Variables and Methods

2004 Democratic Primaries

The candidates used in this study are:
Wesley Clark
Howard Dean
John Edwards
Richard Gephardt
John Kerry
Joseph Lieberman

Variables

Dependent Variable

• Amount of coverage each candidate recieved on a particular issue
  • stories that appeared in The Washington Post and The New York Times that had the candidate's name and the particular issue

Independent Variables

• Press releases issued by the candidate
  • as listed on the candidate's websites
• The day in the campaign process
  • each day in the data set was assigned a number in chronological order between 1 (September 1, 2003) and 185 (March 3, 2004)
• The number of candidates in the race
  • the number of candidates in the race on a particular day. Ranged from 1 (only John Kerry) to 6 (all candidates still in race)
• The day of the week
  • whether it was Monday, Tuesday, Wednesday, Thursday, Friday, Saturday, or Sunday
• State of the Union address
  • whether it was the day of the address
  • whether it was the day after the address
  • whether it was 2 days after the address
• Press conferences and speeches given by the candidates
  • as found in FDCH documents (Federal Document Clearing House) in LexisNexis
• Primaries/Caucuses (Iowa, New Hampshire, and South Carolina)
  • the number of days before the Primary/Caucus
  • the number of days after the Primary/Caucus
• Day of Debate
  • whether it was the day of a debate
• Issue Specific
  • Security Alerts
  • Rulings on Gay Marriage
  • Labor Department Reports

  Methods

  Regression analysis

  Dependent Variable was analyzed with each Independent Variable

  • As long as candidate was still in race
    • Some candidates dropped out of the race a month before the data set was over. Not accounting for the candidate's absense would have skewed the data. So, we only ran models when candidate was still running.
  • Day in the campaign process and Primary/Caucus variables were not in the same model
    • All of these variables were very highly correlated and keeping them in the same model skewed the data. So, 4 separate models were run using only one of the variables at the same time.

   

By Carrie Rafer

© Davidson College, 2004, Department of Political Science, Davidson College, Davidson, NC 28035
Send comments, questions, and suggestions to Patrick Sellers
Created: 4/27/2004. Last updated: 5/2/2004.