3. Results dicted emotional eating ( = −0.492, 95 % CI [−0.637; −0.343], p =
.001). These associations were found while controlling for mothers’ 3.1. Examining concordance between mothers’ and daughters’ BMI, which negatively predicted their body esteem ( = −0.488, 95 BMI, self-compassion, body esteem, and emotional eating % CI [−0.587; −0.357], p = .005), but did not significantly predict self-compassion ( = −0.093, 95 % CI [−0.245; 0.058], p = .208) nor Examination of intraclass correlations between mothers’ and emotional eating ( = 0.047, 95 % CI [−0.106; 0.214], p = .547). daughters’ scores revealed significant correlations for BMI (r = .522, Second, intrapersonal associations were examined for daugh- p < .001), self-compassion (r = .448, p < .001), body esteem (r = .245, ters. Daughters’ self-compassion positively predicted their body p < .05), and emotional eating (r = .522, p < .001), suggesting some esteem ( = 0.527, 95 % CI [0.412; 0.613], p = .003), which nega- degree of concordance within dyads. tively predicted emotional eating ( = −0.399, 95 % CI [−0.529; Paired t-tests were then conducted to compare mothers’ and −0.254], p = .003). These associations were found while controlling daughters’ levels of BMI, self-compassion, body esteem, and for daughters’ BMI, which negatively predicted their body esteem emotional eating. As can be seen in Table 1, average BMI was signif- ( = −0.439, 95 % CI [−0.549; −0.325], p = .002) but did not signifi- icantly higher for mothers (M = 26.53, SD = 4.45) than for daughters cantly predict self-compassion ( = 0.005, 95 % CI [−0.107; 0.109], (M = 24.47, SD = 4.31), t(1,190) = 5.71, p < .001. Mothers (M = 3.47, p = .943) nor emotional eating ( = 0.018, 95 % CI [−0.116; 0.143], SD = 0.60) also showed higher levels of self-compassion than their p = .787). daughters (M = 3.04, SD = 0.60), t(1,190) = 8.29, p < .001, and lower Third, interpersonal associations were examined. Mothers’ and levels (M = 2.05, SD = 0.86) of emotional eating than their daughters daughters’ BMI ( = 0.353, 95 % CI [0.224; 0.476], p = .002), self- (M = 2.30, SD = 0.86), t(1,190) = −3.46, p < 001. Mothers (M = 3.21, compassion ( = 0.288, 95 % CI [0.164; 0.408], p = .002), and SD = 0.84) and daughters (M = 3.25, SD = 0.81) did not report signif- emotional eating ( = 0.309, 95 % CI [0.184; 0.419], p = .002) were icantly different levels of body esteem, t(1,190) = −0.61, p = .55. positively related. Mothers’ and daughters’ body esteem were not significantly related ( = −0.016, 95 % CI [−0.123; 0.088], p = .799). 3.2. Dyadic analyses We also tested an alternative model with different directional- ities where the paths from mothers’ to daughters’ variables were Dyadic analyses using APIM were conducted to examine reversed. The chi-square value for this model was non-significant, whether mothers’ and daughters’ levels of self-compassion were 2 (df = 14, N = 191) = 21.01, p = .101, and other fit indices were linked to their own and each other’s emotional eating (see Table 3). adequate: NFI = .95, NNFI = .97, GFI = .97, CFI = .98, SRMR = .06, and The APIM regression model used to test this hypothesis contained RMSEA = .05 [.00; .09]. This model provided a satisfactory overall four predictors: mothers’ and daughters’ BMI and self-compassion. fit to the data, but the fit indices of the initial model were slightly Two actor effects and one partner effect were observed. Specifi- better. In addition, the Akaike’s Information Criterion of the initial cally, daughters’ emotional eating was significantly and negatively model (i.e., −12.70) was lower than that of the alternative model predicted by their own self-compassion ( = −.319, p < .001) as (i.e., −6.991), which suggests that the former has a better fit and well as by their mothers’ self-compassion ( = −.152, p < .05). Con- should be preferred. versely, mothers’ emotional eating was significantly and negatively predicted by their own self-compassion ( = −.288, p < .001) but 3.4. Indirect effects not by their daughters’ self-compassion ( = −.072, p = .317). Second, we tested whether mothers’ and daughters’ body Bootstrap analyses were performed to examine whether body esteem predicted their own and each other’s emotional eating (see esteem mediated the negative relation between self-compassion Table 4). The APIM regression model used to test this hypothesis and emotional eating. As can be seen in Table 5, the analyses contained four predictors: mothers’ and daughters’ BMI and body revealed that mothers’ body esteem significantly mediated the neg- esteem. Two actor effects and one partner effect were observed. ative relation between their self-compassion and emotional eating Specifically, daughters’ emotional eating was significantly and neg- (bootstrapping point estimate = −.259, 95 % CI [−.354 – −.171], atively predicted by their own body esteem ( = −.412, p < .001) p = .01). In addition, daughters’ body esteem significantly mediated as well as by their mothers’ body esteem ( = −.174, p < .05). Con- the negative relation between their self-compassion and emotional versely, mothers’ emotional eating was significantly and negatively eating (bootstrapping point estimate = −.210, 95 % CI [−.300 – predicted by their own body esteem ( = −.466, p < .001) but not −.127], p < .01). by their daughters’ body esteem ( = −.059, p = .405). Bootstrap analyses were also performed to examine whether mothers’ and daughters’ characteristics mediated the negative rela- 3.3. Main analyses tion between mothers’ self-compassion and daughters’ emotional eating. To distinguish the indirect effect through these two paths, In accordance with previous research on intergenerational the phantom model approach was used (see Macho & Ledermann, transmission of eating behaviors (Arroyo et al., 2017) and the above 2011). In other words, the specific indirect effect of mothers’ self- dyadic analyses showing that mothers’ characteristics (i.e., self- compassion on daughters’ emotional eating through mothers’ body compassion and body esteem) significantly predicted daughters’ esteem and emotional eating could be examined by constraining to eating behavior whereas daughters’ characteristics did not signif- zero the paths from mothers’ self-compassion to daughters’ emo- icantly predict their mothers’ eating behavior, we specified a path tional eating through daughters’ self-compassion and body esteem, analysis model where the direction of effects goes from mothers to and vice versa. As can be seen in Table 5, analyses revealed that daughters (Fig. 1). The tested model provided a satisfactory over- mothers’ body esteem and emotional eating significantly medi- all fit to the data. The chi-square value was non-significant, 2 (df = ated the negative relation between mothers’ self-compassion and 14, N = 191) = 15.30, p = .358, and other fit indices were adequate: daughters’ emotional eating (bootstrapping point estimate = −.089, NFI = .97, NNFI = .99, GFI = .98, CFI = .99, SRMR = .04, and RMSEA = .02 95 % CI [−.142 – −.050], p < .01). In addition, daughters’ self- [.00; .08]. compassion and body esteem significantly mediated the negative The standardized solution of the final model is presented in relation between mothers’ self-compassion and daughters’ emo- Fig. 2. First, intrapersonal associations were examined for mothers. tional eating (bootstrapping point estimate = −.066, 95 % CI [-.113 Mothers’ self-compassion positively predicted their body esteem – −.035], p < .01). These results suggest that daughters of self- ( = 0.525, 95 % CI [0.417; 0.634], p = .001), which negatively pre- compassionate mothers are less prone to emotional eating because
Marketing Research and Analysis-II (Application Oriented) Prof. Jogendra Kumar Nayak Department of Management Studies Indian Institute of Technology - Roorkee Lecture - 31 Non-Parametric Test - VII