Ecological Momentary Assessment of the Snacking Environments of Children from Racially/Ethnically Diverse Households

Abstract

Children consume nearly one-third of their daily energy intake as snacks (i.e., eating occasions that occur between meals); thus there is a growing interest in understanding what snacking occasions look like in the homes of young children. This study makes use of ecological momentary assessment (EMA) to 1) examine differences in the contextual factors, including location, food preparation style, people present, presence of media devices, and overall atmosphere, between meal and snack occasions; and 2) explore differences in the context of snacking occasions across children’s gender and weight status. Data for the current study came from the Family Matters Study, which included 150 families with children aged 5–7 years old (n = 25 from each of the following groups: Black/African American, Hispanic, Hmong, Native American, Somali, White). Parents completed an 8-day EMA observation period, during which they were surveyed after each eating occasion with the study child; questions explored contextual factors including location, food preparation style, people present, presence of media devices, and the overall atmosphere of each eating occasion. Differences between meals and snacks were observed; a smaller percentage of snacks (compared to meals) were prepared by the parent, consisted of only homemade food, and were planned ahead of time, as opposed to being served in response to a child’s request. Snacks were more likely than other meals to be eaten on the couch and in the presence of a screen. Furthermore, important differences in snacking context were observed by child gender and weight status. Findings illuminate opportunities to improve children’s overall dietary intake via interventions focused on improving the quality of foods served during snacks, as well as the contextual environment in which snacks are eaten.

The overall diet quality among children in the United States (US) and many other countries is poor, particularly among low income and racial/ethnic minority subgroups, making the identification of ways to intervene on and improve children's eating patterns and dietary intake a public health priority.¹ Because US children consume nearly one-third of their daily energy intake as snacks (e.g. eating occasions that occur between meals), researchers have begun to explore how snacking contributes to the overall quality of children's dietary intake.^2-5 Snacking represents a key source of the high energy, lower nutrient-dense foods consumed by children⁶; several studies have identified desserts, salty snack foods, and sugar-sweetened beverages, as top foods consumed by children during snacking occasions.^4,7-9 At the same time, snacks also offer a potential opportunity for children to consume more healthful foods. Indeed, several research studies, including one by our study team,¹⁰ have found that foods consumed during snacking occasions contribute positively to children’s intake of necessary vitamins, minerals, fruit, dairy and protein foods.^2,4,7,9,11 Results from our study team were quite nuanced: among children who consumed snacks, snacking was found to be an important source of fruit and dairy, but also contributed significantly to children’s consumption of refined grains and sugar-sweetened beverages. Overall, snacks contributed positively to children’s overall Healthy Eating Index. Because snacks substantially contribute to children’s overall caloric intake, and because snacks consumed by children often contain at least some high energy, lower nutrient-dense foods, there is a growing interest in better understanding of what snacking occasions look like in the homes of young children.

To date, the bulk of research on snacking among young children has focused on the types of foods served and consumed, as well as the predictors of engaging in snacking behaviors, including parenting practices (e.g., restriction, indulgence) and the presence of unhealthy foods within the home food environment.¹² For example, a review of the snacking literature conducted by Blaine and colleagues reported that both parental restriction of food and availability of unhealthy foods within the home environment were positively associated with increased consumption of snacks by children. Blaine also found that within the literature to date, snacking was often assessed using food frequency items and defined post-hoc based on nutritional characteristics of foods eaten (e.g. energy-dense, sugary foods, unhealthy, etc.). The timing was rarely included in the definition of a snack (i.e. chips eaten between meals vs. with lunch). The choice of how to define snacking (i.e. timing of eating vs type of food consumed) within research studies has important implications for how snacking research can be interpreted. Lack of a standard snacking definition contributes, in part, to some of the disparate findings regarding the impact of snacking on dietary intake and weight status.⁷

A small, but emerging, body of research suggests that parents of young children may approach snack times differently than mealtimes with their child, including marked differences in non-nutritive practices. ^12-16 For example, a qualitative study conducted by Davison and colleagues found that the low-income, non-Hispanic white, African American, and Hispanic caregivers in their sample reported frequent use of permissive feeding practices during snacks.¹³ Specifically, parents were less likely to have specific rules or limits around snacks than around meals, were less involved in the regulation of their child's snack intake, and reported context-driven (e.g. pressure from a grandparent) and emotion-based (e.g. snacks to show love, celebrate) provision of snacks to their child.¹³ These qualitative findings suggest that parents might rely more heavily on the use of permissive feeding practices in the context of snacking occasions, as compared to mealtimes. Indeed, findings from a series of qualitative focus groups conducted by Fisher (fourth author) and colleagues revealed that parents are often more flexible with the quality of food served during snacking occasions and tend to think of snacking occasions as serving an inherently different purpose than that of meals; mothers reported that snacks often served a more important role in management of children’s behavior than in providing nutrition.¹⁵ More recently, Loth (first author) and colleagues conducted a study aimed at understanding how parents divide up the responsibilities of feeding and eating (e.g. what, when, where, whether, and how much) with their preschool-aged child; interviews revealed stark differences in how parents handled the responsibilities of feeding at mealtime, as compared to snack time.¹⁶ Overall, parents allowed their child more freedom to dictate what foods were eaten for snacks, when snacks were offered and eaten, and where snacks were consumed, whereas they took primary responsibility for making these same decisions for their child during mealtimes. Findings from these studies all point to the idea that parents’ approach to snacks differs substantially from their approach to meals, and that this difference in approach might impact the quality of foods consumed by children during snacking occasions.^13-16 However, to our knowledge, there has not been quantitative research that directly compares the contextual characteristics (e.g. location, structure, who is present, overall atmosphere) of snacking occasions (defined by timing) to the contextual characteristics of other meal times, to illuminate these different parental approaches.

