The concept
An experiment compares outcomes under deliberately varied conditions. Random assignment helps distribute other influences across groups without choosing them one by one. This supports causal reasoning when the study is well conducted. It differs from random sampling, which concerns who enters the study and therefore helps determine whom the results may represent.
Worked example
A fictional library tests two reminder messages. It randomly assigns consenting participants to receive message A or B, sends both at the same time, and measures returns over the same period. If staff instead send A only to long-standing members and B to new members, differences could reflect membership history. Random assignment reduces that particular selection problem and makes the intended comparison clearer.
Limits and judgment
Chance can still produce imbalanced small groups. Attrition, people receiving the wrong message, or unequal measurement can weaken the comparison after assignment. Specify the outcome and analysis beforehand, inspect implementation, and report uncertainty. A result among participating library users may not generalize to people who never joined the study, even if the internal comparison was randomized correctly.
FROM UNDERSTANDING TO ACTION
Put it to work.
Design a fair comparison of two event reminders, naming assignment, outcome, timing, and one possible implementation failure.
RETRIEVE. REASON. REMEMBER.
Knowledge check.
Choose one answer for each question, then check your reasoning.
Sources & further reading.
Links support the core concepts or identify the book. Examples, exercises, and critical analysis are original Institute material. Source checks are not an independent expert review of every claim.
- NIST: Completely randomized designs ↗
Statistical core concept; all numerical scenarios and interpretations are original instructional examples. Source reviewed 2026-10-19.