R/vote_pres_national.R

In AmberlyDziesinski/ds: Files for Vanderbilt Data Science 1000

#'
#' National Presidential Election Polling Data
#'
#' Loads clean version of presidential general election polling data into the environment. This dataset includes polling data and elections outcomes for presidential elections in 2004, 2008, 2010, 2012, and 2018.
#' @usage data(vote_pres_national)
#'
#' @format
#' Data are structured as one observation per poll.
#' Variables include:
#' \describe{
#'   \item{year}{Year of election}
#'   \item{race}{Type of election race}
#'   \item{pollster}{Organization conducingthe poll and weighting}
#'   \item{end_date}{The last day the respondant was able to take the poll}
#'   \item{sample_size}{The number of respondents who completed the poll}
#'   \item{moe}{Reported margin of error for the surveys; calculated as the sample proportion of respondents supporting the Democratic candidate in unreported (commonly unreported for internet surveys) }
#'   \item{dem_poll}{The percent of respondents supporting the Democratic candidate in the poll}
#'   \item{rep_poll}{The percent of respondents supporting the Republican candidate in the poll}
#'   \item{ind_poll}{The percent of respondents supporting the (leading) independent candidatein the poll (if relevant)}
#'   \item{other_poll}{The percent of respondents supporting "Other" in the poll (if relevant)}
#'   \item{undecided}{The percent of respondents "Undecided" in the poll (if relevant)}
#'   \item{poll_margin}{Poll margin between candidates, calculated as (dem_poll-rep_poll)}
#'   \item{poll_vote_margin}{Difference between the poll margin and vote margin, calculated as ((dem_poll-rep_poll)-(dem_vote-rep_vote))}
#'   \item{error_on_margin}{Error on poll-vote margin, calculated as the absolute value of ((dem_poll-rep_poll)-(dem_vote-rep_vote))}
#'   \item{error_on_candidate}{The difference in error due to each candidate, calculated as ((dem_poll-rep_poll)-(dem_vote-rep_vote))/2}
#'   \item{dem_vote}{The certified vote supporting the Democratic candidate in the election}
#'   \item{rep_vote}{The certified vote supporting the Republican candidate in the election}
#'   \item{ind_vote}{The certified vote supporting the leading independent candidate in the election}
#'   \item{vote_margin}{Difference between Democrat and Republican vote, calculated as (dem_vote-rep_vote)}
#'   \item{winner_election}{Party of the candidate who won the election}
#'   \item{winner_poll}{Party of the candidate who was predicted to win in the poll}
#'   \item{winner_projected}{Indicator that the poll correctly predicted the winner: 1=yes, 0=no}
#'  }
#'
#' @details
#' These data were cleaned for the purpose of Data Science 1000 in the following way: variables with substantial missing data were removed (for example, if a variable was not reported in most years), and the margin of error was calculated where unreported using the sample proportion of respondents supporting the Democratic candidate. All years prior to 2004 were removed because they were missing sample size.
#'
#' @source
#' The historical data comes from three sources.  General election data (national presidential, statewide presidential, senate, and governor) prior to 2012 are from the National Council on Public Polls (http://www.ncpp.org/) website.  National and Statewide presidential polls in 2016 are from the 2016 AAPOR Task Force on Pre-Election polls, as are the primary polls from 2000, 2004, 2008, and 2012.  Data for the 2018 Midterm Elections was collected by a task force created in 2018 at the request of AAPOR Council President David Dutwin consisting of Evans Witt (PSRAI), Scott Clement (Washington Post) and Ariel Edwards-Levy (Huffington Post).
#'
#' @examples
#' # Run descriptive statistics on the data:
#' #hist(moe)
#' #table(winner_election)
#' #mean(sample_size)

"vote_pres_national"