Introduction
Eating disorders (ED) such as anorexia nervosa (AN) and bulimia nervosa (BN) are severe mental disorders that frequently develop in females during adolescence (American Psychiatric Association, 2013; Herpertz-Dahlmann, 2015; van Eeden et al., 2021). The DSM-5 criteria for both AN and BN encompass symptoms such as vomiting/dieting, although another core feature is a disturbed body image (American Psychiatric Association, 2013), defined as difficulties in body-related perception (e.g., overestimation of weight/body size), behavior (e.g., body avoidance/checking), and cognition and affect (e.g., distorted thinking/feelings/evaluation regarding one’s body; Thompson et al., 1999). The most prominent manifestation of a disturbed body image is body dissatisfaction, described as negative thoughts and feelings about one’s body (Grogan, 2022). Body dissatisfaction is reported by female adolescents and women with and without ED (Coker & Abraham, 2014; Laporta-Herrero et al., 2018; Wang et al., 2019) and is known as a risk factor for ED (Rohde et al., 2015).
According to the tripartite influence model of body image and eating disturbance (Thompson et al., 1999), media, peers, and parents are important sources for the development and maintenance of body dissatisfaction. In particular, a direct and indirect transmission of parental body image and dysfunctional eating to the offspring is assumed (Abraczinskas et al., 2012). While direct transmission encompass parental comments about their children’s weight and eating behaviors, e.g., criticism of weight/shape/appearance and verbal encouragement to diet, indirect transmission include modeling behaviors such as imitation of parental body-related comments, behaviors, and attitudes, e.g., expressed body dissatisfaction and dieting (Abraczinskas et al., 2012; Crowther et al., 2015; Rodgers & Chabrol, 2009). Parents’ influence on their daughters has been found to be substantial (Rodgers & Chabrol, 2009), with research demonstrating that mothers have a greater influence than fathers on the development of body dissatisfaction and problematic eating behavior in girls and female adolescents (Kluck, 2010; Neumark-Sztainer et al., 2010). In more detail, concerning direct influences, maternal feedback, criticism, and encouragement to diet were found to be strongly associated with greater weight control behavior such as taking dieting pills or deliberate vomiting (Bauer et al., 2013; Neumark-Sztainer et al., 2010), higher body dissatisfaction (Cooley et al., 2008; Story et al., 2022), and more eating problems/binge eating (Cooley et al., 2008; Neumark-Sztainer et al., 2010) in daughters. Concerning indirect influences, maternal modeling of weight loss, talking about their own weight/size/shape, and eating behavior have been associated with more restraint eating (Snoek et al., 2009; Vincent & McCabe, 2000), binge eating, unhealthy/extreme weight control behavior (Neumark-Sztainer et al., 2010), higher drive for thinness (Wertheim, 2002), and higher body dissatisfaction (Cooley et al., 2008; Wertheim, 2002) in daughters. Besides these cross-sectional studies, maternal influences have also been demonstrated in longitudinal studies (Rodgers et al., 2021; Van den Berg et al., 2010). For instance, maternal weight concerns and related behaviors during daughters’ early adolescence significantly correlated with higher body dissatisfaction and more weight loss attempts in daughters five years later (Van den Berg et al., 2010). Accordingly, research suggests that mothers have a significant direct and indirect influence on their daughters’ body image.
While previous research largely focused on questionnaire data, a laboratory study in female adolescents and their mothers found evidence for the familial transmission of attentional biases (Bauer et al., 2017), which according to the cognitive-behavioral model of eating disorders (Williamson et al., 2004) are known to play a crucial role in the development and maintenance of eating disorders and body image disturbance. In this eye-tracking study (Bauer et al., 2017), daughters and their mothers were shown pictures of their own and an unknown peer’s body, while their gaze patterns were recorded. Additionally, mothers also viewed their daughter’s and the adolescent pictures. In line with the assumption of direct transmission, daughters’ viewing pattern of their own and the adolescent body correlated positively with their mothers’ viewing pattern of the same bodies. Moreover, the more the daughters received negative maternal feedback, the less self-serving were their viewing patterns. In line with the assumption of indirect transmission, daughters’ viewing pattern of their own and an adolescent peer’s body correlated positively with their mothers’ viewing pattern of their own and an adult peer’s body (Bauer et al., 2017). While these results demonstrate familial transmission beyond questionnaire data, to our knowledge, no other study has investigated familial transmission of other body image-related biases in females.
Besides attentional biases, evaluation biases have also been shown to be relevant to body image disturbance (Williamson et al., 2004), with recent studies demonstrating that women with and without eating disorders, as well as men, apply double standards (DS) in body evaluation (Quittkat et al., 2023; Voges et al., 2018, 2019, 2022). DS are defined as an evaluation that differentiates between two individuals even though both show similar extents of a certain attribute (Foschi, 1996, 2000). DS have already been reported in other fields, such as intelligence (Szymanowicz & Furnham, 2013) and perceived competence (Foschi, 1996). More recent research has also found DS in body evaluation, with a presented body being evaluated differently depending on its face (Quittkat et al., 2023; Voges et al., 2018, 2022). In these studies, participants evaluated pictures of artificial and real bodies presented with their own and another person’s face. Women rated an overweight build as having more body fat, as less attractive, and showed more negative emotions and higher arousal when it was presented with their own compared to another face (Quittkat et al., 2023; Voges et al., 2018, 2019, 2022). Furthermore, when these pictures depicted real bodies (Voges et al., 2022), women also reported higher negative emotions and higher arousal, and showed lower attractiveness ratings when viewing their own as opposed to another face on their own body, a thin body, and an average-weight body. Such self-deprecating DS have been found in women with AN and BN (Voges et al., 2018), women with binge-eating disorder (Quittkat et al., 2023), and women without any ED (Quittkat et al., 2023; Voges et al., 2019, 2022). To date, DS have not been examined in female adolescents, despite their particular vulnerability to developing ED (American Psychiatric Association, 2013) and the fact that comparisons with others begin in early adolescence (Sebastian et al., 2008).
The present study therefore examined DS and their potential familial transmission in female adolescents and their mothers. Daughter-mother dyads were presented with pictures of their own body and three different age-matched builds depicted with their own and a peer’s face, and rated them regarding arousal, valence, body attractiveness, and body fat. The mothers additionally provided ratings for their daughters’ body and the three other adolescent builds.
