Primary dependent variables were children's sucrose preference, sweet food liking score, pain tolerance and pain threshold. Sucrose preference and sweet food liking score were analyzed by separate 2 × 2 analyses of variance (ANOVAs), with FH (FHP, FHN) and depression category (PDEP, NDEP) as the between-subjects variables. Logarithmic transformation was performed on the pain tolerance and threshold data to approximate a normal distribution as these data were skewed positively [5]. The transformed data were then analyzed by separate two-way mixed ANOVAs with condition (sucrose, water) as the within-subjects factor and child's depression category (PDEP, NDEP) and FH (FHP, FHN) as the between-subjects factors. Additional Pearson's χ2 tests or one-way ANOVAs were conducted to determine whether the four groups differed on a variety of measures, including demographics and maternal characteristics. When significant, post-hoc tests [Fisher's least squares difference (LSD) tests for ANOVAs and partition squares for χ2 tests] were conducted.