A clearer understanding of how snacking occasions, as compared to other meal occasions, look within the homes of young children, will provide insight to clinicians and public health professionals regarding the types of interventions that might be needed to improve the healthfulness of foods consumed as snacks. Given the impact of snacks on children’s overall caloric intake, it is of interest to understand if the context of snacking occasions looks similar or different within the homes of children of varying weight status. Additionally, given the emergence of evidence suggesting gender differences in foods consumed during snacking occasions¹⁰, an examination of potential differences in the context of snacking occasions is warranted. Finally, an exploration of snacking patterns within a racially/ethnically and socioeconomically diverse population of young people might help to provide insight into observed disparities in dietary intake and weight status across population-level subgroups.

Thus, this study makes use of ecological momentary assessment (EMA) to explore the following study aims within a racially/ethnically and socioeconomically diverse population of boys and girls: 1) to examine differences in the contextual factors, including location, food preparation style, people present, presence of media devices, and overall atmosphere, between meal and snack occasions; and 2) to explore differences in the context of snacking occasions across children’s gender and weight status. EMA methodology is a powerful technique that uses a smartphone-based Web application to record details about behaviors (e.g. meals or snacks consumed) and/or details about the context in which these behaviors occur (e.g. location, people present, mood). These recordings are made in the moment and the participant's daily life, allowing researchers to observe behaviors as they unfold in the real world, moment-by-moment, to capture dynamic changes in behavior over time and across contexts. This approach avoids the limitations inherent in other survey-based measures; for example, asking a parent to report on the “usual” location where their child eats snacks, might yield a very different response than asking the parent to report on the location where their child ate each snack throughout a particular day or week. Study findings will expand upon our current understanding of what snacking occasions look like among school-aged children from low-income and racially/ethnically diverse households, including describing how snacking occasions differ from meal occasions and will illuminate opportunities for potential nutrition interventions aimed at promoting healthful snacking patterns as a means of improving children’s overall dietary intake.

Methods

Data for the current study are from Family Matters a National Institutes of Health-funded study, Family Matters is a 5-year incremental (Phase I = 2014-2016.; Phase II = 2017-2019), mixed-methods (e.g., video-recorded tasks, EMA, interviews, surveys) longitudinal study designed to identify novel risk and protective factors for childhood obesity in the home environments of racially/ethnically diverse and primarily low-income children.¹⁷ Phase I included an in-depth, mixed-methods, cross-sectional examination of the family home environment of diverse families (n=150). Phase II was a longitudinal epidemiological cohort study with diverse families (n=1200). In-depth details regarding both Phases of the Family Matters study have been published elsewhere.¹⁷

Data utilized in the current analysis are from Phase I of the Family Matters study. In Phase I, a mixed-methods analysis of the home environments of racially/ethnically and socioeconomically diverse children ages 5-7 years old was conducted to identify individual, dyadic, and familial risk and protective factors for childhood obesity. The University of Minnesota Institutional Review Board Human Subjects Committee approved all protocols used in both phases of the Family Matters study.

Recruitment and Eligibility Criteria

Eligible children (n=150) and their families were recruited from primary care clinics in the Minneapolis/St. Paul metropolitan area between 2015-2016. Families of children who had a recent well-child visit and reported being part of one of the six racial/ethnic groups included in this study (non-Hispanic white, African American, Hispanic, Native American, Hmong, and Somali) were sent a letter by their family physician asking them to participate in the Family Matters study. Children were eligible to participate in the study if they were between the ages of 5 and 7 years old, had a BMI percentile ≥ 5, had a sibling between the ages of 2 and 12 years old living in the same home, lived with their parent/primary guardian more than 50% of the time, shared at least one meal (home-cooked or otherwise) per day with the parent/primary caregiver, were away from home during the day (e.g. school, summer camp), and were from one of six racial/ethnic categories (non-Hispanic white, African American, Hispanic, Native American, Hmong, and Somali)ere In total, 150 families with children ages 5–7 years old from six ethnic/racial groups participated in the study (n = 25 from each; African American, Hispanic, Hmong, Native American, Somali, non-Hispanic white). Further, the study implemented a stratified sampling strategy; within each ethnic/racial group, half of the sample children recruited had a body mass index (BMI) ≥85 percentile while the other half had a BMI between 5 and 85 percentile. The sample was intentionally stratified by race/ethnicity and weight status (overweight/obese=BMI ≥85%ile; non-overweight=BMI >5%ile and <85%ile) to identify potential weight- and/or race/ethnic-specific home environment factors related to obesity risk. Additional, in-depth details regarding recruitment and study design are published elsewhere.¹⁷

Procedures and Data Collection

A 10-day observational study was conducted with each family, including two in-home visits, three 24-hour dietary recalls, and an 8-day direct EMA observational period in between the home visits. While the Family Matters Study collected many measures described elsewhere,¹⁷ the measures used in this analysis (described below) come from direct measurement of height and weight of the study child and parent respondent by trained staff at the first home visit, from a single online survey completed by the parent at the second home visit, and from an 8-day ecological momentary assessment (EMA) conducted in between home visits. During the 8-day EMA observation period, parents filled out an EMA survey after each meal (defined as breakfast, lunch, dinner, or snack) eaten with the study child on a study-provided iPad mini. Parents were required to complete at least one mealtime survey per day, however, parents completed on average 3 mealtime surveys per day. The average mealtime survey took participants three minutes to complete. EMA survey measures were identified by examining pre-existing, validated instruments and adapting it for EMA, as detailed below.^17-21 All study materials were translated into Spanish, Somali, and Hmong and bilingual staff were available at all home visits, allowing families to participate in their preferred language. Because of the level of study involvement required, families were compensated with the iPad mini used to record EMA data (~ $300) and additional gift card opportunities (up to $100) if all elements of the study were completed (e.g., all days of EMA, all three dietary recalls).