First, in line with previous studies (Voges et al., 2019, 2022), we hypothesized that daughters and mothers would show self-deprecating DS for their own and other bodies, meaning that they would rate a body presented with their own face more negatively than the same body presented with an unknown face (question 1). Potential differences between mothers and daughters were investigated from an exploratory perspective.
Second, we aimed to investigate whether mothers show DS for pictures of their own daughters. As research on DS has not previously investigated families, it was unclear whether mothers would evaluate their daughter’s body more positively when it was presented with their daughter’s face compared to with another face, or whether they would show more critical DS for their own child’s body. Therefore, mothers’ DS for their daughter’s body and DS for the other adolescent bodies were examined from an exploratory perspective (question 2).
Third, in line with the assumptions of Bauer et al. (2017) regarding direct familial transmission, we hypothesized that daughters’ ratings of their own and the other adolescent builds presented with their own and another face would be reflected in their mothers’ ratings of their daughters’ body and the other adolescent builds presented with their daughters’ face and another face. Accordingly, we expected daughters’ DS for their own and the adolescent peer builds to correlate positively with their mothers’ DS for their daughters’ and the adolescent peer builds (question 3). Moreover, in line with previous research on direct transmission, we assumed an association of maternal feedback with daughters’ evaluation of the builds presented with their own and another face. Thus, we expected that negative maternal feedback would be positively related to daughters’ DS of arousal and body fat and to daughters’ scores on measures of eating pathology/body image (question 4) and negatively related to daughters’ DS of valence and body attractiveness for their own and the three adolescent peer builds (question 5). Correlation analyses for positive maternal feedback were conducted from an exploratory perspective. With regard to indirect transmission, consistent with Bauer et al. (2017), we assumed that mothers’ evaluation of their own and the adult builds would be reflected in daughters’ evaluation of their own and the adolescent builds. Consequently, we expected daughters’ DS for their own and the adolescent peer builds to be positively correlated with mothers’ DS for their own and the adult peer builds (question 6). Finally, in line with previous research (Cooley et al., 2008; Neumark-Sztainer et al., 2010), we hypothesized positive correlations between daughters’ and mothers’ scores on measures of eating pathology and body image (question 7).
Methods
Participants
Participants were recruited through flyers/advertisements, press releases, regional newspapers, social media, and email lists (i.e. schools, university). For all participants, the inclusion criterion was no acute suicidal/self-harm behavior. For daughters as the index participants (Bauer et al., 2017), further inclusion criteria were age between 13 and 18 years, the absence of any mental disorder, and a body mass index (BMI) between their 10th and 97th age percentiles (Kromeyer-Hauschild et al., 2001). The latter criterion was applied to obtain a sample that was as representative as possible but also to avoid including participants with severe underweight/overweight in the sample. N = 46 daughter-mother dyads provided informed consent, of which n = 2 dyads were excluded due to a mental disorder of the daughter. Therefore, n = 44 daughter-mother dyads fully participated; of these, n = 6 mothers reported one or more current mental disorder (Fig. 1, supporting information).
Measures and Materials
Questionnaires and Body Ratings
To assess the extent of body image and eating disorder pathology, participants completed the German-language version of the Eating Disorder Examination Questionnaire (EDE-Q; Hilbert & Tuschen-Caffier, 2016), comprising four subscales: Restraint (R), Eating Concern (EC), Weight Concern (WC), and Shape Concern (SC). Items are rated on a 7-point Likert scale (0 = no days/none of the times/not at all to 6 = every day/every time/markedly). High internal consistencies for the subscales have been previously reported (0.80 ≤ α ≤ 0.90; Hilbert et al., 2012). In the present study, Cronbach’s α were αdaughters = 0.66/αmothers = 0.77 for R, αdaughters = 0.19 /αmothers = 0.68 for EC, α = .82daughters/αmothers = 0.63 for WC, and αdaughters = 0.92 /αmothers = 0.84 for SC. Additionally, we applied the subscales Body Dissatisfaction (BD) and Drive for Thinness (DT) from the German-language version of the Eating Disorder Inventory – 2 (EDI-2; Paul & Thiel, 2005) to examine participants’ dissatisfaction with their bodies and desire for thinness. Items on both subscales are rated on a 7-point Likert scale (0 = never to 6 = always), and both subscales have shown good internal consistencies (0.86 ≤ α ≤ 0.89; Paul & Thiel, 2005). Internal consistencies in the present study were likewise high (BD: αdaughters = 0.87/αmothers = 0.89; DT: αdaughters = 0.81/αmothers = 0.88). Furthermore, participants completed a German-language version of the Body Appreciation Scale – 2 (BAS-2, self-translation by our department), which measures acceptance of, favorable opinions toward, and respect for one’s body. The BAS-2 has shown high internal consistency in previous research (0.85 ≤ α ≤ 0.94; Quittkat et al., 2019; Voges et al., 2022); internal consistency in the present study was likewise high (αdaughters = 0.88 and αmothers = 0.95). To measure the perceived negative and positive body-related feedback from their mothers, only daughters completed adapted versions of the subscales Negative Weight and Shape (NWS, nine items) and Positive Weight and Shape (PWS, five items) from the German-language version of the Verbal Commentary on Physical Appearance Scale (VCOPAS; Bauer et al., 2017), with their mother as the source of commentary. This adaptation has previously shown good internal consistency (0.70 ≤ α ≤ 0.84; Bauer et al., 2017); internal consistency in the present study was also high (NWS: α = 0.80; PWS: α = 0.72). Finally, all participants rated the body stimuli according to arousal (1 = very calm to 9 = very arousing), valence (1 = very negative to 9 = very positive), body attractiveness (1 = very unattractive to 9 = very attractive), and body fat (1 = very little body fat to 9 = very much body fat) (Voges et al., 2019).
