qesR

Version française

qesR loads the Quebec Election Studies, with the panels and polls that accompanied them, into R: seven provincial elections from 1998 to 2022, each study by its code, from its original file. Their codebooks, question wording and a search across all studies ship with the package and work offline, in English and French. qesR also harmonizes 11 of the studies into one data frame, so that a question about 25 years of Quebec elections can be answered within each study and then compared.

In 2007 the youngest francophones were the cohort most likely to vote Yes; in 2022 they were the least likely

Party choice is far less tied to the referendum vote than in 2012, and left-right has not taken its place

Each estimate is made within one study, with that study’s weight where it has a validated one and a 95% confidence interval; nothing is pooled across studies. Get started goes from a study code to a weighted estimate, offline, on a small synthetic study that ships with qesR.

Examples

Each page takes a common belief about Quebec elections and sets it against the studies.

• From two parties to four: the vote fragmented, mainly among francophones, and within each sovereignty camp rather than across them.
• Are young Quebecers still the most pro-independence?: the youngest francophones were the most likely to vote Yes in 2007 and the least likely in 2022, mostly through change within cohorts.
• Two dimensions of competition: the national question structures the vote less than in 2012, and the left-right scale has not taken its place.
• Who votes?: the young report voting less at every election since 1998, and interest in politics accounts for only a small part of the gap.
• Are Quebec elections decided during the campaign?: the same respondents before and after the vote show that a campaign moves many voters but, on net, few votes.
• Do surveys miss the Liberals?: the Liberal vote was under-reported from 2007 to 2014, but the gap faded in 2018 and 2022.
• Recipes: the language divide in the Liberal vote: the code, step by step, from one harmonized variable to an estimate for every study.

Installation

# install.packages("remotes")
remotes::install_github("ThomasGareau/qesR")

Once qesR is accepted on CRAN, install.packages("qesR") will do the same. A first session (the lines marked “downloads” fetch studies from their deposits the first time; options(qesR.cache = "disk") keeps them):

library(qesR)
qes_studies()                         # the studies, offline
qes2018 <- get_qes("qes2018")         # downloads: one study, from its original file
qes_search("souverain|sovereign")     # a question, in every study, offline
h <- qes_harmonize(targets = "vote_choice")  # downloads: the vote, in six studies
d <- qes_decon()                      # downloads: every study in one flat file, relaxed

The harmonization rules are experimental and may change from one version of qesR to the next. Keep qes_provenance(h, level = "spec") with your results and pin the version of qesR you used.

Citing qesR

Work that uses qesR should cite the package and each study it draws on. qes_cite() writes both, with each dataset’s DOI:

qes_cite("qes2018")                   # qesR and the 2018 study
qes_cite("qes2018", style = "bibtex")

Citing qesR and the studies gives the citation of every study.

Licence

The code of qesR is under the MIT licence. The data are not part of the package: qesR downloads each study from its deposit on Borealis or the Harvard Dataverse. Most studies are released under CC0; the 2022 study, and the metadata qesR ships from it, are under CC BY-NC 4.0 (attribution, no commercial use). Details and the required attribution.