Meta-analysis of literature is used in animal nutrition research to gain a more comprehensive understanding of the response being studied. Using weighted, mixed-effects, regression, most meta-analytical data sets can be evaluated. In these analyses, data are first gathered using clearly defined search criteria. Collected data should include the response variables of interested, standard errors reported, and explanatory variables under consideration. Once data are compiled, they should be checked for outliers and possible errors when transferring data. To remove individual study statistical analysis effects, data should be partitioned into mixed- and fixed-effect analyses and standardized. When data are clean and errors are standardized, backward, stepwise regression with fixed-effects for variables of interest and random-effects for things like study and location can be conducted. Variables should be removed from the model according to a predetermined cutoff, usually a P-value of 0.05. When all variables in the model are significant, removed variables can be iteratively re-tested to ensure that factors were not removed due to model instability. Correlation between factors can then be assessed using variance inflation factors (VIF). Parameters with a VIF above 10 should only be kept when parameters are correlated by calculation. When variables are correlated, the variable with highest VIF can be removed. Backward, stepwise regression, elimination at a significance cutoff, and elimination based on correlation should be iterated until model has only significant parameters and is not highly correlated. This procedure provides a framework for most animal nutrition meta-analyses, but may require some adjustments based on available data.