Stimuli
Standardized stimuli comprised pictures of three female adolescents and three women wearing standardized grey underwear in four poses against a white background (Fig. 1). Since the age percentiles of adolescents must be taken into account when calculating their BMI, the adolescent models corresponded broadly to the respective BMI-for-age criteria for thin, average-weight, and overweight (Kromeyer-Hauschild et al., 2001). For the adult models, the BMI criteria of the World Health Organization for a thin, an average-weight, and an overweight build were applied (World Health Organization, 2010; Table 1, supporting information). All models provided written informed consent for their pictures without their face to be used as stimulus material and received €50 as reimbursement. The models’ bodies were cut out using GIMP (2.10.12), matched with an adolescent (46_F13_Neutral; Meuwissen et al., 2017) or an adult face (136_03; DeBruine & Jones, 2021), and colored gray using MatLab (R2020a). Small discrepancies between body and face were adjusted in GIMP using Clone, Heal and Smudge. The same procedure was applied to the participants’ pictures. Furthermore, each participant’s face was merged with the three age-appropriate bodies, and the age-appropriate face was placed on the participant’s body. In total, this led to a stimulus set of 32 pictures for daughters and 32 pictures for mothers, each including 16 pictures of all bodies with the participant’s own face (i.e. four bodies in four poses) and 16 pictures of all bodies with the unknown face (i.e. four bodies in four poses).
Table 1
Descriptive statistics and group comparisons of daughters and their mothers regarding demographic characteristics and body- and eating-related symptoms
Daughters (n = 44) | Mothers (n = 44) | Group comparison | |||||
|---|---|---|---|---|---|---|---|
M | SD | M | SD | t | p | d | |
Age (years) | 16.20 | 1.74 | 47.09 | 6.08 | -35.03 | < 0.001 | -5.28 |
Height (m) | 1.65 | 0.05 | 1.67 | 0.06 | -1.54 | 0.139 | -0.23 |
Weight (kg) | 60.09 | 7.35 | 70.20 | 9.49 | -6.67 | < 0.001 | -1.01 |
BMI (kg/m2) | 21.99 | 2.51 | 25.26 | 3.28 | -6.03 | < 0.001 | -0.91 |
BMI percentile | 62.32 | 24.95 | 49.31 | 24.16 | 2.62 | 0.010 | 0.395 |
BMI-SDSLMS | 0.63 | 1.92 | -0.04 | 0.75 | 2.33 | 0.077 | 0.352 |
EDE-Q | |||||||
Restraint | 0.39 | 0.59 | 0.90 | 1.15 | -2.51 | 0.018 | -0.37 |
Eating Concern1 | 0.36 | 0.42 | 0.26 | 0.35 | - | - | - |
Weight Concern | 0.84 | 1.03 | 0.92 | 0.83 | -0.42 | 0.662 | -0.06 |
Shape Concern | 1.07 | 1.14 | 1.20 | 1.06 | -0.56 | 0.563 | -0.08 |
EDI-2 | |||||||
Body Dissatisfaction | 21.11 | 8.54 | 27.57 | 10.47 | -3.49 | 0.002 | -0.53 |
Drive for Thinness | 13.71 | 5.62 | 16.14 | 6.85 | -1.91 | 0.061 | -0.29 |
BAS-2 | 3.94 | 0.65 | 3.78 | 0.82 | 1.22 | 0.221 | 0.181 |
VCOPAS | |||||||
Negative Weight and Shape | 1.47 | 0.53 | - | - | - | - | - |
Positive Weight and Shape | 2.18 | 0.85 | - | - | - | - | - |
Procedure
The procedure was in line with previous studies (Bauer et al., 2017; Voges et al., 2022). Interested daughter-mother dyads were informed about the study in a telephone call and in writing, and provided written informed consent. Next, they separately underwent an online (using the video system RED Medical) diagnostic interview based on DSM-5 criteria (American Psychiatric Association, 2013) with a licensed psychotherapist. The Diagnostic Interview for Mental Disorders in Children and Adolescents (Kinder-DIPS; Margraf et al., 2017; Schneider et al., 2017) was administered with the daughters and the Diagnostic Interview for Mental Disorders (DIPS; Margraf et al., 2017a, b) with the mothers. Separate feedback was subsequently provided and in the case of a diagnosis, participants received contacts to begin psychotherapy.
The laboratory assessment followed a hygiene protocol due to the COVID-19 pandemic (e.g., wearing FFP-2 masks). In the absence of their mothers, daughters were measured and the pictures were taken as described above (Fig. 1), without their mask for the first pose. Next, daughters completed the questionnaires via Unipark (QuestbackGmbH; Cologne, Germany) while their pictures were prepared (see Stimuli) and mothers were measured and photographed. Subsequently, daughters were presented with the 32 adolescent stimuli in a randomized order for five seconds in a 1920 × 1080 px format using E-Prime 3.0 (Psychology Software Tools, Inc.; Sharpsburg, PA, United States). Daughters were told that they would see different bodies with different faces, including their own body and face, and should evaluate each body, independently of its face, regarding arousal, valence, body attractiveness, and body fat. The same procedure was followed with the mothers, who were additionally presented with the 32 adolescent stimuli. Finally, daughter-mother dyads were debriefed and received €60 reimbursement.
Fig. 1
Illustration of the stimuli, as presented to the participants. Four different poses of adolescent and adult thin, average-weight and overweight build as well as of the participant’s body were taken, represented here by the adult average-weight build. Double standard scores were computed by calculating the body evaluation of the build with the participant’s own face (illustrated by the face of the first author) minus the body evaluation with an age-appropriate unknown face (from a database)
×
![]()
Statistical Analyses
Descriptive and correlation analyses were performed using SPSS (27, IBM; Armonk, NY, USA). The analysis of the scale Eating Concerns revealed a very low internal consistency and was therefore removed from all analyses. Multilevel modeling was conducted in R (4.2.2) (R Core Team, 2022) and RStudio (2023.06.0 Build 421; RStudio Team, 2022), with the packages nlme (version 3.1–160; Pinheiro et al., 2022), emmeans (version 1.8.4-1; Lenth, 2023), car (version 3.1-1; Fox & Weisberg, 2019) and tidyverse (version 1.3.2; Wickham et al., 2019). Differences between body ratings with one’s own and the other face were calculated by subtracting the evaluation of the build with the other face from the evaluation of the build with the one’s own face and used as the DS score. Therefore, for valence and body attractiveness, a negative DS score indicated a self-deprecating evaluation, whereas for arousal and body fat, a positive DS score indicated a self-deprecating evaluation. Following this procedure, a further DS score was calculated for mothers’ evaluation of their daughters’ and the adolescent bodies. To determine the extent of DS for each build and each group, 95% confidence intervals were applied, indicating a substantial DS if differing significantly from zero. To investigate differences in DS between daughters and mothers, multilevel models with restricted maximum likelihood estimators were calculated for each DS of arousal, valence, body attractiveness, and body fat as dependent variables. To account for repeated measures, participant ID and daughter-mother dyads served as a random effect, while group (daughters, mothers), build (thin, average-weight, overweight, own), and their interaction served as fixed effects. Likelihood ratio tests and information criteria (AIC/BIC) revealed that in most cases, the robust models, allowing for heteroscedasticity and all random effects, were most appropriate. In the case of no or little improvement, robust models are nonetheless reported. As effect sizes, Hedges’ gav (within-group effects) and Hedges’ g (between-group effects) were calculated (Lakens, 2013). An α level of 0.05 was used for all tests and corrected for multiple comparisons using Bonferroni-Holm correction. All multilevel analyses were also performed as full Bayesian parameter estimation in R (see supporting information). To analyze direct and indirect familial transmission, Pearson correlations with bootstrapping (n = 1000, BCa method) were conducted.
Results
Descriptive statistics are displayed in Table 1. Mothers were significantly older and had a higher BMI/weight than their daughters. Daughters reported lower levels of restraint eating and body dissatisfaction than their mothers. The results of the multilevel models are displayed in Table 2, DS scores and the results of the post hoc t-tests are presented in Table 3 and Fig. 2. Hereinafter, the results of daughters’ and mothers’ body ratings, the additional maternal ratings of adolescent bodies, and familial transmission are presented.
Table 2
Results of the multilevel analyses with group, build, and the interaction group × build for each double standard score
Main effect Group | Main effect Build | Interaction Group × Build | |||||||
|---|---|---|---|---|---|---|---|---|---|
F | df | p | F | df | p | F | df | p | |
Model I | |||||||||
DS arousal | 0.95 | 1, 43 | 0.336 | 1.10 | 3, 258 | 0.351 | 0.58 | 3, 258 | 0.630 |
DS valence | 0.05 | 1, 43 | 0.823 | 1.54 | 3, 258 | 0.204 | 1.71 | 3, 258 | 0.166 |
DS body attractiveness | 2.89 | 1, 43 | 0.096 | 5.15 | 3, 258 | 0.002 | 2.69 | 3, 258 | 0.047 |
DS body fat | 1.37 | 1, 43 | 0.248 | 7.67 | 3, 258 | < 0.001 | 3.18 | 3, 258 | 0.025 |
Model II | |||||||||
DS arousal | 1.62 | 1,43 | 0.211 | 0.46 | 3, 258 | 0.713 | 0.46 | 3, 258 | 0.710 |
DS valence | 1.03 | 1,43 | 0.315 | 4.18 | 3, 258 | 0.007 | 2.24 | 3, 258 | 0.084 |
DS body attractiveness | 5.71 | 1,43 | 0.021 | 4.58 | 3, 258 | 0.004 | 3.26 | 3, 258 | 0.022 |
DS body fat | 1.69 | 1,43 | 0.201 | 2.11 | 3, 258 | 0.100 | 0.30 | 3, 258 | 0.820 |
Table 3
Means, standard errors, effect sizes as well as their confidence intervals and post-hoc t-test results for the double standards dependent on the factors Group and Build for daughters’ and mothers’ evaluation of their own and their peers’ bodies as well as mothers’ evaluation of their daughter’s and the adolescent bodies
Variables | Daughters (n = 44) | Mothers (n = 44) | Across both groups | ||||||
|---|---|---|---|---|---|---|---|---|---|
eM (CIDS) | SE | g (CIg) | eM (CIDS) | SE | g (CIg) | eM (CIDS) | SE | g (CIg) | |
DS arousal | |||||||||
Thin | |||||||||
Model I | 0.557* [0.222, 0.891] | 0.166 | 0.484 [0.140, 0.828] | 0.091 [-0.070, 0.252] | 0.091 | 0.183 [0.034, 0.331] | 0.324* [0.137, 0.511] | 0.093 | 0.358 [0.168, 0.548] |
Model II | “ | “ | “ | 0.119 [-0.041, 0.280] | 0.080 | 0.225 [0.066, 0.383] | 0.338* [0.148, 0.528] | 0.094 | 0.371 [0.180, 0.563] |
Average weight | |||||||||
Model I | 0.330* [0.052, 0.607] | 0.138 | 0.347 [0.064, 0.631] | 0.165 [-0.040, 0.370] | 0.165 | 0.232 [0.019, 0.444] | 0.247* [0.073, 0.421] | 0.086 | 0.298 [0.123, 0.472] |
Model II | “ | “ | “ | 0.125 [-0.044, 0.294] | 0.084 | 0.226 [0.061, 0.391] | 0.227* [0.060, 0.394] | 0.083 | 0.294 [0.132, 0.457] |
Overweight | |||||||||
Model I | 0.631* [0.342, 0.920] | 0.143 | 0.655 [0.368, 0.943] | 0.352* [0.072, 0.632] | 0.352 | 0.388 [0.116, 0.659] | 0.491* [0.289, 0.694] | 0.100 | 0.527 [0.331, 0.723] |
Model II | “ | “ | “ | 0.227* [0.032, 0.423] | 0.097 | 0.345 [0.149, 0.542] | 0.429* [0.253, 0.605] | 0.087 | 0.512 [0.336, 0.688] |
Own | |||||||||
Model I | 0.347* [0.131, 0.562] | 0.107 | 0.454 [0.226, 0.682] | 0.108 [-0.074, 0.290] | 0.108 | 0.185 [0.010, 0.359] | 0.227* [0.084, 0.370] | 0.071 | 0.334 [0.191, 0.477] |
Model II | “ | “ | “ | 0.091 [-0.059, 0.241] | 0.075 | 0.183 [0.034, 0.331] | 0.219* [0.082, 0.356] | 0.068 | 0.338 [0.202, 0.474] |
Across all builds | |||||||||
Model I | 0.466* [0.306, 0.623] | 0.078 | 0.789 [0.613, 0.966] | 0.179* [0.052, 0.306] | 0.063 | 0.466 [0.352, 0.581] | 0.322* [0.218, 0.426] | 0.052 | 0.630 [0.523, 0.738] |
Model II | “ | “ | “ | 0.141* [0.046, 0.232] | 0.045 | 0.499 [0.415, 0.583] | 0.303* [0.212, 0.394] | 0.045 | 0.626 [0.524, 0.729] |
DS valence | |||||||||
Thin | |||||||||
Model I | -0.233 [-0.502, 0.036] | 0.133 | -0.254 [-0.528, 0.020] | -0.301 [-0.613, 0.011] | 0.155 | -0.291 [-0.600, 0.017] | -0.267* [-0.473, -0.061] | 0.102 | -0.277 [-0.480, -0.075] |
Model II | “ | “ | “ | 0.017 [-0.166, 0.200] | 0.091 | 0.028 [-0.155, 0.211] | -0.108 [-0.273, 0.057] | 0.082 | -0.139 [-0.302, 0.025] |
Average weight | |||||||||
Model I | -0.125 [-0.345, 0.095] | 0.109 | -0.173 [-0.389, 0.043] | -0.159 [-0.372, 0.054] | 0.105 | -0.237 [-0.438, -0.037] | -0.142 [-0.295, 0.011] | 0.076 | -0.207 [-0.361, -0.062] |
Model II | “ | “ | “ | 0.034 [-0.198, 0.266] | 0.115 | 0.044 [-0.187, 0.275] | -0.045e [-0.204, 0.114] | 0.079 | -0.061 [-0.271, 0.095] |
Overweight | |||||||||
Model I | -0.614* [-0.961, -0.267] | 0.172 | -0.529 [-0.876, -0.183] | -0.432* [-0.703, -0.161] | 0.134 | -0.467 [-0.743, -0.192] | -0.523* [-0.743, -0.303] | 0.109 | -0.504 [-0.722, -0.268] |
Model II | “ | “ | “ | -0.205* [-0.358, -0.051] | 0.076 | -0.399 [-0.253, -0.246] | -0.409*d [-0.599, -0.220] | 0.094 | -0.451 [-0.641, -0.260] |
Own | |||||||||
Model I | -0.528* [-0.828, -0.229] | 0.149 | -0.514 [-0.821, -0.207] | -0.085 [-0.348, 0.178] | 0.130 | -0.094 [-0.365, 0.178] | -0.307* [-0.506, -0.107] | 0.099 | -0.312 [-0.519, -0.106] |
Model II | “ | “ | “ | 0.210* [0.023, 0.397] | 0.093 | 0.336 [0.149, 0.523] | -0.159 [-0.339, 0.021] | 0.089 | -0.173 [-0.366, 0.020] |
Across all builds | |||||||||
Model I | -0.375* [-0.536, -0.214] | 0.080 | -0.644 [-0.818, -0.470] | -0.244* [-0.397, -0.092] | 0.076 | -0.467 [-0.623, -0.311] | -0.310* [-0.421, -0.199] | 0.055 | -0.564 [-0.679, -0.448] |
Model II | “ | “ | “ | 0.014 [-0.083, 0.111] | 0.048 | 0.050 [-0.035, 0.135] | -0.180* [-0.269, -0.092] | 0.044 | -0.366 [-0.470, -0.263] |
DS body attractiveness | |||||||||
Thin | |||||||||
Model I | 0.074e [-0.217, 0.364] | 0.144 | 0.080 [-0.197, 0.357] | -0.148 [-0.400, 0.104] | 0.125 | -0.200 [-0.421, 0.021] | -0.037c [-0.229, 0.155] | 0.095 | -0.044 [-0.219, 0.131] |
Model II | “ | “ | “ | -0.114 [-0.401, 0.174] | 0.143 | -0.118 [-0.406, 0.170] | -0.020 [-0.220, 0.180] | 0.099 | -0.021 [-0.218, 0.176] |
Average weight | |||||||||
Model I | -0.364* [-0.665, -0.062] | 0.149 | -0.373 [-0.664, -0.082] | -0.051c [-0.267, 0.164] | 0.107 | -0.068 [-0.293, 0.156] | -0.207* [-0.393, -0.022] | 0.092 | -0.238 [-0.421, -0.055] |
Model II | “ | “ | “ | 0.074 [-0.154, 0.302] | 0.113 | 0.097 [-0.131, 0.325] | -0.145 [-0.330, 0.040] | 0.092 | -0.163 [-0.350, 0.024] |
Overweight | |||||||||
Model I | -0.489*c [-0.819, -0.158] | 0.164 | -0.418 [-0.767, -0.069] | -0.523*d, f [-0.788, -0.257] | 0.132 | -0.557 [-0.837, -0.276] | -0.506c, f [-0.718, -0.294] | 0.105 | -0.484 [-0.704, -0.265] |
Model II | “ | “ | “ | -0.148f [-0.303, 0.007] | 0.077 | -0.293 [-0.444, -0.143] | -0.318*f [-0.504, -0.133] | 0.092 | -0.352 [-0.542, -0.163] |
Own | |||||||||
Model I | -0.261*a [-0.489, -0.034] | 0.113 | -0.338 [-0.569, -0.108] | 0.068e [-0.169, 0.306] | 0.118 | 0.084 [-0.158, 0.326] | -0.097e [-0.261, 0.068] | 0.082 | -0.121 [-0.289, 0.047] |
Model II | “ | “ | “ | 0.284*b, e [0.071, 0.498] | 0.106 | 0.412 [0.207, 0.618] | 0.011e [-0.144, 0.166] | 0.077 | 0.015 [-0.147, 0.177] |
Across all builds | |||||||||
Model I | -0.260*a [-0.436, -0.084] | 0.087 | -0.441 [-0.617, -0.265] | -0.163* [-0.321, -0.006] | 0.078 | -0.313 [-0.469, -0.512] | -0.212* [-0.330, -0.094] | 0.059 | -0.384 [-0.500, -0.268] |
Model II | “ | “ | “ | 0.024b [-0.103, 0.152] | 0.063 | 0.059 [-0.063, 0.181] | -0.118* [-0.218, -0.018] | 0.050 | -0.227 [-0.336, -0.118] |
DS body fat | |||||||||
Thin | |||||||||
Model I | 0.080 [-0.105, 0.264] | 0.091 | 0.132 [-0.047, 0.312] | -0.182e [-0.467, 0.103] | 0.141 | -0.196 [-0.473, 0.080] | -0.051d, e,f [-0.221, 0.119] | 0.084 | -0.066 [-0.229, 0.098] |
Model II | “ | “ | “ | -0.068 [-0.216, 0.080] | 0.074 | -0.137 [-0.286, 0.012] | 0.006 [-0.111, 0.123] | 0.058 | 0.010 [-0.105, 0.126] |
Average weight | |||||||||
Model I | 0.358* [0.129, 0.587] | 0.114 | 0.474 [0.248, 0.700] | 0.159e [-0.095, 0.413] | 0.126 | 0.183 [-0.077, 0.443] | 0.259*c [0.087, 0.430] | 0.085 | 0.320 [0.150, 0.490] |
Model II | “ | “ | “ | 0.170 [-0.012, 0.353] | 0.090 | 0.277 [0.093, 0.461] | 0.264* [0.119, 0.410] | 0.072 | 0.385 [0.241, 0.529] |
Overweight | |||||||||
Model I | 0.244* [0.028, 0.461] | 0.107 | 0.332 [0.113, 0.552] | 0.562*c, d,f [0.368, 0.757] | 0.096 | 0.845 [0.646, 1.044] | 0.403*c [0.258, 0.549] | 0.072 | 0.568 [0.419, 0.718] |
Model II | “ | “ | “ | 0.165 [-0.054, 0.383] | 0.108 | 0.227 [0.010, 0.444] | 0.205* [0.050, 0.359] | 0.076 | 0.284 [0.132, 0.435] |
Own | |||||||||
Model I | 0.250* [0.043, 0.457] | 0.102 | 0.360 [0.153, 0.567] | 0.176*e [0.034, 0.318] | 0.071 | 0.367 [0.223, 0.510] | 0.213*c [0.088, 0.339] | 0.062 | 0.362 [0.238, 0.485] |
Model II | “ | “ | “ | -0.011 [-0.179, 0.156] | 0.083 | -0.020 [-0.187, 0.146] | 0.119 [-0.013, 0.252] | 0.066 | 0.188 [0.055, 0.321] |
Across all builds | |||||||||
Model I | 0.233* [0.116, 0.350] | 0.058 | 0.618 [0.506, 0.731] | 0.179* [0.055, 0.303] | 0.061 | 0.451 [0.333, 0.570] | 0.206* [0.121, 0.291] | 0.042 | 0.539 [0.459, 0.619] |
Model II | “ | “ | “ | 0.064 [-0.033, 0.161] | 0.048 | 0.201 [0.107, 0.269] | 0.148* [0.075, 0.222] | 0.036 | 0.419 [0.345, 0.494] |
Fig. 2
Means and standard errors displayed as error bars for the DS scores of valence, arousal, body attractiveness, and body fat for daughters (black) and mothers (grey) and the builds thin, average weight, overweight and own. Bar charts in the upper row of each dependant variable correspond to the results of Model I of daughters’ and mothers’ body ratings for the adolescent (in the case of the daughter) and adult bodies (in the case of the mother). Bar charts in the lower row of each dependant variable represent daughters’ and mothers’ body ratings for the adolescent bodies (Modell I). Asterisks denote significant Bonferroni-Holm-corrected post hoc effects between mothers and daughters for the respective build. Crosses indicate a substantial DS, if differing significantly from zero
×
![]()
Daughters’ and Mothers’ Ratings of the Adolescent and Adult Female Bodies (Model I)
Regarding DS in arousal, daughters indicated significantly higher arousal for all builds when these builds were presented with their own than with another face, while mothers reported significantly higher arousal only for the overweight body with their own compared to another face. However, the main effects of Group, Build and the Group × Build interaction were non-significant. Thus, higher arousal was reported for builds with participants’ own compared to another face.
Regarding DS in valence, daughters reported higher negative feelings for the overweight and their own body with their own than with another face, while this was only the case for the overweight body in mothers. However, in terms of group differences, the main effects of Group, Build, and the Group × Build interaction were non-significant. Accordingly, across all builds, daughters and mothers showed higher negative feelings for the builds with their own compared to another face.
Regarding DS in body attractiveness, daughters reported lower attractiveness for the average-weight, overweight, and their own body with their own compared to another face, while mothers only showed a self-deprecating DS for the overweight body. In terms of group differences, the main effect of Build and the Group × Build interaction were significant, while the main effect of Group was non-significant. For daughters, this was reflected by the self-deprecating DS for the overweight body, that differed significantly from the non-significant DS for the thin build (gav = 0.533). For mothers, the self-deprecating DS for the overweight body differed significantly from the non-significant DS for the average-weight (gav = 0.554) and own body (gav= 0.674).
Regarding DS in body fat, daughters rated the average-weight, overweight, and their own body with more body fat when these bodies were presented with their own than with another face, whereas mothers only showed self-deprecating DS for the overweight and own build. The analysis of group differences revealed a significant main effect of Build and a significant Group × Build interaction emerged, whereas the main effect of Group was non-significant. Mothers showed a significantly higher DS for the overweight build compared to the thin (gav = 0.923), average-weight (gav = 0.521), and own build (gav= 0.666).
Mothers’ Ratings of the Adolescent Bodies (Model II)
Regarding DS in arousal, like their daughters, mothers reported higher arousal for all builds with their daughters’ face compared to another face and showed higher arousal only for the overweight body with their daughters’ compared to another face. However, the main effects of Group, Build and the Group × Build interaction were non-significant.
Regarding DS in valence, mothers felt more negative about the overweight build with their daughters’ than with another face, but showed more positive feelings about their daughters’ body when presented with their daughters’ own compared to another face. Regarding group differences, only the main effect of Build was significant. Mothers and daughters felt more negative about the overweight build with the daughters’ face than with another face, and this self-deprecating DS differed significantly from the non-significant DS for the average-weight build (gav = 0.439).
Regarding DS in body attractiveness, mothers rated their daughters’ body as more attractive with their daughters’ than with another face. Further, the main effects of Group, Build and the Group × Build interaction were significant. Across all builds, daughters evaluated the bodies as less attractive with their own than with another face, differing significantly from their mothers’ ratings, who showed no DS (gav = 0.560). The significant enhancing DS of mothers for their daughter’s body differed significantly from their daughters’ self-deprecating DS (gav = 0.745). Furthermore, mothers’ enhancing DS for their daughters’ body differed significantly from the non-significant DS for the overweight build (gav = 0.715).
Regarding DS in body fat, across both groups and all builds, more body fat was reported for builds with one’s own than with another face. However, the main effects of Group, Build, and the Group × Build interaction were non-significant.
