Merging
Add cases, add variables, or join on a key for two files. The agent can chain merges in one plan. Fuse is for waves.
There are three routes.
Merge Datasets — two files
Data → Data preparation → Merge Datasets joins the open file to one other dataset. Preview the unmatched rows before Merge datasets is enabled. The result opens as a new dataset.
Add cases stacks the second file under the first. Columns that exist on only one file stay, with blanks on the other side.
Add variables puts the second file’s columns next to the first when both have the same number of rows. Matching is by row order, not by a key. Names that collide take the file name as a suffix.
Join on keys matches on columns that exist on both files (_row_uid, _source_row_uids, and _weight are not offered as keys). Inner (default), Left, Right, or Outer.
The Merge Datasets dialog offers join-on-keys, add variables, and stack. If both files share an obvious ID column, the dialog defaults to a keyed join on that column. Add variables warns that rows are matched by position.
Unmatched report
Preview match shows:
- How many rows the result will have
- On a keyed join: unmatched counts on this file and the other file, with sample key values
- Duplicate key counts on each file, with a warning that duplicate keys multiply rows
- Columns only on this file, and only on the other file, when those lists are not empty
The merge button stays disabled until the preview matches the current selection. Changing the other file, the join type, or the keys clears the report. Saving a keyed merge still refuses duplicate keys unless you ask the agent to keep a many-to-many join.
The agent — chaining merges in one plan
When you attach several files and ask to merge and weight them, the agent proposes one plan: left-merge each extra file onto the first, then rake. You approve once. That is how three or more files are joined on a key without opening Merge Datasets for each pair.
Fuse — waves
Data → Data preparation → Fuse Waves (tracker files) or Fuse Datasets (separate respondent files). Select at least two datasets. The open file is pre-checked. Optional key / wave column.
With a key, files are outer-merged one after another. Without a key, rows are stacked and a _wave column is added.
Fuse opens the result the same way Merge Datasets does.