We used descriptive statistics to evaluate BE prevalence and frequency in this sample. Due to the ordinal nature of BE frequency measure, we used Spearman’s rho to investigate correlations of BE frequency with BMI, ED symptoms, negative affect, and consumption of nutritious foods. Finally, we used a binomial logistic regression model to investigate the degree to which BE frequency contributed to risk for obesity status while controlling for age, race, and ethnicity. Regarding BE frequency and prevalence (Table 1), 48% of the sample reported never BE. Using the clinical frequency criterion of BE once per week or more, 26.5% of the sample reported BE at least weekly. As a secondary measure of BE frequency in this sample, we examined rates of BE using the BE item on the EDE-Q Short. One participant’s answer did not align with their response on the VA-BES, resulting in 25.5% of the sample endorsing at least one BE episode in the past week. BE frequency was correlated with higher BMI (⍴ = .40, p < .01), more ED symptoms (⍴ = .65, p < .001), greater negative affect (⍴ = .43, p < .01), and less frequent consumption of nutritious foods (⍴ = .16, p < .05). Finally, BE frequency contributed significant risk for obese status based on self-reported BMI (OR = 2.08; 95% CI [1.52; 2.82], p < .001). Age, race, and ethnicity categories were non-significant in the model.
Study 2 (Replication): Sample Living with Food Insecurity
Participants
This sample consisted of 64 older women (ages 66+) who were clients at food pantries of the local food bank who participated in a larger study of individuals living with food insecurity (32). Age was only captured in clusters; thus, only women who selected the “age 66 or over” option were included in this study. Younger participants and men were not included. The majority of participants identified as Hispanic (65.1%; Table 1). Over a third (39.1%) reported disabled status, and 48.4% had less than a high school education or equivalent (GED). Almost half (46.9%) reported a household annual income of less than $10,000/year. Access to current health data is often low among individuals living with extreme poverty; scales are considered luxury items and often are not in the home (32). Thus, requesting current height and weight in this sample was likely to elicit inaccurate data. Therefore, we do not have BMI data in this sample.
Procedure
This study received IRB approval and was run in collaboration with the local food bank. See Becker et al. (32) for details regarding the full study and community partnership. All questionnaires were available in either English or Spanish. Following informed consent, participants completed questionnaires in person; undergraduate research assistants facilitated reading as needed (i.e., read questions aloud in language of choice) and answered questions. Participants received a $5 gift card to a local grocery store as compensation.
Measures
Consistent with working with a marginalized, low-education population and incorporating a socially conscious lens (33,34), we evaluated survey complexity. Guided by a colleague with extensive experience working with marginalized populations, we employed best practices delineated by Stonewall and colleagues (35) to evaluate and modify questionnaire language. We modified or removed items based on reading level or that may impact comprehension (for detailed rationale and procedure for measures modification, see Becket et al. (32)). This process ensured that our survey used appropriate language in order to be inclusive and culturally sensitive (33).
Food Insecurity
To assess severity of food insecurity, we used the Radimer Cornell Food Insecurity Measure (RCFIM: 36; 37). This measure uses a Likert scale, and global scores indicate level of food insecurity: 1) not food insecure (i.e., did not meet criteria for food. insecurity); 2) household food insecurity (i.e., anxiety about food, eating the same thing repeatedly due to lack of resources, and food running out but no one going hungry in the home); 3) individual food insecurity (i.e., adult reports being hungry at times because they lack food); and 4) child hunger food insecurity (i.e., adult reports inability to adequately feed their children secondary to lack of resources). For this study, we only included participants in the two highest levels of food insecurity (individual and child hunger), as these are the most severe. Internal consistency was good in our sample (Cronbach’s α = .92).
Binge Eating (BE)
We used the BE item from the Eating Disorder Diagnostic Scale for DSM 5 (EDDS-5) to assess frequency of BE over the past month (38). The EDDS is a brief self-report measure designed to assess the spectrum of EDs. The BE item asks, “How many times in the past month (30-31 days) on average have you eaten an unusually large amount of food and experienced a loss of control?” A report of ≥ 4 times in the past month equated to weekly binge eating.
