Export Formats
qualcode.ai offers three export formats, each designed for different use cases. Choose the format that best matches your analysis workflow.
Statistical Format (SPSS/R)
The statistical format is optimized for quantitative analysis in tools like SPSS, R, Stata, or Python. It includes numeric codes and all metadata needed for statistical analysis.
Included Columns
| Column | Type | Description |
|---|---|---|
row_index | Integer | Original row number from your data file |
original_text | String | The original response text |
status | String | Processing status (agreed, disagreed, etc.) |
status_code | Integer | Numeric code for status (see below) |
final_codes | String | Final assigned category/categories |
rater_a_codes | String | Categories assigned by Rater A |
rater_b_codes | String | Categories assigned by Rater B |
rater_a_confidence | Float | Confidence score from Rater A (0-1) |
rater_b_confidence | Float | Confidence score from Rater B (0-1) |
rejected_reason | String | If rejected, the reason (empty, too_short, etc.) |
Status Codes
The status_code column contains numeric values for easier filtering and analysis:
| Code | Status | Meaning |
|---|---|---|
1 | Agreed | Both raters assigned the same category |
2 | Disagreed | Raters assigned different categories (not reconciled) |
3 | Reconciled | Human resolved a disagreement |
-1 | Rejected | Pre-filtered as invalid |
-2 | Unclassifiable | Could not be classified by either rater |
0 | Pending | Not yet processed |
Use statistical format when: You need to perform statistical analysis, calculate your own agreement metrics, or import into SPSS/R/Stata. This format includes ALL records, including rejected ones.
Detailed Format
The detailed format provides a complete audit trail of every classification decision. It's human-readable and includes all the information you need to review the coding process.
Included Columns
Same columns as Statistical format, but with human-readable labels and without numeric status codes. Ideal for:
- Quality assurance review
- Auditing classification decisions
- Sharing results with stakeholders who don't use statistical software
- Manual review in Excel or Google Sheets
Confidence scores help identify edge cases: Sort by confidence to find responses that were difficult to classify. Low confidence (even on agreed items) often indicates borderline cases worth reviewing.
Compact Format
The compact format is streamlined for simple analysis - just your original data with final classifications added. It excludes rejected records and most metadata.
Included Columns
| Column | Description |
|---|---|
row_index | Original row number |
original_text | The original response |
final_codes | Final assigned category/categories |
Key Differences
- Rejected records excluded: Only valid, classified responses are included
- No confidence scores: Simplified for end-user analysis
- No rater breakdown: Just the final result
Compact format hides detail: Use this only when you don't need to audit the classification process. For research requiring methodological transparency, use Statistical or Detailed format.
File Formats
Each export type is available in two file formats:
CSV (.csv)
- Universal compatibility with any software
- Plain text, easy to version control
- Smaller file size
- May have encoding issues with special characters
Excel (.xlsx)
- Opens directly in Excel, Google Sheets, etc.
- Preserves column formatting
- Includes summary statistics on a separate sheet
- Better handling of special characters and long text
Which Format Should I Use?
| Use Case | Recommended Format |
|---|---|
| Statistical analysis (SPSS, R, Stata) | Statistical (CSV) |
| Quality review / auditing | Detailed (Excel) |
| Sharing with non-technical stakeholders | Compact (Excel) |
| Academic research requiring transparency | Statistical or Detailed |
| Quick frequency analysis | Compact (CSV or Excel) |
| Importing to another system | Statistical (CSV) |
Next: Learn about pre-filtering options to automatically handle invalid responses before classification.