1.1. The Satter Eating Competence model
The Satter eating competence model (ecSatter) is an evidence-based and practice-based conceptualization of the interrelated spectrum of eating attitudes and behaviors [
3]. According to Satter [
3] (p. S142), competent eaters are “positive, comfortable, and flexible with eating and are matter-of-fact and reliable about getting enough to eat of enjoyable and nourishing food”. The Satter eating competence model breaks eating competence down into four components: (1) attitudes about eating and food; (2) food acceptance skills; (3) internal regulation skills; and (4) skills and resources for managing the food context and orchestrating family meals [
3] (p. S142). Positive eating attitudes refer to a positive interest in food/eating, attunement to inner (e.g., appetite) and outer (e.g., food attractiveness and availability) food experiences, and self-trust about managing food [
3]. Food Acceptance skills refer to feeling comfortable in the presence of novel food, and being willing to experiment with unfamiliar food and learn to like it, i.e., not being a picky eater [
3]. Internal Regulation skills refer to the experiential processes of hunger, appetite, and satiety, which attend to physiological homeostatic mechanisms that support constitutionally appropriate body weight [
3]. Dieting requires ignoring these homeostatic mechanisms and triggers counterregulatory mechanisms leading to weight gain [
3]. Contextual Skills refer to meal planning (structure) and support in choosing preferred food and eating it in amounts that satisfy hunger/appetite [
3]. Maintaining a pattern of regular meals depends on the three other competencies, i.e., positive attitude towards eating, accepting food, and being attuned to signs of hunger, appetite and satiety (regulation) [
3].
1.2. The construction of the Eating Competence Satter Inventory (ecSI) and its different adaptations (ecSI/LI, ecSI 2.0, ecSI 2.0TM)
To assess the four aforementioned eating competencies, Satter developed the Eating Competence Satter Inventory (ecSI) [
3]. The ecSI consisted of 16 items divided over four subscales, capturing the four eating competences: Eating Attitudes (5 items), Food Acceptance (3 items), Internal Regulation (3 items), and Contextual Skills (5 items). Each item needs to be answered on a five-point scale with the response options
always (score 3),
often (score 2),
sometimes (score 1),
rarely (score 0) and
never (score 0). The total ecSI score can range from 0 to 48, and an ecSI total score ≥ 32 indicates eating competence [
4]. Lohse et al. [
4] validated the ecSI in a community sample of 863 adults (78.7 % female) and could replicate the theoretical four-factor structure of the ecSI utilizing exploratory factor analysis. Results showed that all four subscales were reliable: Eating Attitudes (α=.84), Food Acceptance (α=.65), Internal Regulation (α=.75), and Contextual Skills (α=.74) [
4]. Participants scoring high on eating competence (ecSI total score ≥ 32) were generally older, reported lower BMI, fewer eating disorder symptoms, more physically activity, and made more healthy food/ meal choices [
4]. Stotts and Lohse [
5] investigated the test-retest reliability (interval 2 to 6 weeks) of the ecSI in 259 white females, which revealed moderate-to-high Spearman rank correlation coefficients: ecSI total score (
rs = .68), Eating Attitudes (
rs = .70), Food Acceptance (
rs = .65), Internal Regulation (
rs = .52), and Contextual Skills (r
s = .70).
In 2011, Krall and Lohse [
6] investigated the validity of the ecSI in women with low income (N=25). Four items of the ecSI were misinterpreted due to wording and clarity problems [
6]. These four misinterpreted items were revised based on participants’ feedback, retested and combined with the 12 unaltered items in the ecSI for Low-Income (ecSI/LI). Krall and Lohse [
6] validated the ecSI/LI in 507 females with low income and showed that eating competent women (ecSI/LI total score ≥ 32) reported more physical activity, more fruit and vegetable intake, better food planning, a lower BMI, less body weight dissatisfaction and less eating disorder symptoms compared to non-eating competent women.
In 2015, Lohse [
7] examined whether the ecSI/LI could also be used in the general population, which was not considered low-income. She administered both the ecSI and the ecSI/LI (with four altered items) to 127 participants; the correlation between both versions of the instruments was very high (
r = .98). The author concluded that the ecSI/LI was also applicable in the general population, and renamed the ecSI/LI as Eating Competence Satter Inventory 2.0 (ecSI 2.0). In 2015, Tilles-Tirkkonen and colleagues [
8] validated a preliminary Finnish translation of the ecSI 2.0 in a Finnish sample of 976 adolescents (54% girls) aged 10-17 years old. They could replicate the four-factor structure of the ecSI 2.0 using confirmatory factor analysis and showed that the total scale and the four subscales were reliable: Total ecSI 2.0 (α=.92), Eating Attitudes (α=.87), Food Acceptance (α=.78), Internal Regulation (α=.83), and Contextual Skills (α=.81). Eating competent adolescents (ecSI 2.0 total score ≥ 32) reported a higher level of self-esteem and a stronger sense of identity coherence; were less dissatisfied about their body size, and had less often tried to lose body weight. They also reported higher meal frequency, more consumption of fruits and vegetables, and more health-promoting family meals.
