jamesmartherus / debates

Tidy presidential debate transcripts

debates: US Presidential Debate Transcripts

Presidential debates are an important opportunity for candidates to share their platforms. debates provides easy access to debate transcripts from Presidential, Vice Presidential, and primary candidate debates. The current version includes Presidential and Vice-Presidential debate transcripts starting in 1960, and for most debates from the 2012, 2016, and 2020 primary elections. debates includes one dataset, debate_transcripts, as a compact rda object. Once the package is installed and loaded, the dataset can be loaded using the data() function.

debate_transcripts includes speaker-level and debate-level data. Each row in debate_transcripts represents one statement. Along with the text of the statement, each row includes the speaker's name and an indicator variable that identifies whether or not the speaker is a candidate (as opposed to being a moderator, an announcer, or someone asking a question). Each row also indicates the date, location, and type of debate. To suggest additional fields, please open an issue.

For more information on how the dataset was compiled, see the file TUPD.pdf, also available here: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3611815

Installation

To install debates, use the install_github function from the devtools package:

library(devtools)
install_github("jamesmartherus/debates")

Alternatively, you can download transcripts.rda directly from the data folder.

Examples

library(debates)

data(debate_transcripts) #Load Transcript Data File

Acknowledgments

• Transcripts were gathered from a variety of sources including Rev.com, debates.org, and a variety of news sites.

debates's Issues

Some speakers are misclassified as `candidate==0`

James, thank you for putting this great resource together!

FYI, there appear to be 3 cases in 2020 primary debates when a speaker is classified as not being a candidate

• John Delaney
• "Bill deBlasio" (but "Bill DeBlasio" rows are classified correctly)
• Kamala Harris (but only in one dabate: 2020-10-08)

(I noticed this while running debate_transcripts %>% filter(election_year==2020) %>% group_by(candidate) %>% count(speaker) %>% View() )

John Edwards observations coded as presidential debates

Hi @jamesmartherus, thanks so much for your work in creating this amazing resource!

I think I found a small error: there are two observations for John Edwards coded as "Pres" in the type variable. For example, running the following:

library(debates)
data(debate_transcripts)
table(debate_transcripts$speaker[debate_transcripts$type=="Pres" & debate_transcripts$candidate==1], exclude = F)

Gets you this:

I was focusing only on the general election presidential debates data, so I'm not sure if there are also issues for John Edwards in other type values (e.g. in primary or VP debates data).

Unrelated question: just wanted to check, but the missing debates between 1960 and 1976 was intentional/something you were aware of, right? Again, just want to make sure it wasn't supposed to be in the data or anything.

Gerald Ford is misclassified

Hi,

First, thank you very much for making this awesome resource so easily usable! It was extremely valuable for us.

I found that all rows by president Gerald Ford are classified as candidate=0 and, if you can't find him, that's because his speaker value is incorrectly labeled as Gerlad Ford 😅 .

Kind regards