Data packages

Data packages

Currently in alpha testing, ghosted allows you to de-identify interview transcripts in batches with one local app. It:

  • asks for input/output directories, known interviewer names, known participant names, and other terms to redact.

  • scans the input directory, reviews likely names detected with local rule-based matching, and writes redacted transcripts to the output directory.

  • includes options for output format, filename suffix, redaction token, blank lines between turns, console reports, common-name redaction, and the completion notice.

  • runs locally on your computer and does not use AI or remote services.

Access the repo here.

To cite this package: Shankar A, Canizares C, Cardozo F (2026). ghosted: Deidentify Transcript Files with a Local App. R package version 0.0.2.0, https://abiraahmi.github.io/ghosted/.

DedooseR is an R package that connects with Dedoose to support the analysis of qualitative data. It was built to help researchers streamline workflows, explore qualitative data flexibly, and conduct qualitative coding and analysis with rigor.

DedooseR currently has 8 key functions that allow you to:

  • clean_data: standardizes column names, keeps the highest ranked coder per transcript, drops range/weight columns, prefixes code variables with c_, and returns both a cleaned dataset and codebook

  • recode_themes: combines selected codes into a single logical column and updates the codebook.

  • view_excerpts: create an interactive, filterable datatable to view the excerpts behind each code

  • wordcloud: filters excerpts for a selected code, removes common stop words, and renders the result into a beautiful word cloud

  • create_code_summary to summarize code counts and the proportion of transcripts/media objects they come from, set a min count or proportion for the summary output and plot counts and proportions (or both!)

  • set_saturation: uses the output of create_code_summary to filter and visualize codes that meet minimum appearance targets

  • compare_saturation: builds on the summary table to check multiple threshold sets at once - useful when you want compare a strict bar versus a more liberal bar. You can also plot these different bars against each other

  • cooccurence: helps you see which codes travel together within the same transcript or media title, building both a matrix and a network plot

Access the package via CRAN.

To cite this package: Shankar, A., Cañizares, C., & Cardozo, F, Stockmans, S & Morris-Perez, P. (2025). DedooseR: Monitoring and Analyzing Dedoose Qualitative Data Exports (Version 2.0.0.1) CRAN. https://cran.r-project.org/web/packages/DedooseR/