Editverse Free Calculator
Randomization Sequence Generator for Clinical Trials
Generate a reproducible allocation sequence for a trial — simple, block, or stratified randomisation with a fixed seed — and copy the exact R command behind the result. A full guide follows below. Validated against R 4.3.3.
In Simple Words
Randomisation assigns participants to treatment groups by chance, which is what removes selection bias and balances both known and unknown confounders across arms. A proper sequence is generated by a defined random process, not by alternation, dates, or anything a recruiter could predict.
Block randomisation keeps the group sizes balanced as recruitment proceeds by randomising within small blocks, and stratified randomisation balances important prognostic factors by randomising separately within each stratum. A recorded random seed makes the whole sequence reproducible and auditable.
When to Use Each Randomisation Method
Use simple randomisation for large trials where chance alone keeps groups balanced. Use block randomisation when you need balanced group sizes throughout recruitment, especially in smaller trials, and stratified randomisation when a few strong prognostic factors must be balanced across arms.
Whatever method you choose, the sequence must be concealed from those enrolling participants — generation and allocation concealment are distinct safeguards, and both are needed. Record the seed and parameters so the sequence is reproducible. Pair the design with an appropriate sample size. Unsure which scheme fits your trial? Our Test Selector can help.
set.seed({seed})
arms <- trimws(strsplit("{arms}", ",")[[1]]); blocks <- c({blocks}); n <- {n}
strata <- trimws(strsplit("{strata}", ",")[[1]]); scheme <- "{scheme}"
gen <- function(n) {
if (grepl("Simple", scheme)) return(arms[sample.int(length(arms), n, replace = TRUE)])
out <- character(0)
while (length(out) < n) {
b <- blocks[sample.int(length(blocks), 1)]
out <- c(out, sample(rep(arms, length.out = b)))
}
out[1:n]
}
if (grepl("Stratified", scheme)) lapply(setNames(strata, strata), function(s) gen(n)) else gen(n)User guide
When to use. Generate a reproducible allocation list (simple, permuted-block or stratified) with a recorded seed. The R command reproduces the exact list.
Assumptions. The list is held by someone independent of recruitment (allocation concealment); the seed and list are archived.
Common mistakes. Do not use a single fixed block size in unblinded trials (the last assignments become predictable). Simple randomisation can leave small trials unbalanced. Dynamic methods such as minimisation need dedicated software.
Worked example
Click Load example data above, then Run Analysis to see a fully worked result with figure, formula and interpretation for this test.
References
- Schulz KF, Grimes DA. Generation of allocation sequences in randomised trials. Lancet. 2002;359:515-519.
- Kim J, Shin W. How to do random allocation (randomization). Clin Orthop Surg. 2014;6(1):103-109.
Worked Examples
Example 1 — Block randomisation
For a two-arm trial using blocks of four, each block contains two A and two B assignments in random order. After every four participants the groups are exactly balanced, so an early stop still leaves comparable arms — useful in smaller trials.
Example 2 — Stratified randomisation
If disease severity strongly affects outcome, randomise separately within mild and severe strata, each with its own block sequence. This guarantees both arms have similar severity mixes, preventing chance imbalance on a key prognostic factor.
Assumptions to Check
A genuine random process. The sequence must come from a defined random generator with a recorded seed, never from alternation, birth dates, or admission order.
Allocation concealment. Those enrolling participants must not be able to foresee the next assignment; a generated sequence is only safe if it is concealed until allocation.
Sensible block size. Blocks that are too small and of fixed size become predictable near the end of each block; varying block sizes reduces this risk.
Common Mistakes
Confusing generation with concealment
A perfect random sequence can still be subverted if recruiters can see upcoming assignments. Generation and concealment are separate and both essential.
Using predictable fixed blocks
Small fixed blocks let staff guess the last assignment in each block. Use random or larger block sizes to keep allocation unpredictable.
Pseudo-randomisation by date or order
Assigning by alternation, day of week, or admission number is not randomisation and reintroduces selection bias.
Not recording the seed
Without the random seed and parameters, the sequence cannot be reproduced or audited, weakening the trial’s transparency.
Frequently Asked Questions
What is the difference between simple and block randomisation?
Simple randomisation assigns each participant independently; block randomisation balances group sizes by randomising within small blocks.
Why stratify?
Stratification balances strong prognostic factors across arms, preventing chance imbalance on variables that strongly affect the outcome.
What is allocation concealment?
It is keeping the upcoming assignment hidden from those enrolling participants, so the random sequence cannot be subverted at the point of allocation.
Does the result match R?
Yes. The generator uses standard R randomisation with a recorded seed and shows the command so the exact sequence can be reproduced.
Related Calculators
References
- Schulz KF, Grimes DA. Generation of allocation sequences in randomised trials: chance, not choice. Lancet. 2002;359(9305):515–519. PMID 11853818.
- Schulz KF, Grimes DA. Allocation concealment in randomised trials: defending against deciphering. Lancet. 2002;359(9306):614–618. PMID 11867132.
- Schulz KF, Altman DG, Moher D. CONSORT 2010 Statement: updated guidelines for reporting parallel group randomised trials. BMJ. 2010;340:c332.
Need This Analysis Done for Your Manuscript?
Our team builds reproducible randomisation schedules — simple, block, or stratified with concealed allocation — and documents the seed and parameters for full auditability.