Meta-Analysis

Client Context

A public health research team at a UK university needed a systematic review and meta-analysis of 22 studies examining the effectiveness of school-based mindfulness interventions on adolescent anxiety. The review was intended for publication in a high-impact journal.

The Challenge

The studies used diverse outcome measures (GAD-7, RCADS, SCARED), different comparison groups (waitlist, active control, treatment-as-usual), and varied follow-up periods. Several studies had high risk of bias, and one had incomplete outcome data. The research team needed a statistically rigorous approach that would withstand peer review and inform policy recommendations.

Our Approach

We conducted the meta-analysis using a random-effects model with restricted maximum-likelihood estimation. We standardised all outcomes to Hedges' g, assessed heterogeneity using I² and prediction intervals, and performed subgroup analyses by comparison type, intervention duration, and age group. Publication bias was evaluated through funnel plots, Egger's regression, and trim-and-fill analysis. Sensitivity analyses excluded high-risk-of-bias studies. The entire analysis was documented in an R Markdown report with reproducible code.

Results

The pooled effect was small but significant (Hedges' g = 0.28, 95% CI: 0.12 to 0.44), with substantial heterogeneity (I² = 62%). Subgroup analysis revealed that studies using active controls showed no significant effect, while waitlist-controlled studies showed moderate effects. Publication bias was detected but the trim-and-fill adjusted estimate remained significant. The research team submitted the manuscript to the Journal of Child Psychology and Psychiatry, where it was accepted with minor revisions.

Client Testimonial

"We came to AnalyticsScholar with a messy evidence base and they turned it into a clear, publishable narrative. The subgroup analyses were exactly what the reviewers asked for."

— Dr. Sarah K., Senior Lecturer in Public Health

[Explore More Case Studies → /case-studies/]

[Discuss Your Meta-Analysis → /contact/]