Sample demographics.

The study sample included diverse families who were equally distributed across the six racial/ethnic groups recruited in the study (African American, American Indian, Hispanic, Hmong, Somali, White). Additionally, families were from low-income households, with 70% of families earning less than $35,000 per year. The majority of participants were mothers (91%) who were approximately 35 years old (mean = 34.5; sd = 7.1) with children aged 6 years old (mean = 6.4; sd= 0.08). Over half of the mothers worked full or part-time and 61% had a high school diploma or less. About half of the mothers were married and 64% of households had two parents. Additional participant details can be found in Table 1.

Table 1.

Family Matters Study Phase I Demographic Characteristics (N=150)

	Primary Caregiver (n=150)	Study Child (n=150)
Participant Characteristics
	N (%)	N (%)
Female	137 (91%)	71 (47%)
Age in years (sd)	34.5 (7.1)	6.4 (0.8)
Adult BMI (BMI %ile child)	30.9 (7.2)	75.9 (23.1)
Weight Status
Nonoverweight	35 (23%)	77 (51%)
Overweight/Obese	115 (77%)	73 (49%)
Race/Ethnicity
American Indian or Alaskan Native	21 (14%)	25 (17%)
Asian	25 (17%)	25 (17%)
Black or African American	22 (15%)	25 (17%)
White	27 (18%)	25 (17%)
Somali	25 (17%)	25 (17%)
Hispanic	23 (15%)	25 (17%)
Mixed/Other	7 (5%)	--
Household Characteristics
Household Structure
One Parent (no other adults)	37 (25%)
One Parent (w/ other adults)	18 (12%)
Two Parents (no other adults)	78 (52%)
Two Parents (w/ other adults)	17 (11%)
Primary Caregiver Marital Status
Married	78 (51%)
Committed dating relationship or engaged	31 (21%)
Casually dating	2 (1%)
Separated or Divorced	6 (4%)
Widowed	1 (1%)
Single/never married	31 (21%)
Missing	1 (1%)
Educational Attainment
Middle school or junior high	15 (10%)
Some high school	17 (11%)
High school or GED	60 (40%)
Vocational, technical, trade certificate program, or associate degree	28 (18%)
College degree (e.g., bachelor, masters, or doctorate)	25 (16%)
Other	4 (3%)
Missing	1 (1%)
Primary Caregiver Work Status
Working full-time	63 (42%)
Working part-time	32 (21%)
Stay at home caregiver	25 (17%)
Currently unemployed, seeking work	18 (12%)
Not working for pay (unable to work, retired, student, etc)	11 (7%)
Not Applicable	1 (1%)
Household Income
Less than $20,000	50 (33%)
$20,000 - $34,999	55 (37%)
$35,000 - $49,999	16 (11%)
$50,000 - $74,999	12 (8%)
$75,000 - $99,999	7 (5%)
$100,000 or more	9 (6%)
Missing	1 (1%)
Household Income Source (all that apply)
Wages from self	85 (57%)
Wages from other guardian	34 (23%)
Another family member	14 (10%)
Unemployment compensation	4 (3%)
Worker's compensation	0 (0%)
Social Security	11 (7%)
Public assistance (e.g., MFIP)	34 (23%)
Alimony/Child support	17 (11%)
Other sources	21 (14%)

Note: Percentages may not sum to exactly 100% due to rounding

Measures

Ecological momentary assessment (EMA).

Multiple daily measures of EMA over eight days were collected on parents, allowing for participants to respond to questions about variables such as parenting practices, eating, and physical activity behaviors, mood, and stress, within different contexts that occurred day-to-day, moment-by-moment, in families’ lives. Standardized EMA data collection protocols from prior studies were used in the study including: (1) signal contingent (i.e., researcher-initiated surveys completed randomly throughout the day; questions on these surveys were focused on parent mood and stress and recent parent/child activities) and (2) event contingent (participant-initiated surveys completed after the parent and child ate a meal or snack together; questions on these surveys focused on foods consumed as well as the meal/snack environment and context) EMA surveys.^22-24 iPad minis were provided to parents to enter responses to the EMA surveys during the eight-day observation period. All EMA responses were time-stamped. Participants’ were assigned additional days of EMA if several EMA prompts were missed within a day to obtain a minimum of eight full days of EMA data with at least four complete EMA surveys per day (i.e., at least 2 signal contingent surveys; at least 1 event contingent survey; 1 end-of-day survey). The current study uses data collected from the event contingent EMA surveys as these asked questions focused on foods consumed as meals and snacks, as well as questions on the environment and context in which the meal or snack was consumed.

