sebastianbarfort / sds

Social Data Science, course at University of Copenhagen
http://sebastianbarfort.github.io/sds/
12 stars 17 forks source link

Group 18: Project Description #76

Closed ThorsteinTN closed 8 years ago

ThorsteinTN commented 9 years ago

Get by with a little help from a friend: Presidential candidates and Super PACs

1 Idea

The purpose of this project is to examine the Super PACs and, if possible, their influence on politics in the US.

A Super PAC is a relatively new type of political action committee that arose following two federal court decisions that found that limitations on corporate and individual contributions are unconstitutional violations of the First Amendment right to free speech. The Super PACs may raise and spend unlimited amounts of money from associations, individuals, unions and corporations to advocate for or against political candidates or parties. The Super PACs are not allowed to make donations directly to the political candidates or parties, but can engage in unlimited political spending and support independently of the political campaigns.

The size of the Super PACs varies greatly. Some of the Super PACs have raised as much as 150 million dollars over two years, which must be considered to be a large enough amount of money to influence the elections of house representatives, senators and presidents. The Super PACs are therefore relevant to examine.

This project will examine a wide range of problems:

To our knowledge such an analysis has not been made before. The Centre for Responsive Politics has made some descriptive analysis describing the Super PAC but has, as far as we know, not made the suggested analysis.

2 Data

The Super PACs are required to report their donors to the Federal Election Commission on a monthly or semi-annual basis in off years, and monthly in election-years. This makes it doable to examine the Super PACs. The Centre for Responsive Politics (http://www.opensecrets.org/) has collected data for the Super PACs for the cycles of 2010, 2012, 2014 and 2016.

We will collect this data by scraping the webpage, using two loops (as in the second home assignment) for each of the four years and for every Super PAC. Each year has a separate page containing data on:

  1. The name of the Super PAC
  2. Who the Super PAC supports/opposes
  3. The total independent expenditure
  4. The viewpoint (conservative or liberal)
  5. The total amount of money raised.

Further, each Super PAC also has a separate page with data for every contribution made, containing information on:

  1. The politician the Super PAC supports/opposes
  2. The state of the politician
  3. The party of the politician
  4. Which office (house, senate or presidential) the politician is running for
  5. The size of the expenditure
  6. The amount of money spent on supporting or opposing the politician.

Each page also contains information about the total amount spent for and against each of the two parties (democrats and republicans).

The database does not contain information about the date of the submitted contribution or when they were spent for the elections in 2010, 2012 and 2014, which could have been a very interesting feature to add to our analysis. There are however a slightly expanded dataset for the 2016 elections where they have added independent expenditure made within the last two weeks. It includes candidate, race (house, senate or presidential), size of expenditure, the payee, whether it was for supporting or opposing, the date and a note about how the money was spent.

3 Method

Parts of the analysis will be descriptive and will consist of graphs, maps etc. This is relevant for all parts of the analysis except the examination of the relationship between Super PACs and the results of the election. To examine the relationship between the use of Super PACs and political view, we will further make significant-tests. As of now, we haven’t thought of a specific estimation model since it depends on the final structure of our data.

The examination of the relationship between the support from Super PACs and the results of the elections will address problems of endogeneity. E.g. to control for vote-getting ability across candidates, we aim to examine elections in which the same two candidates face one another on more than one occasion; differing eliminates the influence of any fixed candidate or district attributes.

Also, we touch upon potential problems of reverse causation. Getting by with a little help from a friend can refer to two friends: Either it refers to the politician getting elected with the help from super PACs, more specifically the financial resources. The other interpretation, however, is equally likely. Assuming - as it often is - that political actors are rational, utility maximizing entities, super PACs might very well support Politicians in order to gain access after a succesfull election. Thus, reverse causation cannot be ruled out, as super PACs might simply play a bandwagon strategy, supporting an already popular candidate. An observation supporting the last interpretation is if data indicates that (some) super PACs support politicians on both sides of the aisle. However, as the alternative to the given politician is unknown, such behavior cannot rule out that super PACs support the most likeminded politicians.