republicans generally seeing more volatility compared to Democrats.
By 2020, both parties show a decrease in yearly changes, reflecting a downturn in PAC donations heading into the 2024 election cycle.
The graph highlights that while both parties experience significant changes in donations, the timing and magnitude of these changes are influenced by the prevailing political climate and the party in the presidency.
Model Creation
This model aims to predict the political party of the president for the year 2024 using historical data on political contributions from 2000 to 2024, with training limited to data up to 2012. The process begins by loading necessary libraries for data manipulation and machine learning. Historical data is then prepared by calculating the lagged values of contributions, which represent contributions from the previous cycle, and computing the ratios of current to lagged contributions to capture relative changes. A Random Forest classifier is trained on this processed data, using the election cycle, sums of contributions for each party, and the computed ratios as predictor variables. For the 2024 prediction, a new dataset is created with projected contribution sums, and ratios are calculated based on the latest available training data. The trained model then predicts the political party of the president for 2024 based on this new data. The final output includes the year, projected contribution sums, calculated ratios, and the predicted political party. This model leverages historical trends and relative changes in political contributions to forecast the political outcome for the 2024 presidential election.
| Year Cycle | Democratic Sum | Republican Sum | Democratic Ratio | Republican Ratio | Predicted President Party |
|---|---|---|---|---|---|
| 2024 | 4272336 | 6192659 | 0.5231894 | 0.5274752 | Republican |
The prediction for the 2024 presidential election indicates that the Republican Party is expected to win. This is based on a Random Forest model that used data from previous election cycles. The data shows that the Democratic contributions sum to 4,272,336, while the Republican contributions sum to 6,192,659. The ratios of current to previous contributions are 0.523 for Democrats and 0.527 for Republicans, suggesting a slightly higher momentum for the Republicans, leading to the predicted outcome.
This graph illustrates the variable importance in a Random Forest model by showing the contribution of each variable to the model’s predictive performance. It uses the “Mean Decrease Gini” score to quantify the influence of each factor, with a higher score indicating a more influential variable.
These plots show the partial dependence of different variables in the model on the model’s predictions. Each plot represents how a specific variable, such as “dems_sum,” “repubs_sum,” “dems_ratio,” and “repubs_ratio”, influences the model’s output while keeping all other variables constant.
Conclusion
Ultimately, it’s important to keep in mind that such model and analysis only attempt to use one single factor, PAC Donation, to try to offer insights into an extremely complicated human event, US election. Therefore, it’s important to understand the constraint and limitness in terms of information this model receive when making its prediction. This project have prove that foreign influence through Political Action Committee does seems exist and also hold a significant stake in the party composure of the US Congress as seen with the high correlation factor. Addditionally, during presidential election years, the party that seems to have more momentum as define by the ratio of PAC Donation have always led to the winner of the election since 2000. If this trend continue, it is predicted that a Republican President will came to office of the 2024 election if the current PAC donations to each party maintains the same.