Event contingent surveys

were self-initiated by parents whenever they shared an eating occasion with the child participant. Parents were asked to use their iPad mini to respond to the following quantitative questions about eating occasion characteristics: 1) what type of eating occasion it was (breakfast; lunch; dinner; snack), 2) who prepared the food (Myself; My partner/spouse; A child in the household; Another adult in the household; Other person(s) (e.g., potluck, visiting a friend); Food establishment (e.g., fast food, restaurant, grocery store deli); Other), 3) how was the food prepared (Fast food/take-out (eating at home or at a restaurant); Pre-prepared foods (e.g., macaroni and cheese, frozen meals) or purchased snacks (e.g., fruit snacks, chips, granola bars, cereal); Homemade/freshly prepared (include fresh fruits or vegetables)), 4) what influenced the person’s decision to serve the food at the eating occasion (Quick and easy to make; Child/family likes; Child asked for a specific food or meal; Desire to avoid conflict with child or a family fight; It was food I had available at home; It was a healthy option; Stressful day/busy schedule; Too tired to cook; It was a planned meal; It was available at the place we ate (e.g., restaurant, celebration/gathering); Other), 5) how many children were present (1-10+), 6) who were the children (Child in the study; Older sibling(s); Younger sibling(s); Other family members (e.g., cousin); Non-family members (e.g., friend, neighbor)), 7) how many adults were present (1-10+), 8) who were they (Main parent; Other caregiver (e.g., partner, spouse); Other family members (e.g., grandparent, aunt, uncle); Non-family members (e.g., friend, neighbor)), 9) where did the meal/snack take place (Around a table or counter at home; On couch/chair in living area; Scattered throughout house; Standing up; In the car; At a restaurant; Other), 10) what was happening during the meal/snack (Conversation; Watching TV; TV on in background; Playing a video game; Using a cell phone; Using a tablet; Using a computer; Reading/looking at a book; Listening to headphones; None of the above), and 11) what was the emotional atmosphere like (Chaotic; Rushed; Tense; Relaxed; Enjoyable; Neutral)? These EMA questions were developed for the Family Matters study based on qualitative data showing the importance of collecting more detailed information about contextual characteristics of eating occasions collected from parents as a part of a previous study the authors conducted.²⁵

Ethnicity/Race, Sex, and Age.

One of the criteria for study eligibility was pre-stratification on race/ethnicity across six racial/ethnic groups (i.e, African American, American Indian, Hispanic/Latino, Hmong, Somali, white). Parents self-reported their own and their child’s race/ethnicity according to the six racial/ethnic groups. Parent and child age was calculated using self-reported birth date and survey completion date. Parent and child sex was determined by parent self-report of their own and the child’s sex.

Statistical Analysis

The current study described snacking occasion characteristics including food preparation, logistics, and contextual features, in addition to examining differences between characteristics of meals versus snacking occasions (aim 1). The second aim examined whether snacking occasions differed for boys or girls and for nonoverweight or overweight children. For each aim, the overall frequency of these characteristics was calculated in the sample, and between-child and within-child statistics were computed from the EMA data to characterize how common or rare features of meals were in the sample and how infrequent or frequent these characteristics were reported at within-child occasions. Meal characteristics were dummy-coded (dichotomous dependent variable) and generalized estimating equations were fitted to deal with correlated participant error terms to examine differences in the reported prevalence of each characteristic for snacks relative to meal occasions, boys compared to girls, and overweight children compared to nonoverweight children. Huber-white robust standard errors were computed, a binomial variance family was specified, and an independence working correlation structure was set. Significance testing was performed after controlling for parent and child sex, child age and race/ethnicity, household structure (e.g., single parent with/without other adults and multiple parents with/without other adults) and reported annual household income. Because the aims of this manuscript were primarily descriptive, adjustments for multiplicity were not made and the significance level was set at P<0.05. All data management and statistical analysis were performed in Stata 15.1 MP (College Station, TX).

The a priori decision to focus on differences by weight status and sex of the child was based on previous research suggesting there are important potential differences in snacking behaviors across these child characteristics that should be examined further.¹⁰ The decision to not additionally stratify by race/ethnicity or immigrant status was made out of concern that cell sizes for certain subgroups would be too small.

Results

On average, children in our sample reported engaging in 1.46 (SD:0.77; Range: 1-7) snacking occasions per day. On the bulk of days examined (90.2%) children engaged in fewer than three snacking occasions (i.e., one or two); on only 2.4% of days did children engage in more than three snacking occasions (i.e., four to seven). The average number of snacking occasions per day did not differ by child sex (p=0.94); both boys and girls engaged in an average of 1.46 snacking occasions daily (SD boys: 0.76, SD girls: 0.78). The average number of snacking occasions per day did not differ by child weight status (p=0.13); children whose weight fell into a non-overweight range engaged in an average of 1.46 (SD: 0.81) snacking occasions daily and children whose weight fell into the non-overweight range engaged in an average of 1.41 (SD: 0.72) snacking occasions daily.

Contextual Differences Between Meal and Snack Eating Occasions

Several significant differences in the contextual factors surrounding meals and snacks were found; see Tables 2-4 for the full results. With regard to food preparation, foods consumed at snacking occasions were more likely to be prepared outside of the home (e.g., gas station, fast food establishment, restaurant) (46%, as compared to only 32% of meals). Furthermore, a greater percentage of meal occasions consisted of only homemade foods as compared to snacking occasions (55% vs. 38%, respectively), whereas nearly half of snacking occasions (48%) consisted of only pre-prepared foods, as compared to only 23% of meal occasions.

Table 2.

