{{ summaryHeading }} {{ summaryPrimary }} {{ summaryLine }} {{ badge.label }} {{ badge.value }}
Use a short protocol or appendix label.
One arm per line, for example Intervention | 1 | INT.
{{ formatInteger(participant_count) }}
This is a per-stratum count, not the overall trial total.
per stratum
Comma-separated values such as 4, 6 for a 1:1 design.
Example: Site: North, South. Leave blank for one unstratified schedule.
Store this seed with the reviewed schedule assumptions.
Choose whether block completeness or exact row count controls the last block.
Choose the schedule label that reviewers and downstream imports will see.
Letters, numbers, underscore, and hyphen; maximum 12 characters.
Generate an extra opaque code column or leave it empty.
Letters and numbers; maximum 8 characters.
SeqParticipantStratumBlockPositionAllocationCodeCopy
{{ row.sequence }}{{ row.participant_id }}{{ row.stratum }}{{ row.block_number }} / {{ row.block_size }}{{ row.position_in_block }}{{ row.allocation_display }}{{ row.randomization_code || '—' }}
StratumBlockSizeGenerated mixSeed branchStatusCopy
{{ row.stratum }}{{ row.block_number }}{{ row.block_size }}{{ row.allocation_mix }}{{ row.seed_branch }}{{ row.complete ? 'Complete' : 'Trimmed' }}
Random Number Generator Generate random integers or decimals from a stepped range with secure or seeded draws, weighted profiles, exclusions, and repeat controls.
Random Picker Pick random names or rows with reproducible seeds and optional weights, then review group limits and estimated inclusion chances before sharing the draw.
Event Schedule Generator Build an event run sheet from pasted session rows, keep fixed cues or continuous flow, and catch overlaps and tight handoffs before publishing.
Bonus Pool Allocation Calculator Allocate a bonus pool by eligibility and weighted policy, then reconcile floors and caps plus payroll rounding against the final balance.

Random assignment protects a study from letting enrollment order or investigator preference determine treatment. In a small trial, however, unrestricted randomization can produce a large temporary imbalance simply by chance. Permuted blocks limit that drift by constructing short groups with the planned treatment ratio and shuffling the order within each group.

For two equally weighted arms, a block of four contains two assignments to each arm. With a 2:1 ratio, a block of six contains four assignments to the first arm and two to the second. The block preserves the allocation mix when complete; the permutation hides the simple ordering of those assignments.

Stratification creates a separate randomization sequence for each combination of selected factors, such as site and age band. It can protect balance on a small number of important prognostic factors, but too many strata divide enrollment into sparse groups and can leave many incomplete blocks.

  • Fixed block size gives predictable local balance but may make future assignments easier to infer.
  • Several valid block sizes reduce that regularity while keeping each complete block compatible with the arm ratio.
  • Completing the final block preserves its ratio but may create more schedule rows than requested.
  • Trimming the final block keeps the exact row count but can disturb the ratio in that last block.

Randomization and allocation concealment are related but different. A schedule can be random yet still be compromised if staff can view future assignments. Participant labels, arm codes, seeds, and deterministic opaque-looking strings do not provide secure concealment. A live clinical workflow usually needs controlled access, an audit trail, and a process that reveals an assignment only after eligibility and enrollment are fixed.

Balance is also not proof that a design is statistically sound. Block sizes, ratios, strata, analysis methods, and handling of withdrawals should follow a prespecified protocol reviewed by the study statistician. The schedule is one design artifact within that larger plan.

How to Use This Tool:

Translate the approved allocation design into explicit arms, block sizes, strata, and a reproducibility seed.

  1. Enter each treatment arm on its own line as a label, ratio, and optional code, for example Intervention | 2 | INT. Use whole-number ratios.
  2. Set Participants per stratum. This is multiplied by the number of crossed strata, so it is not the overall trial total when stratification is used.
  3. Enter one or more Block sizes. Every size must be a whole-number multiple of the sum of the arm ratios.
  4. Add Stratification factors only when the protocol calls for them. Write one factor per line as Factor: category, category.
  5. Record the Reproducibility seed with the reviewed assumptions, then choose whether the Final block is completed or trimmed.
  6. Review Arm balance by stratum and Block audit. Resolve overrun, trimmed-block, and concealment warnings before releasing the schedule into an approved randomization process.

Interpreting Results:

Compare observed assignments with the ratio-derived expected count within each stratum. A nonzero delta can be normal when the requested count ends partway through a block. Check the block audit to learn whether the difference came from a trimmed block or from completing beyond the requested count.

The seed digest is a short review label for the entered seed, not a secret and not proof that a schedule was independently generated. Likewise, opaque randomization codes help label rows but do not prevent someone with the schedule or seed from learning future assignments.

Technical Details:

Each stratum receives an independent deterministic sequence derived from the shared seed and the stratum code. For every block, one allowed size is selected, the exact arm multiset is constructed from the requested ratio, and that multiset is shuffled.

Rule Core:

Rule Boundary Consequence
Arm design 2 to 8 unique arms; each ratio is an integer from 1 to 20 Repeated labels are combined before the final ratio is evaluated.
Block validity Ratio total to 1,000; every size is divisible by the ratio total Every complete block contains an exact multiple of the planned ratio.
Strata At most 64 crossed combinations Each combination gets its own seeded schedule.
Participants 1 to 5,000 per stratum and at most 20,000 generated assignments Complete final blocks are included in the assignment-limit check.
Final block Complete or trim Complete may overrun; trim may break the last block's ratio.

Formula Core:

Let wi be arm i's integer weight and W the sum of all arm weights. A valid block of size B gives that arm the following number of positions:

ni = B × wiW

Because B must be divisible by W, every ni is a whole number. For Ns generated rows in one stratum, the balance table compares the observed count with:

Ei = Ns × wiW

The displayed delta is observed minus expected. Expected counts may be fractional when a sequence stops between complete ratio cycles.

Mechanism:

The same seed and normalized design reproduce the same schedule. Each stratum branches from the seed, block sizes are selected from the allowed set, and arm positions are shuffled with a deterministic pseudo-random sequence. The method is suitable for repeatable generation and audit, not cryptographic unpredictability.

Limitations:

This is a planning and schedule-generation aid, not a validated clinical randomization service. It does not authenticate users, conceal future allocations, enforce eligibility, record assignment release, or replace statistical review. Do not place participant health information in labels or seeds, and keep the final schedule under the access controls required by the study protocol.

Worked Examples:

Two-to-one allocation across two sites

New care with weight 2 and usual care with weight 1 gives a ratio total of 3, so block size 3 is valid. With three requested participants per site and Site set to North and South, each complete stratum receives two New care assignments and one Usual care assignment. Six rows are generated overall, while the shuffled order differs by stratum and seed.

References: