Skip to main content

Table 2 Effect of including sex as an interactive versus additive term when modeling the influence of diet on body size

From: Sex-specific plasticity and the nutritional geometry of insulin-signaling gene expression in Drosophila melanogaster

Modela DFb AICc BICc LL Model comparsions LRTd Pe
A: Tijkl = Sk + Ci + Ci2 + Pj + Pj2 + CiPj + Rl 9 − 1059.0 − 1017.3 538.48    
B: Tijkl = Ci + Ci2 + Sk (Pj + Pj2) + CiPj + Rl 11 − 1057.7 − 1006.8 539.85 A. v. B 2.73 0.256
C: Tijkl = Sk (Ci + Ci2) + Pj + Pj2 + CiPj + Rl 11 − 1066.7 − 1015.8 544.34 A. v. C 11.71 0.006
D: Tijkl = Sk (Ci + Ci2 + Pj + Pj2 + CiPj) + Rl 14 − 1067.7 − 1002.9 547.85 C. v. D 7.03 0.071
  1. aT is the body size, S is sex, C is carbohydrate, P is protein, R is replicate vial (random factor). The models differ by having sex as an interactive versus an additive factor
  2. bEstimated degrees of freedom for each model
  3. cAIC, BIC, log-likelihood (LL) calculated using ML fit
  4. dLikelihood ratio test (LRT) statistic is for model comparisons
  5. eP-value for LRT is calculated by parametric bootstrapping using ML fit. Significant P-values are shown in bold