Snacking Meal Characteristics Compared with Breakfast, Lunch, and Dinner Meal Characteristics (N=1,080 Snacking Occasions and N=2,759 Non-Snacking Meal Occasions)

Food Preparation	All Non-Snacking Meal Occasions (Breakfast, Lunch, and Dinner)			Snacking-Only Occasions
	N Occasions (Overall %)a	N Participant (Between %)b	Average Participant Frequency (Within %)c	N Occasions (Overall %)a	N Participant (Between %)b	Average Participant Frequency (Within %)c	P Valued
Who prepared the meal?
Parent/partner	1765 (64%)	147 (98%)	65%	583 (54%)	124 (83%)	63%	<0.001
A cook outside home (e.g. fast food, restaurant, another adult)	511 (19%)	124 (83%)	24%	269 (25%)	94 (63%)	37%	0.01
Child participant + parent/partner	483 (18%)	112 (75%)	23%	228 (21%)	78 (52%)	40%	0.02
How was the meal prepared?
Only homemade	1531 (56%)	146 (97%)	56%	411 (38%)	411 (274%)	47%	<0.001
Only pre-prepared	641 (23%)	129 (86%)	29%	515 (48%)	515 (343%)	54%	<0.001
Only fast-food	347 (13%)	104 (69%)	19%	94 (9%)	94 (63%)	27%	0.002
Combination of homemade, pre-pared, fast-food	239 (9%)	72 (48%)	17%	60 (6%)	60 (40%)	24%	0.01
What were the primary decisions about what was served?
Child / family likes	561 (20%)	124 (83%)	24%	245 (23%)	89 (59%)	38%	0.24
Child asked for a specific food or meal	283 (10%)	107 (71%)	16%	228 (21%)	80 (53%)	37%	<0.001
Desire to avoid conflict with child or a family fight	76 (3%)	19 (13%)	19%	33 (3%)	15 (10%)	28%	0.90
Healthy option	369 (13%)	93 (62%)	19%	143 (13%)	67 (45%)	29%	0.67
Planned meal	433 (16%)	115 (77%)	22%	48 (4%)	35 (23%)	17%	<0.001
Using food available in the home	435 (16%)	113 (75%)	21%	135 (13%)	63 (42%)	27%	0.04
Quick and easy to make	327 (12%)	107 (71%)	17%	134 (12%)	68 (45%)	27%	0.66
Other	274 (10%)	106 (71%)	16%	114 (11%)	58 (39%)	27%	0.86
Logistics of the Eating Occasion	All Non-Snacking Meal Occasions (Breakfast, Lunch, and Dinner)			Snacking-Only Occasions
	N Occasions (Overall %)a	N Participant (Between %)b	Average Participant Frequency (Within %)c	N Occasions (Overall %)a	N Participant (Between %)b	Average Participant Frequency (Within %)c	P Valued
How many children were present?
Child participant	303 (11%)	86 (57%)	21%	181 (17%)	71 (47%)	37%	0.002
Two children	995 (36%)	113 (75%)	46%	393 (36%)	102 (68%)	51%	0.63
Three children or more	1461 (53%)	134 (89%)	59%	506 (47%)	105 (70%)	63%	0.11
Who were these children?
Child participant	300 (11%)	86 (57%)	21%	181 (17%)	71 (47%)	37%	0.001
Child participant + siblings(s)	2077 (75%)	137 (91%)	80%	736 (68%)	127 (85%)	75%	0.003
Child participant + siblings(s)+others	379 (14%)	107 (71%)	21%	163 (15%)	67 (45%)	33%	0.72
Note for others: extended family members or non-family members
How many adults were present?
Parent participant	1300 (47%)	139 (93%)	52%	617 (57%)	130 (87%)	64%	<0.001
Parent participant + another adult(s)	1459 (53%)	138 (92%)	57%	463 (43%)	111 (74%)	55%	<0.001
Who were these adults?
Parent/primary caregiver	1294 (47%)	139 (93%)	51%	616 (57%)	130 (87%)	64%	<0.001
Parent + other family member(s)	1355 (49%)	137 (91%)	53%	421 (39%)	103 (69%)	53%	<0.001
Parent + other family member + non-family member	105 (4%)	50 (33%)	12%	42 (4%)	29 (19%)	22%	0.82
Note for other family member: second caregiver and grandparent
Features of the Eating Occasion	All Non-Snacking Meal Occasions (Breakfast, Lunch, and Dinner)			Snacking-Only Occasions
	N Occasions (Overall %)a	N Participant (Between %)b	Average Participant Frequency (Within %)c	N Occasions (Overall %)a	N Participant (Between %)b	Average Participant Frequency (Within %)c	P Valued
Where did the meal take place?
Around the table	2153 (78%)	147 (98%)	79%	488 (45%)	127 (85%)	55%	<0.001
On couch/chair in living area	281 (10%)	67 (45%)	23%	284 (26%)	91 (61%)	42%	<0.001
Scattered throughout house or standing-up	122 (4%)	53 (35%)	13%	165 (15%)	60 (40%)	32%	<0.001
Outside house (e.g. in the car, at a restaurant, other)	202 (7%)	81 (54%)	15%	143 (13%)	59 (39%)	30%	<0.001
Note for other: at a park, school's picnic table/event, neighbor's home, relative's home, or friend's home.
What kinds of foods were eaten?
All Grains	1832 (66%)	147 (98%)	37%	343 (32%)	107 (71%)	41%	<0.001
Dairy	1253 (45%)	140 (93%)	49%	331 (31%)	108 (72%)	43%	<0.001
Fruit and Vegetables	1597 (58%)	146 (97%)	58%	528 (49%)	120 (80%)	55%	<0.001
Meat, Beans, and Nuts	1538 (56%)	144 (96%)	59%	200 (19%)	81 (54%)	33%	<0.001
Added Sugar (e.g., sugary drinks, cake, candy)	354 (13%)	107 (71%)	18%	259 (24%)	96 (64%)	34%	<0.001
What was happening during the meal or snack?
Conversation only	1502 (54%)	140 (93%)	58%	437 (40%)	116 (77%)	52%	<0.001
Screentime activities only	458 (17%)	89 (59%)	27%	255 (24%)	92 (61%)	38%	<0.001
Screentime + conversation	485 (18%)	102 (68%)	27%	212 (20%)	73 (49%)	37%	0.43
Conversation + other	57 (2%)	22 (15%)	12%	25 (2%)	12 (8%)	22%	0.57
No conversation + other	256 (9%)	73 (49%)	20%	151 (14%)	59 (39%)	31%	<0.001
Note for screentime activities: Listening to headphones; playing videogames; watching TV and/or TV in background; using cell phone, computer, or tablet
Note for other: Reading/looking at a book or none of the above
What was the atmosphere like?
Relaxed	763 (28%)	122 (81%)	36%	342 (32%)	107 (71%)	43%	0.02
Enjoyable	1244 (45%)	141 (94%)	48%	479 (44%)	115 (77%)	54%	0.42
Neutral	491 (18%)	97 (65%)	25%	180 (17%)	65 (43%)	38%	0.70
Rushed	133 (5%)	56 (37%)	12%	43 (4%)	26 (17%)	24%	0.54
Chaotic	74 (3%)	30 (20%)	14%	21 (2%)	11 (7%)	24%	0.40
Tense	53 (2%)	29 (19%)	10%	15 (1%)	13 (9%)	16%	0.22

