Reading focus
Syllabus: How Can Couples, Families and Groups Become Settings for Change?
Distinguish assignment, delivery and observation, keeping the substantive question in view as the notation becomes technical.
Why you’re reading it. Understand why the number of people alone does not determine the strength of group-treatment evidence.
A question to consider. How would a trial with 100 people in five groups differ from one with 100 people in twenty groups?
Why this reading
Randomizing individuals does not make their subsequent experiences independent. This methodological review explains how group delivery can change the uncertainty in a treatment comparison and audits whether published studies recognized the problem.
Before you read
An intraclass correlation describes similarity within groups. The unit of randomization, unit of treatment delivery and unit of observation can differ. A false positive is an apparent treatment effect arising when the relevant null hypothesis is true.
Read for
- Distinguish individually randomized group treatment from cluster randomization.
- Connect the design distinctions and variance discussion with the review’s coding rules.
Consider afterwards
- Why might adding participants to existing groups help less than adding new groups?
- What information would you need to judge an analysis when only one condition meets in groups?
Editorial context
The audit covers 34 reports from six journals in 2002–2006. It documents a design and reporting problem, not a current prevalence estimate or proof that every affected treatment is ineffective. The equations formalize the central issue: shared experiences can make conventional precision estimates misleading.