Skip to main content

Table 4 Number of participants and clusters required in a multilevel study

From: Study protocol of a factorial trial ECHO: optimizing a group-based school intervention for children with emotional problems

 

Assumed ICC

Design effect

N indiv

n/cluster

N schools

A

0.00 – Base model

1

400

20

20

B

0.05 – Two-level model

1.95

800

20

40

  1. Design effect = 1 + (nc − 1)
  2. ICC intraclass correlation coefficient, n number of pupils
  3. *ICC (nc = average number of individuals in a cluster = 20)
Back to article page

BMC Psychology

ISSN: 2050-7283

Contact us