Abstract #6

# 6
Meta-analysis: Part II (exercises).
Douglas M. Liebe*1, Robin R. White1, 1Virginia Tech, Blacksburg, VA.

This meta-analysis workshop will work through an example data set using a common analysis procedure to illustrate the usefulness of meta-analysis as a tool in ruminant nutrition research. The workshop will focus on use of R and R Studio for conducting meta-analysis. The example data set includes literature reporting how microbial N outflows from the rumen are influenced by dietary nutrient intakes, marker type, sampling location, rumen pH and rumen volatile fatty acid or ammonia concentrations. The workshop will walk through a multi-step process used to evaluate data for common errors, correct standard errors, and derive models using a backward, stepwise regression procedure. Packages reviewed will include those required to read in data (xlsx, XLConnect, googlesheets), those required to handle data (dplyr, reshape2), those required to visualize data (ggplot2), and those required to fit linear mixed effect models (nlme, lme4, lmerTest). The workshop will walk participants through how data should be structured in text files or spreadsheets; how packages can be used to read data in from these external formats; how data can be handled and queried once read into R or R Studio; how data can be visualized using the ggplot2 package; and how models can be derived using linear mixed-effects models weighted for the inverse of study standard errors. At the end of the workshop, participants should be able to (1) organize data for use in a meta-analysis; (2) read data into R from a variety of formats; (3) visualize data distributions for assessment of common data entry errors; (4) calculate weights for use in meta-analysis; (5) derive a model using a multi-step backward elimination approach; and (6) evaluate the fit of a model using common fit statistics.

Key Words: meta-analysis, nutrition, regression