Direct and Indirect Familial Transmission
Regarding direct transmission, there was either no significant association between daughters’ DS scores and their mothers’ additional DS scores (all p > .05) or zero was included in the bootstrapped confidence intervals, indicating no relevant finding. The correlation analyses between negative maternal feedback (VCOPAS NWS) and daughters’ DS scores revealed significant associations for valence regarding one’s own body (r = − .527, p < .001, one-tailed, CI [-0.726; − 0.054]) and for body attractiveness (r = − .375, p = .006, one-tailed, CI [-0.600; − 0.005]) and body fat (r = .319, p = .017, one-tailed, CI [0.045; 0.545]) regarding the overweight build. Correlation analyses between positive maternal feedback (VCOPAS PWS) and daughters’ DS revealed significant associations regarding arousal (r = − .314, p = .038, CI [-0.528; − 0.092]), valence (r = .417, p = .005, CI [0.145; 0.607]), and body attractiveness (r = .331, p = .028, CI [0.049; 0.578]) for one’s own body. Furthermore, significant associations emerged between the VCOPAS NWS and daughters’ SC (r = .457, p = .001, one-tailed, CI [.124; .660]), DT (r = .433, p = .002, one-tailed, CI [.067; .695]), and BD (r = .446, p = .001, one-tailed, CI [.068; .722]) and between the VCOPAS PWS and daughters’ EC (r = -.325, p = .016, one-tailed, CI [-.576; -.023]) and BAS-2 (r = .310, p = .020, one-tailed, CI [.061; .522]). Other correlations were either non-significant (p > .05) or zero was included in the bootstrapped confidence interval.
Regarding indirect transmission, all correlations between daughters’ and mothers’ DS scores and between daughters’ and mothers’ eating pathology/body image scores were non-significant or indicated no relevant effect, except for a significant correlation between daughters’ and mothers’ BAS-2 scores (r = .266, p = .040, one-tailed, CI [-0.069; 0.517]).
Discussion
This study explored DS in daughter-mother dyads and investigated whether such biases are transmitted from mothers to daughters. Consistent with our assumptions regarding question 1, we found significant DS in daughters and their mothers regarding arousal, valence, body attractiveness, and body fat. These results underline that not only adult women (Voges et al., 2022) but also female adolescents apply self-deprecating DS to bodies, even when they report low levels of body dissatisfaction and eating pathology (Bauer et al., 2017; Hilbert & Tuschen-Caffier, 2016; Hilbert et al., 2007; Kappel et al., 2012). However, the results regarding the emotional evaluation of arousal and valence differed from the results regarding body ratings of body attractiveness and body fat.
Regarding the extent of DS for arousal, for each build, daughters showed higher arousal when the bodies were presented with their own than with another face, while mothers only showed this DS for the overweight body. In terms of valence, the tests of DS against zero indicated that daughters had higher negative feelings about the overweight body and their own body when presented with their own than with another face, while this was only the case for the overweight body in mothers. Although these patterns of DS for arousal and valence suggested that multilevel models would reveal effects of group and build, the main and interaction effects were non-significant. Thus, self-deprecating DS were not specific for builds and groups; rather, daughters and mothers showed overall self-deprecating DS for arousal and valence, insofar as they reported higher arousal and higher negative feelings when a body was presented with their own than with another face.
Regarding body attractiveness and body fat, we found DS differing from zero, but also main effects of Build and significant interactions of Group and Build. In more detail, daughters reported significantly lower levels of body attractiveness and higher levels of body fat for the average-weight, overweight, and own build with their own face than with another face. However, daughters showed no DS for the thin build, which is contrary to findings in adult women (Voges et al., 2022). To explain this discrepancy, it may be suggested that the daughters have not yet fully internalized the Western thin-body ideal (Crossley et al., 2012), as thin-ideal internalization increases in females from pre-puberty to adulthood (Suisman et al., 2014). Moreover, they might have perceived this build as equally ideal for themselves and others, meaning that they did not differentiate between the faces for the thin build while showing self-deprecating DS of body fat and body attractiveness for the other builds, which did not correspond to the Western ideal. For mothers, the analyses of body attractiveness and body fat showed that mothers rated the overweight build as less attractive with their own than with another face and evaluated the overweight build and their own build with more body fat with their own than with another face. This contradicts the previous study by Voges et al. (2022), who found that young women showed self-critical DS of valence, arousal, and body attractiveness but not of body fat for their own build. However, age might be an influencing factor here, as female bodies change and BMI increases with age (German Federal Statistical Office, 2023, June). While on average, the mothers were only slightly overweight, they might have perceived their own body as having too much body fat, resulting in stricter evaluations for their own than for another face. At the same time, the mothers may place greater importance on the functionality of their body, focusing more on physical ability and experiences (e.g., giving birth, overcoming illness) while accepting that their body is aging (Alleva & Tylka, 2021; Tiggemann, 2015). In contrast, mothers showed self-deprecating DS for the overweight build for all variables, underlining a biased evaluation of overweight builds in general, which is in accordance with previous studies (Quittkat et al., 2023; Voges et al., 2019, 2022) and with their daughters’ evaluations in the present study. As people with overweight are often stigmatized in Western societies through derogatory attributions (e.g., “disorganized”, “lazy”; Hu et al., 2018), both daughters’ and mothers’ evaluations might have been influenced by these prejudices, resulting in a stronger rejection of this build when presented with their own than with another face.
Regarding question 2, and mothers’ DS regarding their daughter’s body, we aimed to investigate whether mothers might display different DS than their daughters with respect to their daughter’s body and the adolescent builds. We found that the mothers’ evaluation differed significantly from their daughter’s evaluation: Mothers reported higher levels of body attractiveness for their daughter’s body with their daughter’s face than with another face, whereas daughters evaluated their own body as less attractive with their own face compared to another face. This positive DS of mothers is in line with a previous study in which parents rated their own child more positively than another child (Wenger & Fowers, 2008). Some maternal DS differed significantly from zero, indicating relevant DS, such as more positive feelings for their own daughter’s body and more negative feelings and more arousal for the overweight build with their own daughter’s face than another face. Although the different direction of maternal DS indicated a trend towards group differences, neither effects of group nor a significant interaction of group and build were found for arousal, valence and body fat.
