Publications reporting several experiments were dealt with by
assigning a specific code for each experiment. Each observation in
the meta-analysis corresponded to the mean or the Least Square
mean (LSmean) of each treatment group. Treatment groups were
then defined as the groups of treatments within an experiment that
changed only due to the dependent variable of interest (Sauvant,
Schmidely, & Daudin, 2005).
The selected publications were further required to provide an indication
of intra-experimental variation, expressed either as SEM
(or SE) or SD, for the mean of each treatment group. Publications in
which this information wasmissing were excluded. Using this criterion
restricted the bibliographic search yield to a total of 50 studies. Further
selection based on reported effects of fasting, lairage, transport and
the interaction between fasting and transport on the four analyzed
attributes narrowed this number down to 27 publications published
between 1985 and 2010 and included a total of 94 treatment groups
(see Appendix Table A1), representing 6526 pigs. In the sensitivity
analysis, the influence of each individual study was evaluated
by estimating the overall effect in the absence of one or more studies.