^a.Overall tabulations count the frequency of response at the sample population level. For example, of the 2,759 breakfast, lunch, or dinner occasions, 64% (N=1,765) were prepared by the parent or partner.

^b.Between tabulations compute the fraction of the full sample (N=150 participants) who reported a response at least once for the respective meal occasion. For example, 147 participants had at least one non-snacking meal occasion in which the parent or partner prepared a meal, which means that three participants never had a meal prepared by a parent/partner over the course of the study. Only 29 families had at least one meal occasion they considered to be a tense atmosphere.

^c.Within-participant statistics report the fraction of the time that participants reported the same meal characteristic over multiple meals. For example, the most frequent response about the snacking atmosphere was “enjoyable” (44%) and 77% of the sample (N=115) reported the meal characteristic at least once. Of the 115 participants who had an “enjoyable” snacking occasion, they reported the same response about other snacking occasions 54% of the time -- approximately 1 in every 2 snacks.

^d.Response values were dummy coded to test if the probability of the response differed between snacks and non-snacking occasions. Generalized estimating equation models were fitted to account for repeated observations and adjusted for parent and child sex, child age and race/ethnicity, household structure (e.g., single parent with/without other adults and multiple parents with/without other adults) and reported annual household income.

Table 4.

Snacking Meal Characteristic Differences Stratified by Child Overweight and Non-Overweight Status (N=1,080 Snacking Occasions; N=73 Overweight and N=77 Nonoverweight)

