# Step 4 · Schema

The schema step arranges your arms and phases into the **randomisation sequences**
participants follow. It is shown as a grid: one column per phase across the trial
timeline, one row per sequence.

## Phases across the top

Add the phases that make up the trial in order, for example a baseline, one or more
intervention periods, washouts between them, and the post-trial debrief. Baseline and
debrief are single, fixed phases; intervention and washout phases can repeat.

## Sequences down the side

Each sequence is one ordering of arms:

- In a **parallel** trial you typically have one sequence per active/sham arm, each
  participant follows one arm throughout.
- In a **crossover** trial each sequence is a different order of the same arms, every
  participant passes through all of them, but in a randomised order.

For each intervention cell you pick which arm fills it; for each washout cell you pick
which washout arm (nothing or unmodulated) applies.

## How participants are placed

When a participant enrols, the portal assigns them to a sequence at random, weighted by
the completion targets you set in {doc}`3-treatment-phases`. The washout assignments in
this grid then determine, for each rest period, whether that participant listens to an
unmodulated sound or does not listen at all.

:::{note}
Weighted randomisation is defined per arm. In a parallel design each sequence is a
single arm, so the weighting is exact. Crossover designs are supported, but the
per-arm shares apply across the orderings; see {doc}`../enrolment` for the detail.
:::

Next: **{doc}`5-sessions`**.