External variables (data)

Introduction

Importing external variables lets you append existing respondent data from outside Discover to your survey records for analysis. This is useful when you already have a database of respondent information and want to avoid asking for the same details again in a survey question.

Note that imported variables are only available in the Analysis area — for example, in crosstabs and segmentation — and cannot be used in the Create area for survey authoring.

External Variable Manager with imported variables in Discover

Data and file formatting

Imported data is treated as categorical, meaning all values are bucketed into categories. For example, if values 1, 2, 4, and 1 are imported across four respondents, the two respondents with a value of 1 are grouped together.

For the file to import correctly, follow these rules:

  • The first row must contain variable names, one per column.
  • The first column must contain the unique respondent identifier matching the ID type selected in the import dialog. If using Discover's internal record ID, this corresponds to the Record ID variable. If using a unique pass-in variable, use the variable name defined in the variable manager.
  • Every respondent in the survey should be included, identified by their unique ID with data for each external variable.
  • The file must contain 100,000 cells or fewer.

Troubleshooting

Additive imports

Importing is additive. If a file contains a variable name matching a previously imported variable, the new data is merged with the existing data rather than replacing it. A confirmation dialog appears in this case.

Examples:

  • Importing 1,000 respondent records for a variable, then another 1,000 different records, merges the two sets. If any record IDs overlap, the existing data for those records is overwritten by the new data.
  • Importing records with one set of values (1, 2, 3) then another set with different values (one, two, three) merges both sets rather than replacing the existing values.

Unmatched IDs

The unmatched IDs report shows the number of unmatched IDs found, specifying whether they were in the import file or the survey database.

To resolve, remove unmatched IDs from the import file and reupload. Alternatively, you can proceed without changes. The software will ignore missing IDs in the import file and categorize missing survey record IDs as “Other” in analysis.

Formatting errors

  • Missing unique identifier column: The import stops and an error identifies the missing column. Add the unique identifier column as described above and reimport.
  • Missing respondent ID: The import stops and an error identifies the row with the missing ID. Add the missing ID and reimport.
  • Duplicate respondent IDs: The import stops and an error identifies the rows with duplicate IDs. Remove or merge the duplicates and reimport.
  • Missing variable name: The import stops and an error identifies the column with the missing name. Add the variable name to the first row and reimport.
  • Oversized file: Files must contain 100,000 cells or fewer to avoid server timeout issues. Examples of supported sizes include 1 variable with 100,000 respondents, or 10 variables with 10,000 respondents. If your data exceeds this limit, split it into multiple files and import them separately. Since imports are additive, the data will merge correctly. For example, with 20,000 respondents and 10 variables, upload one file with the first 10,000 respondents and another with the second 10,000 for the same variables.

File reading error

If the file cannot be read due to a connection or technical issue, try importing again. If the issue persists, contact our support team.