2.4. Statistical analysis
The results are presented as means±S.E. Data were analyzed
using the GLM procedure of SAS (SAS, 1999). Basal glucose,
insulin, glucagon and cortisol concentrations were calculated as
the average of pre-injection values (the values at −15 min and
just before the injection (time 0)) in each animal. A randomized
block design (each animal was treated as a block) was used to
evaluate both the significance of the differences among the
samples obtained at the same time following the infusion
(comparison of the effectiveness between ghrelin and control
injection: Xij=μ+αi+βj+εij (Xij; sum of squares, μ; mean, αi;
effect of animal factor, βj; effect of chemical treatment, εij; error
of a main plot)) and the significance of the differences between
the basal value and the values obtained after each injection
(effects of each reagent administration: Xij=μ+αi+βj+εij (Xij;
sum of squares, μ; mean, αi; effect of animal factor, βj; effect of
sampling time, εij; error of a main plot)). The analyses were
made for each physiological stage separately. Significance was
accepted at a probability value b0.05. When Pb0.1, the result
was considered to show a tendency.