Master’s Thesis Abstract by Karen E. Kopper (2003)

Meta-analysis, a relatively new statistical approach in forest science, was used to quantify the effectiveness of prescribed burning in reducing surface fuels in ponderosa pine (Pinus ponderosa) forests of the western United States. An aggregated data set was compiled from 8 published reports (locations) that contained data from 65 fire treatment units. Fixed-effects and mixed-effects models were used to analyze the data and develop inferences, with effects size measured by Hedges' d and the log response ratio. A mixed-effects model with the log response ratio was the only model that stabilized the variance of the grand mean, allowing the effects of fire to be robustly interpreted. The primary inference from this model was that prescribed fire significantly reduces quantities of surface fuels, but with no additional interpretation possible due to variability in site conditions, operational aspects of burning, and season of burning. Meta-analysis applications in natural resources merit further consideration because meta-analysis allows inferences to be developed across data sets reported by multiple authors. Meta-analysis is typically superior to Analysis of Variance (ANOVA), because it does not violate the assumption of equal variance across sites as is typical of ANOVA. A mixed-effects model will be most successful for data sets from multiple sites with high variability, although fixed-effects models should be considered for situations in which variability is relatively low. Standardized methodology, consistent measurement protocols, and complete reporting of both significant and nonsignificant results will greatly assist future synthesis efforts using meta-analysis.