Regarding familial transmission of DS, there were no significant associations between daughters’ and mothers’ DS for the adolescent builds (direct, question 3) and between daughters’ and mothers’ DS for their own and the respective peer group builds (indirect, question 6). This is contrary to Bauer et al. (2017), who reported evidence of direct and indirect transmission of attentional patterns from mothers to daughters. However, Bauer et al. (2017) examined implicit gaze patterns, whereas familial transmission might not apply to explicit body evaluation such as DS. Nevertheless, we found significant associations of maternal feedback with daughters’ DS and body image (direct transmission, questions 4 and 5), suggesting a central role of maternal feedback. Specifically, the greater the negative maternal feedback, the more self-deprecating the daughters’ DS of valence for their own body and of body attractiveness and body fat for the overweight body. Moreover, the greater the positive feedback, the less self-deprecating the daughters’ DS of arousal, valence, and body attractiveness for their own body. Whereas greater negative maternal feedback was further associated with higher shape concerns, a higher drive for thinness, and higher body dissatisfaction in daughters, greater positive maternal feedback was related to higher body appreciation. Additionally, there was a positive correlation between mothers’ and daughters’ body appreciation, suggesting indirect transmission (question 7). Although not all associations between mothers’ and daughters’ body image were significant, these results underline the well-documented influence of maternal feedback and role-modeling on daughters’ body image (Abraczinskas et al., 2012; Crowther et al., 2015; Rodgers & Chabrol, 2009). Nevertheless, other influences might have an even greater effect on daughters’ DS, especially as female adolescents report high daily screen time on social media (Nesi et al., 2023). In line with this, current research applying adaptations of the tripartite model highlights social media as another source of appearance pressure (Jarman et al., 2021; Roberts et al., 2022). As a recent study revealed that females with and without ED report increased body dissatisfaction after viewing Instagram pictures of thin and toned bodies (Ladwig et al., 2024), the constant evaluation processes of one’s own and other bodies that occur when using social media might foster the development of DS in adolescents. Thus, peers and social media might have a greater influence on explicit body evaluation than mothers.
Some limitations should be considered when interpreting these results. Participants were asked not to wear special make-up or clothes, or to do anything else in advance of the photo session that they would not do in their daily life. However, it cannot be ruled out that they prepared themselves in some way before the session (e.g., not drinking or eating, applying special body lotion) and therefore looked slightly different in the photos than usual. Although the photographs were standardized, the bodies differed in their characteristics (e.g., size/shape/skin/scars). All pictures were colored gray to account for some differences, but this restricts ecological validity. Furthermore, this study was not pre-registered. Additionally, it cannot be ruled out that the study was underpowered to detect small effects. As participants were informed in advance that they would be photographed in underwear, individuals with high body dissatisfaction may not have participated, which might have led to a selection bias. The daughters had a low score of 21.11 for body dissatisfaction, which corresponds to the 30th percentile of the norms for 13-20-year-old girls (Kappel et al., 2012). Moreover, the adolescent and adult sample covers a higher range of BMI than the standardized stimulus material. Therefore, some participants’ BMIs were further away from the different control bodies while others were closer, and this greater discrepancy might have influenced identification with the models and consequently body evaluation. However, this discrepancy reflects the participants’ usual environment, as their BMI also differs from the BMI of other bodies to which they are exposed in everyday life (e.g. in class, at work, in advertising, when participating in sport, etc.). Nevertheless, future studies should examine DS using stimulus material that is idiosyncratically matched to the participants. Additionally, daughters reported a mean BMI that corresponded to the 62nd BMI percentile and was slightly higher than the average-weight build. The BMI might have varied more among our female adolescents because height and weight vary strongly in puberty (e.g., due to hormonal and genetic influences, time of onset of puberty). However, it cannot be ruled out that daughters with a lower BMI, but who still met our criteria, may not have participated. Furthermore, our findings are only cross-sectional and results for familiar transmission are only based on correlational analyses, whereas longitudinal data on DS and familial transmission are lacking. Finally, these results are not generalizable. For instance, the presence of other family members in the home, such as brothers, sisters, or fathers, might influence the development and strength of DS. Furthermore, as adolescents with ED were found to report more negative interpersonal beliefs than adolescents without any mental disorder (Harper et al., 2023), DS might function differently in this group, as previously reported for women with AN, who showed stricter DS than women without ED (Voges et al., 2018). Therefore, longitudinal studies on DS and their familial transmission including adolescents with and without ED, and living in various different family constellations (e.g., with other parents, brothers and sisters), are required.
Nevertheless, this study is the first to provide evidence that DS exist not only in adult women with and without ED (Quittkat et al., 2023; Voges et al., 2018, 2022) but also in female adolescents and their mothers. As young adolescents begin to compare themselves with others and attach greater importance to others’ judgments (Sebastian et al., 2008), female adolescents might be particularly vulnerable to develop biased body evaluation for their own, an average-weight, and overweight build. To raise awareness of DS and enhance positive body image, psychoeducation on DS and preventive interventions in schools might be helpful. Moreover, as negative maternal feedback impaired daughters’ body image, parental training on body-related behavior regarding body shaming, comments on bodies/appearance, weight stigmatization, and appraisal of thin builds might reduce the harmful influences of maternal behaviors on daughters. Finally, cognitive bias modifications (Dietel et al., 2020) of DS and psychoeducation on body functionality might be promising approaches to reduce DS in female adolescents and their mothers.
Acknowledgements
We would like to thank Sarah Mannion for correcting the English-language manuscript.
Declarations
Language Statement
In the following article, we use the terms “underweight”, “average weight” and “overweight” in line with the body mass index (BMI) ranges defined by the World Health Organization. We recognize the frequently negative connotations of “overweight” and the inherent prejudice and weight stigmatization associated with this term within the general population. We distance ourselves completely from any form of stigmatization, insults, and offensive language, and wish to emphasize that we use these terms only to describe the BMI ranges categorized by the WHO. None of the descriptions in this article are intended to stigmatize any individual or to attribute the term “overweight” as an identity description, prejudice, or negative stereotype to the participating daughter-mother dyads or other groups of people.
Ethics Approval
The study was approved by the ethics committee of Osnabrück University and is in line with the ethical principles of the Declaration of Helsinki. As the participants included minors, their written informed consent as well as written informed consent from their parents was mandatory.
Conflict of Interest
The authors report no conflict of interest.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
