Power Analysis in Social Work Intervention Research: Designing Cluster-Randomized Trials

In cluster-randomized trials (CRTs) of social work interventions, groups are assigned to treatment conditions. Conducting a power analysis ensures that enough groups are sampled for testing hypotheses. The power analysis method and inputs must be informed by the hypotheses, effects to be tested, and...

Full description

Saved in:
Bibliographic Details
Published in:Social work research 2009-03, Vol.33 (1), p.43-52
Main Authors: Rose, Roderick A., Bowen, Gary L.
Format: Article
Language:English
Subjects:
Clinical outcomes
Cluster Analysis
cluster-randomized trial
Control Groups
Data analysis
Design
Design analysis
Educational Research
Heuristics
Hypotheses
Hypothesis Testing
Intervention
intervention design
Methods
Modeling
Multilevel models
Outcomes of education
Power
Power analysis
Research Design
Research Methodology
Sampling
Social Work
Statistical Analysis
Statistical discrepancies
Statistical Significance
Statistical variance
Studies
Citations: Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:In cluster-randomized trials (CRTs) of social work interventions, groups are assigned to treatment conditions. Conducting a power analysis ensures that enough groups are sampled for testing hypotheses. The power analysis method and inputs must be informed by the hypotheses, effects to be tested, and the data analysis plans. The authors present a heuristic that can be combined with existing power analysis guidelines to select the appropriate power analysis method for a CRT given a well-defined research hypothesis. Using existing data, the authors demonstrate the heuristic in the context of a whole-school intervention program. The results indicate that the minimum number of schools needed to test the effect of the program can vary due to differences in hypotheses, making the heuristic an important research planning tool.
ISSN:1070-5309
1545-6838
DOI:10.1093/swr/33.1.43