Background

In the 2012 House Elections, our data recorded 2,178 candidates vying for a seat in the House of Representatives. Each candidate has different supporters from their respective state. And these supporters all come from their own blend of unique socio-economic conditions and political attitudes.

Working through this information, we became interested in how fundraisers, donations, and supporter finances varied among different candidates.

The House of Representative seats have unlimited two year terms per candidate, and therefore we can expect that certain candidates have been holding onto their seat comfortably for many years. These comfortable candidates would be less likely to seek out funds from individual supporters. Vice-versa, a supporter may be less likely to donate to a representative who’s been winning the seat for 20-odd years. Other candidates may be in a heated campaign, trying hard to amass supporter donations and win against their worthy competitor.

Research Question

The question we asked from the information was:

“Which candidates won a seat in the House with the lowest and greatest supporter contributions in the 2012 election?”

Data

In compiling and tidying the data, we selected two tables of interest to help us answer the question.

“House Elections”

“House Elections” has 2,178 observations. The observations in the “House Elections” data set include candidate ID number, candidate name, party allegiance, if they won election, and how many primary/general/run-off votes they received.

This data set is useful to us because it provides us with information on the running candidates and whether they have won a seat or not. In particular, we can merge this information with the contributions set (below) to learn how much money a candidate received from supporters AND if they actually won a seat.

“Contributions”

“Contributions” has 386,369 observations. Each observation is a contribution from a supporter (some of which happen on the same day from the same fundraiser/individual) that went to a candidate running for either a committee, the House, or the Senate.

We intentionally filtered the information through a left_join to get only contributions going to candidates running for House elections. The left_join function also gave us the merged data-set from which our information analysis stems from.

To merge, we found that the candidate ID number was a shared feature in both data sets.

#Loading the relevant data. 

library(tidyverse)
house_elections <- read_csv("house_elections.csv")
load("contributions.rda") 

#Renaming columns to merge data sets. 

names(house_elections)[1] <- "cand_id" 

#Specifying useful columns from House Elections data set. 

HouseWinnings <- 
house_elections %>% 
  select(candidate_name, state, primary_votes, cand_id, candidate_name, 
         party, general_votes, ge_winner) %>%
  arrange(cand_id) 

#Creating a new table with total contributions as a new column. 

Contributions <- contributions %>% 
  group_by(cand_id) %>% 
  summarise(Total_Contributions = sum(transaction_amt)) 

#Merging two data sets.

WinContrib <- 
  left_join(HouseWinnings, Contributions, by = "cand_id") %>% 
  filter(!is.na(cand_id))

#Filtering to only those with House seats and arranging to reveal highest/lowest contributions. 

LowCostWinners <- 
WinContrib %>% 
  select(candidate_name, state, ge_winner, Total_Contributions, 
         party, primary_votes) %>%
  filter(ge_winner == "W")%>% 
  arrange(Total_Contributions)

HighCostWinners <- 
  LowCostWinners %>% 
  arrange(desc(Total_Contributions))
 
#Filtering to only those that didn't win a seat, and arranging to reveal highest contributions (biggest losers). 

HighCostLosers <- 
WinContrib %>% 
  select(candidate_name, state, ge_winner, Total_Contributions, 
         party, primary_votes) %>%
  filter(ge_winner == "N")%>% 
  arrange(desc(Total_Contributions))

head(LowCostWinners)
## # A tibble: 6 × 6
##           candidate_name state ge_winner Total_Contributions party
##                    <chr> <chr>     <chr>               <dbl> <chr>
## 1  Smith, Christopher H.    NJ         W                8001     R
## 2       Smith, Robert N.    TN         W                9754     R
## 3 Faleomavaega, Eni F.H.    AS         W               10750     D
## 4       Curson, David A.    MI         W               15629     D
## 5         Sarbanes, John    MD         W               19089     D
## 6 Bordallo, Madeleine Z.    GU         W               33905     D
## # ... with 1 more variables: primary_votes <int>
head(HighCostWinners)
## # A tibble: 6 × 6
##     candidate_name state ge_winner Total_Contributions party primary_votes
##              <chr> <chr>     <chr>               <dbl> <chr>         <int>
## 1     Renacci, Jim    OH         W             6909165     R         66487
## 2 Duckworth, Tammy    IL         W             5619978     D         17097
## 3      Latham, Tom    IA         W             5363782     R         27757
## 4     Barrow, John    GA         W             5296867     D         41587
## 5    Bustos, Cheri    IL         W             5293962     D         18652
## 6    Benishek, Dan    MI         W             5252284     R         64411
head(HighCostLosers)
## # A tibble: 6 × 6
##               candidate_name state ge_winner Total_Contributions party
##                        <chr> <chr>     <chr>               <dbl> <chr>
## 1             Critz, Mark S.    PA         N             6972650     D
## 2               Lungren, Dan    CA         N             6368881     R
## 3              Sutton, Betty    OH         N             5816877     D
## 4 Canseco, Francisco "Quico"    TX         N             5733298     R
## 5             Cravaack, Chip    MN         N             5695571     R
## 6             Bilbray, Brian    CA         N             5690358     R
## # ... with 1 more variables: primary_votes <int>

Results

The conclusion can be simplified to this (single-sentenced) fact: In the 2012 House Elections, Christopher H. Smith (from New Jersey) was the seat-winning Representative to receive the lowest amount of contributions from supporters, while Jim Renacci (from Ohio) was the seat-winning Representative to receive the highest amount of contributions from supporters.