More recently, Godleski, Lohse, and Krall [
9] investigated the factor structure of the ecSI 2.0 in 2010 adults. The findings of the confirmatory factor analysis confirmed the four-factor structure of the ecSI 2.0, but suggested a migration of item 9 “
I trust myself to eat enough for me” from the Internal Regulation subscale to the Eating Attitudes subscale. The migration of item 9 improved model fit and reduced the magnitude of the correlation between the Eating Attitudes and the Internal Regulation subscales (e.g., in sample A from
r = .90 to
r = .75). In sum, the final ecSI2.0
TM still consists of 16 items divided over four subscales but with different item numbers: Eating Attitudes (n
items=5+1=6), Food Acceptance (n
items=3), Internal Regulation (n
items=3-1=2), and Contextual Skills (n
items=5). The correlational patterns of the ecSI2.0/ecSI2.0
TM (sub)scales with other variables remained very similar [
9]. Given that the ecSI 2.0
TM is used both at the total and the subscales level, Godleski et al. [
9] also investigated a second-order factor model in which each of the subscales loaded on one higher-order latent factor (Eating Competence). The data fitted this higher-order model well, and the subscales loaded on the overall higher-order factor as follows: Eating Attitudes (.96), Food Acceptance (.62), Internal Regulation (.83), and Contextual Skills (.82). Finally, de Queiroz and colleagues [
10] investigated the validity and reliability of the Brazilian Portuguese version of the ecSI2.0
TM. Confirmatory factor analysis confirmed the four factor structure of the ecSI2.0
TM BR and the internal consistency coefficients of the subscales were: ecSI2.0
TM BR total score (α =.87), Eating Attitudes (α=.79), Food Acceptance (α =.73), Internal Regulation (α =.53), and Contextual Skills (α=.82) [
10].
As far as we know, no studies investigated (1) the factor structure, (2) reliability, (3) measurement invariance across sex and age, and (4) validity of the ecSI 2.0
TM DUTCH in adolescents. Gaining insight in the eating attitudes and behaviors of adolescents is however important, given that those attitudes and behaviors impact their relationship with food and their bodies. Adolescents who have positive eating attitudes, particularly if they score high in food acceptance skills, may be less vulnerable to food fads and good-food-bad-food thinking. Adolescents who score high in internal regulation are less likely to engage in restrictive dieting and compensatory overeating. Adolescents who give evidence of contextual skills may be more resourceful and confident in providing for their nutritional needs [
3].
Therefore, it is important to investigate whether the ecSI 2.0
TM DUTCH is a valid and reliable instrument to assess eating competence skills in adolescents, whether it is measurement invariant across sex and age, and to examine its’ association with dysregulated eating behaviors, body attitude and identity development, which are closely related with each other [
11,
12]. Up till now, several studies investigated the association between disordered eating and eating competence [
4,
6]; however, only one study investigated the association between identity development and positive eating attitudes and behaviors in adolescence [
8], although both, healthy eating attitudes and behaviors and identity development are core developmental tasks in adolescence [
1,
2].
1.3. The present study
Therefore, in the present study, we investigated (1) the four-factor structure of the ecSI 2.0TM DUTCH and the higher order model with one latent construct (eating competence) as well as the correlations between the latent factors in a sample of Flemish adolescents, (2) the reliability of the total scale and the subscales, (3) the measurement invariance of the ecSI 2.0TM DUTCH across sex and age; and (4) the associations between the ecSI 2.0TM DUTCH total/subscales and Body Mass Index (BMI), eating disorder symptoms (Drive for Thinness, Bulimia, Body Dissatisfaction) and identity measures (diffusion/synthesis), to investigate convergent/divergent validity.
Based on the existing literature, (1) we expected to find a four factor structure of the ecSI 2.0
TM DUTCH with a slightly better fit for a factor solution in which item 9 “
I trust myself to eat enough for me” belongs to the Eating Attitudes scale [
9] compared to a second factor solution with item 9 belonging to the Internal Regulation scale [
7]. We also expected that the correlation between the Eating Attitudes and the Internal Regulation subscales would be lower in the first factor solution (
r = ± .75) compared to the second factor solution (
r = ± .90) based on a study in adults [
9]. Finally, we also hypothesized that a second-order model in which each of the subscales loaded on one higher-order latent factor (i.e., Eating Competence) would fit the data well [
9]. (2) Concerning the reliability of the subscales, we expected acceptable (α ≥ .70) internal consistency for the Food Acceptance subscale, good/excellent (α ≥ .80/90) internal consistency coefficients for the Contextual Skills and Eating Attitudes subscales, and poor internal consistency of the Internal Regulation subscale [
10]. (3) Given the lack of previous studies on the measurement invariance across sex and age of the ecSI 2.0
TM DUTCH , no concrete hypotheses could be forwarded. (4) Finally, with respect to convergent/divergent validity, we expected that the total ecSI 2.0
TM DUTCH and its subscales would be unrelated or negatively related to BMI, and negatively associated with eating disorder symptoms [
4,
6] and identity confusion [
8]; whereas the scales would be positively related to identity synthesis [
8].