counterbalanceA FIELD GUIDE TO FAIR COMPARISONS / 010LOCAL WORKSPACE

REFERENCE STUDY / SYNTHETIC DATA

Two ways to teach one idea.

Does a worked example help more than a second explanation?

12units · 3 blocks · 2 groups

Browser memory only. No upload, account or autosave. Download a study file before leaving.

Synthetic reference loaded. Run a comparison or replace it with your own roster.Revision 0

FIXED COUNTS WITHIN EVERY BLOCK

Compare like with like.

A / Second explanationB / Worked example
01

Morning workshop

2 assigned to B

U0211
U0410
U0111
U0312
02

Afternoon workshop

2 assigned to B

U0613
U0815
U0515
U0714
03

Evening workshop

2 assigned to B

U109
U128
U0910
U119

Unit IDs are labels, not identities. Each row must represent the actual unit of randomization. This design is not suitable for analyzing individual rows when entire clusters were assigned together.

BEFORE THE COMPARISON

Write down the design.

Fixed group B counts

Each remaining unit goes to A. Counts lock once groups are assigned.

of 4
of 4
of 4

Random assignment is a design.
Not a label to add afterward.

Built for learning and transparent small studies—not clinical, regulatory or high-stakes decisions. No result proves representativeness or checks interference. See the design-based foundations and random-permutation testing principles.