Data Preparation

From PDF to Dataframe

Structuring the Data

We organize meta-analytic data in a specific “Long Format” where each row represents an Effect Size (ES).

The Dataframe Structure

Handling Moderators

Moderators are variables that might explain why effect sizes differ. They must be coded carefully.

• Categorical: E.g., Treatment Type (CBT vs. IPT). In R, these become factors.
• Continuous: E.g., Number of Sessions, Mean Age.

Missing Data

Meta-analysis is plagued by missing data (e.g., authors reporting “p < .05” but no exact values).

1. Contact Authors: The gold standard.
2. Imputation: Risky, only if necessary.
3. Conversion: Use compute.es to derive effects from p-values or t-stats.

Next Section: Converting Effects >