Sample Allocation Across Condition Arms

Survey experiments need enough respondents in each condition arm to support the intended analysis. Before publishing, decide how many participants should reach each arm, whether assignment should be equal or weighted, and whether quotas should cap any cells.

1. Step 1: Calculate expected n outside the editor. Decide target completes per arm, expected screen-outs, and any unequal allocation before configuring Flow conditions.

   Step 1: Plan allocation before configuring Flow

2. Step 2: Open Flow. Create or select the condition set that assigns respondents to the arms.

   Step 2: Open Flow and select condition set

3. Step 3: Set assignment percentages. Enter equal or weighted random assignment values and confirm they total 100 percent.

   Step 3: Set assignment percentages

4. Step 4: Add stratification if needed. Use a variable that appears before the split when assignment must be balanced within groups.

   Step 4: Configure stratification variable

5. Step 5: Open Deploy for quotas. Add quota targets for arms or cells when collection should stop or reroute after a count is reached.

   Step 5: Configure quotas in Deploy

6. Step 6: Export pilot data. Confirm condition assignments and quota-relevant fields appear in Export before launch.

   Step 6: Check condition fields in Export

In Flow, random assignment percentages determine expected allocation across arms. Equal splits are common for control and treatment comparisons. Unequal splits can be useful when one arm is more expensive, when a treatment is exploratory, or when the design intentionally oversamples a group.

Percentages alone only describe the target split. Whether the actual sample converges on that target exactly or only on average depends on the separate Balance arm sizes toggle: turned on, Domandata keeps a running count and draws each respondent from whichever arm is currently furthest below its target share (even/block assignment); turned off, each respondent is an independent draw that matches the target only on average, and a small or unevenly-dropped-out sample can end up lopsided. See How Randomization Works for the full explanation and every place this choice appears.

Stratified assignment helps balance arms within important groups. Quotas help manage target counts after responses start. Use both carefully: stratification shapes assignment probabilities, while quotas manage whether a target cell remains open.

• Each condition set has clearly named arms.
• Assignment percentages total 100 percent.
• Stratification variables appear before the split.
• Balance arm sizes is turned on for any design that needs guaranteed, not just expected, cell sizes.
• After a few real published-link responses, the live per-arm count shown next to each arm name (in the Branching panel's Arm names editor) matches what you expect.
• If a stratum's percentages do not total 100%, a Set equal split button appears next to that group so you can fix it in one click.
• Quota targets match the planned sample allocation.
• Exports include condition assignments for analysis.

• How Randomization Works
• Create Experimental Conditions
• Set Survey Quotas
• Export Research Survey Responses
• Preview and Pilot Test a Survey