Participants
The sample consisted of 301 participants (180 female, 1 undisclosed), with a mean age of 41.0 (SD = 11.9) and a mean BMI of 29.8 (SD = 8.6). The majority reported to be omnivores (n = 283), and others reported to follow a vegetarian diet (n = 8), vegan diet (n = 2), and non-disclosed information (n = 8). Most participants were White (n = 273), followed by smaller groups of Black (n = 7), Asian (n = 12), and mixed race (n = 7) participants (2 participants did not disclose any background). Subject-to-variables ratio was assessed for the recruitment of adequate participants, where an acceptable 8 to 1 (Henson & Roberts, 2006) was exceeded bearing in mind the validity testing that was also conducted on the same sample. With significance set at .05, medium effect size, power = .80, and 5 variables in one model, indicated a sample size of 126 participants (Cohen, 1992; Faul et al., 2009). To ensure that consideration would be given to average and obese populations, the recruitment focused on an average weight population until reaching half of the sample, and the rest specifically targeted obese populations. An ability to identify individual differences and developing a scale that would be applicable across different subsamples led to separate identical analyses depending on BMI categorisation.
Materials
The Mindful Eating Behaviour Scale – Trait (MEBS – T; Mantzios, 2022). The scale contains of 10 items that measure two components of mindful eating behaviour: “Sensory Attention” (5 items) and “Non-judgmental Awareness” (5 items). Both factors are descriptive and align with mindful eating behaviour and mindfulness theory. Sample questions include “I fully taste what I am eating” and “I hold my attention on what I am eating, despite recognising the occurrence of thoughts and/or feelings while I am eating”. The scale utilises a 4-point Likert scale with responses ranging from 1 (strongly disagree) to 4 (strongly agree). The Cronbach alpha for the total score of the scale in the present study was .85, for the subscales “Sensory Attention” .85 and “Non-judgmental Awareness” .82.
Five-Facet Mindfulness Questionnaire – 15 (FFMQ-15; Gu et al., 2016). This is a shorter version of the original 39-item FFMQ (Baer et al., 2008), which measures five facets of mindfulness: Observing, Describing, Acting with Awareness, Non-Judging and Non-Reactivity. Sample questions include “I do jobs or tasks automatically without being aware of what I’m doing” and “I find myself doing things without paying attention”. Responses are recorded on a 5-point Likert scale ranging from 1 (never or very rarely) to 5 (very often or always true). A score is combined for each facet of the scale. The Cronbach alpha for the present study for the overall score was .79, and for observing .65, describing .81, acting with awareness .78, non-judging .85 and non-reactivity .72.
Dutch Eating Behaviour Questionnaire (DEBQ; Van Strien et al., 1986). The DEBQ is a 33-item scale containing External Eating, Restrained Eating and Emotional Eating items. Sample items include “Do you have the desire to eat when you are irritated?” and “Do you have a desire to eat when you have nothing to do?”. Responses are recorded on a 5-point Likert scale on a 5-point Likert scale ranging from 1 (never) to 5 (very often). Cronbach’s alpha for the present research was .88 for External Eating, .89 for Restrained Eating and .96 for Emotional Eating.
Grazing Scale (GS; Lane & Szabo, 2013). The 8-item Grazing Scale investigates the repetitive eating of small amounts of food. A sample item is ‘Have you ever felt compelled or driven to eat, even when not hungry?’, and responses range from 1 (rarely) to 5 (all of the time). Cronbach’s alpha for the present research was .92.
Procedure
Participants were provided with a link to an online platform (i.e., Profilic), and were reimbursed for their participation time (£6.00/hr). Participants first viewed the participant information form, followed by the consent form, and upon consenting were exposed to a demographics page and the psychometric material (MEBS-T, DEBQ, GS, and FFMQ-15). When completing the materials, the participants were directed to a debrief page, which concluded their participation.