Smith, the Lowest Cost Winner

Christopher Smith, a moderate Republican from New Jersey, has a very active history in encouraging and pushing legislature. His strongest positions stand for gender equality, harassment protection, and anti-abortion. He has held his position since ’81 (wow!). The data here supports the conclusion that Smith was comfortable in his seat in the House, and neither supporters nor his campaigners deemed it necessary to send or seek more money for the 2012 election. For that reason, Smith won the election with only $8,000 in contributions.

Renacci, the Most Expensive Winner

Jim Renacci, a Republican from Ohio, is a strong business man. Strictly opposed to taxes, he spends most of his position protecting entrepreneurs and small businesses in Ohio. He won his seat in 2012 by a vote count of 181,137 to 165,636. That was extremely close! Our data supports that his campaigners and supporters would feel the urgency to donate significantly large amounts of money to his campaign to aid him through this incredibly competitive race.

Renacci accepted a total of $6,909,165 during his campaign.

Critz, the Most Expensive Loser

Just as a fun aside, we also manipulated and re-filtered the data set to reveal the most donated-to loser in the House elections of 2012. Our result led us to Critz. This candidate accepted just over what Renacci did - $6,972,650 - and still didn’t come up with a seat!… What a bummer.

Visualizing Individual Contributions by Primary Votes in House Candidates

#Assigning factors and numerics to columns to help axis presentation. 

WinContrib$party <- as.factor(WinContrib$party)
WinContrib$state <- as.factor(WinContrib$state)
WinContrib$ge_winner <- as.factor(WinContrib$ge_winner)
WinContrib$Total_Contributions <- as.numeric(WinContrib$Total_Contributions)

#Filtering NA's from the data-set and rearranging. 

WinContrib <- WinContrib %>% 
  filter(!is.na(Total_Contributions)) %>% 
  filter(!is.na(ge_winner)) %>%
  arrange(Total_Contributions)

#Producing the scatterplot. 

CandidatePlot <- ggplot(data = WinContrib, aes(Total_Contributions/1000000, general_votes)) + 
  geom_point(aes(col = ge_winner)) +
  labs(x = "Total Contributions ($ in millions)", 
       y = "General Votes", 
       col = "Election Winners", 
       title = "2012 House Election Candidates by Contributions\nand General Votes Received",
       alpha = NULL) +
  theme_classic()
CandidatePlot

Descisions with Visualization

This scatter-plot helps us visualize the breadth of winners and losers in the 2012 House Elections by the amount of contributions they received from supporters and general votes gained in the actual election.

The x-axis represents the amount of contributions each candidate received. To make this information cleaner, we divided the total contributions by 1,000,000 (this makes the numerics into single digit values). The y-axis represents the amount of general votes each candidate received.

Our information is color-coded by whether or not the candidate won a seat in the House elections. The entries coded red did not receive a seat, whereas the entries coded blue did.

The most important use of this color-coded scatter-plot is that it can help us shape the relationship between the total number of contributions and the prospect of winning the election. We can following along the x-axis as the total contributions to each candidate increases, and determine by the color of each observation whether that candidate managed to snag a seat.

Finally, the choice of a scatter-plot is also useful because we were dealing with two numerical values which can be considered to have a significant correlation. As total contributions from supporters increase, the total number of general votes the candidate receives should also increase. This is most likely because individuals who donate to a candidate are also likely to vote for them in the general elections.

However, it is important to note that this graph isn’t as useful in describing the correlation between votes and contributions as it is between winning a seat and contributions. This is because the total number of votes available varies from state to state. Some candidates in less-populated states may receive fewer votes (and find themselves lower on the y-axis) but may have been quiet popular in their election given the size of their state.

Analysis

The central take-away from this scatterplot is that the total contributions a candidate receives is moderately related to whether or not the candidate managed to win a seat in the 2012 House Elections.

More specifically, this graph gives a few interesting points.

For one, candidates which received contributions nearest to zero were most likely to not win a seat. This is demonstrated by the cluster of red dots nearing the origin of the graph.

Secondly, candidates which received just a little bit of contributions (around $1 million, as opposed to Renacci’s near 7 million donations) were more likely to win. This may be because there is generally some amount of donations given to every winning candidate, and a likable candidate who is facing no competition can anticipate at least some amount of contributions but would probably not be donated excessive amounts (think Renacci).

Lastly, as contributions increase, the amount of red dots and blue dots at a particular point on the x-axis becomes considerably similar compared to the clusters and blues and reds near the origin. This may indicate that the race for the House seat for these candidates were more competitive. More money was given to both competing candidates to support them, but only one could prevail and win the spot.