Food Preparation	Overweight (N=73)			Nonoverweight (N=77)
	N Occasions (Overall %)a	N Participant (Between %)b	Average Participant Frequency (Within %)c	N Occasions (Overall %)a	N Participant (Between %)b	Average Participant Frequency (Within %)c	P Valued
Who prepared the meal?
Parent/partner	258 (53%)	58 (79%)	65%	325 (55%)	66 (86%)	61%	0.70
A cook outside home (e.g. fast food, restaurant, another adult)	128 (26%)	44 (60%)	41%	141 (24%)	50 (65%)	33%	0.47
Child participant + parent/partner	102 (21%)	37 (51%)	41%	126 (21%)	41 (53%)	39%	0.85
How was the meal prepared?
Only homemade	165 (34%)	59 (81%)	44%	246 (42%)	60 (78%)	50%	0.19
Only pre-prepared	260 (53%)	60 (82%)	56%	255 (43%)	58 (75%)	52%	0.03
Only fast-food	41 (8%)	23 (32%)	26%	53 (9%)	29 (38%)	29%	0.40
Combination of homemade, pre-pared, fast-food	22 (5%)	19 (26%)	27%	38 (6%)	23 (30%)	21%	0.11
What were the primary decisions about what was served?
Child / family likes	95 (19%)	43 (59%)	37%	150 (25%)	46 (60%)	40%	0.04
Child asked for a specific food or meal	92 (19%)	36 (49%)	38%	136 (23%)	44 (57%)	36%	0.34
Desire to avoid conflict with child or a family fight	5 (1%)	4 (5%)	15%	28 (5%)	11 (14%)	33%	0.04
Healthy option	70 (14%)	34 (47%)	33%	73 (12%)	33 (43%)	24%	0.27
Planned meal	28 (6%)	20 (27%)	19%	20 (3%)	15 (19%)	15%	0.13
Using food available in the home	58 (12%)	28 (38%)	30%	77 (13%)	35 (45%)	26%	0.93
Quick and easy to make	71 (15%)	32 (44%)	30%	63 (11%)	36 (47%)	24%	0.12
Other	69 (14%)	30 (41%)	28%	45 (8%)	28 (36%)	26%	0.02
Social Dimensions of the Eating Occasion	Overweight (N=73)			Nonoverweight (N=77)
	N Occasions (Overall %)a	N Participant (Between %)b	Average Participant Frequency (Within %)c	N Occasions (Overall %)a	N Participant (Between %)b	Average Participant Frequency (Within %)c	P Valued
How many children were present?
Child participant	86 (18%)	37 (51%)	40%	95 (16%)	34 (44%)	34%	0.37
Two children	158 (32%)	47 (64%)	48%	235 (40%)	55 (71%)	53%	0.46
Three children or more	244 (50%)	55 (75%)	61%	262 (44%)	50 (65%)	65%	0.85
Who were these children?
Child participant	86 (18%)	37 (51%)	40%	95 (16%)	34 (44%)	34%	0.37
Child participant + siblings(s)	318 (65%)	59 (81%)	73%	418 (71%)	68 (88%)	77%	0.21
Child participant + siblings(s)+others	84 (17%)	34 (47%)	39%	79 (13%)	33 (43%)	28%	0.39
Note for others: extended family members or non-family members
How many adults were present?
Parent participant	288 (59%)	64 (88%)	65%	329 (56%)	66 (86%)	63%	0.18
Parent participant + another adult(s)	200 (41%)	55 (75%)	53%	263 (44%)	56 (73%)	56%	0.18
Who were these adults?
Parent/primary caregiver	288 (59%)	64 (88%)	65%	328 (55%)	66 (86%)	63%	0.18
Parent + other family member(s)	181 (37%)	51 (70%)	52%	240 (41%)	52 (68%)	54%	0.21
Parent + other family member + non-family member	19 (4%)	13 (18%)	23%	23 (4%)	16 (21%)	21%	0.82
Note for other family member: second caregiver and grandparent
Features of the Eating Occasion	Overweight (N=73)			Nonoverweight (N=77)
	N Occasions (Overall %)a	N Participant (Between %)b	Average Participant Frequency (Within %)c	N Occasions (Overall %)a	N Participant (Between %)b	Average Participant Frequency (Within %)c	P Valued
Where did the meal take place?
Around the table	209 (43%)	67 (92%)	50%	279 (47%)	60 (78%)	60%	0.30
On couch/chair in living area	156 (32%)	49 (67%)	45%	128 (22%)	42 (55%)	37%	0.02
Scattered throughout house or standing-up	65 (13%)	22 (30%)	30%	100 (17%)	38 (49%)	33%	0.18
Outside house (e.g. in the car, at a restaurant, other)	58 (12%)	25 (34%)	34%	85 (14%)	34 (44%)	26%	0.95
Note for other: at a park, school's picnic table/event, neighbor's home, relative's home, or friend's home.
What kinds of foods were eaten?
All Grains	140 (29%)	47 (64%)	42%	203 (34%)	60 (78%)	40%	0.81
Dairy	147 (30%)	50 (68%)	44%	184 (31%)	58 (75%)	42%	0.92
Fruit and Vegetables	201 (41%)	58 (79%)	51%	327 (55%)	62 (81%)	59%	0.03
Meat, Beans, and Nuts	82 (17%)	34 (47%)	36%	118 (20%)	47 (61%)	32%	0.99
Added Sugar (e.g., sugary drinks, cake, candy)	127 (26%)	48 (66%)	37%	132 (22%)	48 (62%)	31%	0.92
What was happening during the meal or snack?
Conversation only	173 (35%)	53 (73%)	49%	264 (45%)	63 (82%)	55%	0.08
Screentime activities only	149 (31%)	48 (66%)	44%	106 (18%)	44 (57%)	33%	0.002
Screentime + conversation	87 (18%)	36 (49%)	37%	125 (21%)	37 (48%)	36%	0.54
Conversation + other	4 (1%)	3 (4%)	20%	21 (4%)	9 (12%)	23%	0.09
No conversation + other	75 (15%)	29 (40%)	34%	76 (13%)	30 (39%)	28%	0.62
Note for screentime activities: Listening to headphones; playing videogames; watching TV and/or TV in background; using cell phone, computer, or tablet
Note for other: Reading/looking at a book or none of the above
What was the atmosphere like?
Relaxed	150 (31%)	53 (73%)	44%	192 (32%)	54 (70%)	43%	0.54
Enjoyable	204 (42%)	56 (77%)	51%	275 (46%)	59 (77%)	56%	0.08
Neutral	98 (20%)	34 (47%)	42%	82 (14%)	31 (40%)	34%	0.02
Rushed	16 (3%)	11 (15%)	19%	27 (5%)	15 (19%)	28%	0.98
Chaotic	14 (3%)	5 (7%)	38%	7 (1%)	6 (8%)	13%	0.06
Tense	6 (1%)	6 (8%)	16%	9 (2%)	7 (9%)	16%	0.94

^a.Overall tabulations count the frequency of response at the sample population level. For example, of the overweight participants' 488 snacking occasions, 53% (N=258) were prepared by the parent or partner.

^b.Between tabulations compute the fraction of the participants in the weight status stratified sample (N=73 overweight and N=77 nonoverweight participants) who reported a response at least once for the respective snacking occasion. For example, 58 overweight participants (79%) had at least one snacking occasion in which the parent or partner prepared the snack, which means that 15 overweight participants never had a snack prepared by a parent/partner over the course of the study. Only 6 overweight participants had at least one snack that they considered to be a tense atmosphere.