Internalized Weight Stigma
We used the Weight Self-Stigma Questionnaire (WSSQ) to assess internalization of weight bias (39). Items are rated on a 7- point Likert scale (1 = strongly disagree to 7 = strongly agree), and participants rate their level of agreement with explicit weight bias statements (e.g., “I became fat because I’m a weak person” and “People think that I am to blame for being fat.”). Scores are summed for a total score and higher scores indicate greater self-stigma; internal consistency was excellent (α = .97).
Anxiety
We used eight items from the Penn State Worry Questionnaire (PSWQ; 40) to assess anxiety/worry. Items are rated on a 5-point Likert scale (1 = not at all typical of me to 5 = very typical of me); higher scores indicate more anxiety/worry. Internal consistency for this version of the PSWQ in the current sample was excellent (α = .95).
Results
Similar to the index study, we used descriptive statistics to evaluate BE prevalence and frequency in this replication sample. We used bivariate correlations of BE frequency with anxiety and internalized weight stigma. Finally, we examined rates of unhealthy weight control behaviors (e.g., self-induced vomiting, use of laxatives/diuretics for weight control) in this sample. Regarding BE prevalence, 20.4% of this sample reported BE on a weekly or more basis in the past month. Frequency of BE was positively correlated with higher anxiety (r = .37, p = .008) and internalized weight stigma (r = .42, p = .002). Within this sample of older women living with significant food insecurity, 7.9% of women reported self-induced vomiting, and 11% reported use of laxative or diuretics for weight control in the past month.
Study 3 (Replication): Sample with High Education Levels
Participants
Participants were 100 women recruited online for a larger study of body image in a diverse sample of adult women. Participant ages ranged from 55-79 years (M = 60.57, SD = 5.05) and reported a mean body mass index of 26.62 (SD = 6.04). The majority (72.0%) self-identified as White; 2.0% identified as Hispanic/Latina, and 72% were currently married (Table 1). Notably, 50% of women in this sample had a Masters or Doctoral degree.
Procedure
IRB approval was granted for this study and participants were recruited via email, social media, personal and professional networks, and by word of mouth. Recruitment emails described the study as exploring body image and wellness in a diverse population of adult women. All emails and posts requested that women forward the study invitation to their own social and professional networks (i.e., snowball sampling). After providing informed consent, participants completed self-report questionnaires online and had the option to provide their email address to enter a raffle for a $200 Amazon gift card.
Measures
Binge Eating (BE)
We assessed BE with the diagnostic items of the EDE-Q (29), which is a self-report measure of eating behaviors and attitudes over the past 28 days. The BE items inquire about frequency of BE episodes over the past 28 days. A report of ≥4 times in the past four weeks (28 days) equated to weekly BE.
BMI
Participants self-reported their height and weight in order to calculate BMI.
Depressive Symptoms
We used the Beck Depression Inventory- II (BDI-II; 41) to measure depressive symptoms. Due to liability in assessing suicidality anonymously and via online survey, we removed the suicidality item from the BDI-II. Thus, our final measure included 20 items and had good internal validity in the current sample (α = .85).
Body Shame
We administered the Shame subscale of the Objectified Body Consciousness Scale (OBCS; 42). This subscale includes eight items rated from 1 (strongly disagree) to 7 (strongly agree), with a “not applicable” option for items that do not apply. Higher scores indicate greater shame, and 25% or more of items NA or missing qualifies as missing overall score. In the present sample, internal consistency was good (α = .85).
Results
Similar to the previous studies, we used descriptive statistics to evaluate BE prevalence and frequency in this second replication sample. We used bivariate correlations of BE frequency with BMI, depressive symptoms, and body shame. Finally, we used a binomial logistic regression model to investigate the degree to which BE frequency contributed to risk for obesity status while controlling for age, race, and ethnicity. In this sample of women with high education levels, 19% reported BE weekly or more in the past month. BE frequency was positively correlated with greater depressive symptoms (r = .36, p < .001) and higher body shame (r = .44, p < .001), but was not correlated with BMI (r = .20, p = .075). Finally, BE frequency contributed a small, but statistically significant risk for obesity status (OR = 1.12; 95% CI [1.00, 1.24], p = .04). Neither age nor race/ethnicity were significant in the model.