Data Analysis
Data screening was conducted prior to inferential analyses to evaluate whether assumptions were met regarding the presence of outliers, multivariate normality, linearity, and homogeneity of variance. The Kaiser–Meyer–Olkin (KMO) measure of sampling adequacy and Bartlett's test of sphericity were also evaluated preceding any attempts to conduct exploratory factor analyses to ensure data fitness. Once all assumptions were satisfied, Exploratory Factor Analysis (EFA), with principal axis factor extraction and oblique rotation was performed. Scree plot identification, Eigenvalue (>1), and higher item loading greater than 0.30 were criteria to evaluate factor extraction, with the addition of a Monte Carlo PCA for parallel analysis indicative of rejecting or accepting factors (Kaiser, 1960; Tabachnick & Fidell 2007). Once the factor structure was identified, Pearson's correlations between the subscales were performed to investigate the potential of an overall score calculation for the scale, as well as Cronbach's α internal consistency coefficients were calculated for both the overall scale and subscales. All data analyses to this point were analysed using IBM SPSS 28. Data were further analysed using AMOS 24. “Sensory Attention” and “Non-judgmental Awareness” were first-order latent factors that loaded onto a second-order latent factor; that is, “Mindful Eating Behaviour Scale”. Structural Equation Modelling was run using the maximum-likelihood method, and Confirmatory Factor Analysis (CFA) goodness-of-fit was assessed for this one factor, second-order model, which included indexes of fit: a Chi-squared by degree of freedom (χ2 CMIN/df) ratio < 5; root mean square error of approximation (RMSEA) < 0.08; Adjusted Goodness of Fit Index (AGFI), the Goodness of Fit Index (GFI), Tucker–Lewis Index (TLI), Comparative Fit Index (CFI), and Incremental Fit Index (IFI) > 0.9; Parsimony Normed Fit Index (PNFI) > 0.5 (Bentler & Bonett, 1980; Hooper et al., 2008; Kline, 2015).
Results
The acceptability of the factorial structures was assessed by exploring the Kaiser–Meyer–Olkin (KMO) and the Bartlett’s sphericity test. The Kaiser–Meyer–Olkin measure of sampling adequacy was .85, and Bartlett’s Test of Sphericity (p < .001) indicated that the assumptions for a factor analysis were met. Principal component analysis revealed the presence of two components with eigenvalues exceeding 1, explaining 42.9% and 19.2% of the variance. An inspection of the screeplot revealed a break after the second component, highlighting the items that loaded on the two subscales. Parallel analysis indicated that both components should be accepted (see Table 2). Oblimin rotation was performed assuming that there would be an overall correlation between subscales as all of the items were set to reflect and measure mindful eating behaviour. The analysis indicated strong loadings (>.5), apart from items 5 and 7. The two-factor solution explained a total of 62.2% of the variance. The sample was separated, and isolated factorial analyses were repeated for obese and average-weight subsamples. While the findings were analogous between the overall sample and the obese sample, the average weight (BMI<30) identified items 5 and 7 to load onto a third factor (Eigenvalue = 1.007), which violated the theoretical assumption of the two-factor structure of the scale, and items were excluded from the final version of the scale to ensure consistency of comparisons between groups and clarity in identifying individual differences in future research. Correlations between subscales were significant and of moderate strength (r = .42), corresponding to the association that would be expected in a homogeneous scale.
Please Insert Table 2 here
Please Insert Table 3 here
Reliability was assessed by estimating Cronbach’s alpha values for the total score, and the subscales that were specified during the development of the scale. The “Sensory Attention” subscale displayed and an alpha value of .85, and the “Non-judgmental Awareness” subscale .82, while for the total score the alpha value was .85; all exceeding the recommended values for an internally consistent scale.
Correlations are reported in Table 4. Person’s correlations were conducted between the MEBS-T, DEBQ, GS, and FFMQ-15. MEBS-T showed a significant negative correlation to emotional eating and a significant positive correlation to overall mindfulness scores. Interestingly, the Sensory Attention subscale displayed a significant negative relationship to emotional eating and grazing, while the Non-judgmental Awareness subscale displayed a positive significant relationship to restrained eating.
Please Insert Table 4 here
The CFA revealed that the 10-items were not a good fit for the proposed model: CMIN/df = 3.75; RMSEA = .096; AGFI = .87, GFI = .92, TLI = .91, CFI = .93, IFI = .93; PNFI = .69, despite five of the eight indices of fit indicating an adequate fit. Contrary, the removal of Items 5 and 7, as indicated being the weakest in loading for both the EFA and CFA, proposed a better fit: CMIN/df = 3.32; RMSEA = .088; AGFI = .90, GFI = .95, TLI = .94, CFI = .96, IFI = .96; PNFI = .64; all indicating a good fit, apart from the RMSEA marginally exceeding the suggestive value by .008. (see Figure 1 for loadings).
Please Insert Figure 1 here