When converting from CSV to SPSS (.sav) format, you can provide additional metadata to preserve variable definitions and value labels.

The metadata file should be a JSON file with the following structure:

{
  "column_labels": {
    "age": "Age of respondent",
    "gender": "Gender of respondent",
    "income": "Annual household income"
  },
  "value_labels": {
    "gender": {
      "1": "Male",
      "2": "Female",
      "3": "Non-binary"
    },
    "education": {
      "1": "Less than high school",
      "2": "High school",
      "3": "Some college",
      "4": "Bachelor's degree",
      "5": "Graduate degree"
    }
  }
}

Where:

• column_labels: Maps column names to their descriptions
• value_labels: Maps column names to a dictionary of value-label pairs

This is the same format generated when converting from SPSS to JSON, so you can use the JSON output as a template.