Replication Data for "Using Multilingual Language Technology to Classify Open-Ended Survey Responses: Conceptions of Democracy in a Cross-Cultural Survey Setting"

We include 2 zip files, the main archive (multilingual_survey_classification.zip) containing the code and derived data files. The trained models are too big to fit into a single archive without exceeding the individual file upload limit. We add another archive with the two multilingual models trained on all 5 languages for the study (multilingual_models.zip) the structure of which corresponds to the subdirectory that will be created in models/ during replication. The model archive reflects the repository structure. Importantly, the model files are not needed for replication, as they are being created during the full replication. Zip files for the monolingual and cross-lingual experiments can be made available - each about 19 GB in size. We originally planned to include these in the submission, but ran into issues with the file upload both through HTTP and Dropbox upload. In addition we also include additional data on the survey collected for our research.