^c.Within-participant statistics report the fraction of the time that participants reported the same snacking characteristic over multiple occasions. For example, the most frequent response about the snacking atmosphere among participants with girls was “enjoyable” (42%) and 77% of the stratified overweight sample (N=56) reported the snacking characteristic at least once. Of the 56 overweight participant who had an “enjoyable” snacking occasion, they reported the same response about other snacking occasions 51% of the time -- about 1 in every 2 snacks.

^d.Response values were dummy coded to test if the probability of the snacking response differed between boys and girls. Meal occasions were restricted to snacks and generalized estimating equation models were fitted to account for repeated observations and adjusted for parent and child sex, child age and race/ethnicity, household structure (e.g., single parent with/without other adults and multiple parents with/without other adults) and reported annual household income.

Differences with regard to who was present at the different eating occasions were observed; approximately half of children ate meals (53%) with their parent plus another adult, in comparison, 43% of snacks were consumed in the presence of more than one adult. Additionally, children were more likely to eat meals, as compared to snacks in the presence of other children (11% of meals were eaten with only the target child present, compared to 17% of snacks).

Differences also emerged in terms of location and atmosphere of eating occasions. For example, the bulk of meals (78%) were eaten around a table as compared to only 45% of snacks. Snacks were more frequently eaten on the couch (26%), scattered throughout the house (15%) or in a location outside of the house (13%; e.g. in the car, a restaurant). Also, screens were more prevalent at snacking occasions compared to mealtimes as well; parents reported the presence of screens at 35% (17% screentime only + 18% screentime plus conversation) of mealtimes, as compared to 44% of snacking occasions (24 % screentime only + 20 % screentime plus conversation).

Finally, differences emerged with regard to the types of foods eaten at different eating occasions. Grains, dairy, fruits and vegetables, and protein foods were served at a higher percentage of meals, as compared to snacking occasions. In contrast, foods containing added sugar (e.g. sugar-sweetened beverages, cake, candy) were consumed at 24% of snacks, as compared to only 13% of meals. Finally, a meaningful difference was observed with regard to how decisions were made about what to eat for meals versus snacks; for example, 21% of the time, snacks were chosen because the child asked for a specific food, as compared to only 10% of meals. Alternatively, foods served at meals were more likely planned ahead of time (16% as compared to only 4% of snacks).

Snacking Occasions: Contextual Differences by Child Sex and Weight Status

Snacking occasion context looked different in some cases for children of different genders and weight statuses (Tables 2-4); specifically, notable differences emerged with regard to food preparation, location and presence screens at snacks, and types of foods consumed at snacking occasions.

Boys:

Weight status was found to modify the consumption of homemade foods as snacks among boys (p=0.008); specifically, boys at a higher weight status consumed homemade snacks 32% of the time, whereas non-overweight boys consumed homemade snacks 42% of the time. Further, for overweight boys, child/family preference was found to guide the decision about what foods to serve at snacks 25% of the time, as compared to only 15% of the time for normal-weight boys (<0.001). Weight status was found to modify the presence of screentime at snacks among boys; overweight boys engaged in screentime during snacks 55% of the time, as compared to only 36% of the time among non-overweight boys.

Girls:

For girls at a higher weight status, snacks were more frequently chosen because they were healthy options (19% of the time) as compared to only 10% of the time among non-overweight girls; no such difference was seen among boys.

Both boys and girls:

Weight status was found to significantly modify the location of where snacks were consumed and the types of foods eaten at snacking occasions for both boys and girls (<0.001). For example, overweight girls reported consuming 27% of their snacks on the couch, as compared to 16% of the snacks consumed by non-overweight girls. Further, fruits and vegetables were eaten more frequently at snacks consumed by non-overweight boys (56%) and girls (61%), compared to overweight boys (42%) and girls (50%) (p<0.001).

Discussion

This study sought to better understand differences in the contextual factors of a meal- and snacking occasions within a racially/ethnically and socioeconomically diverse population of children and to clarify if there are differences in the approach to snacks by child sex or weight status. To the best of our knowledge, this study is the first to use EMA to directly compare quantitatively measured characteristics of snacking occasions to meals within the same population. Findings illuminated opportunities for potential nutrition interventions aimed at promoting healthful snacking patterns as a means of improving children’s overall dietary intake. Overall, meals and snacks looked quite different with regard to food preparation decisions, the location of where these eating occasions occurred, the presence of screens, and the types of foods served. Further, differences in snacking context were observed by child gender and weight status.

Differences between meals and snacks were observed with regard to food preparation; a greater percentage of meals were prepared by the parent, consisted of only homemade food, and were planned ahead of time, as opposed to being served in response to a specific request from the child. These findings confirm and expand upon the themes that emerged from prior qualitative work that explored snacking occasions among children.^13,15,16 The combined research effort of these separate studies highlight the idea that parents may be less invested in the types of foods served at snacking occasions. Although many parents take considerable care to plan, prepare and serve their children healthful meals in a routine and structured way, parents may serve snacks in reactive or responsive ways without special regard to the nutritional quality of the foods offered at these occasions. Indeed, Fisher et al concluded that parents of young children use snacks for a variety of non-nutritive reasons related to managing children’s behavior.¹⁵ Along these same lines, findings from the current study also revealed that children were much more likely to report consumption of foods containing added sugar (e.g. sugary drinks, cake, candy) and less likely to report consumption of grains, dairy, fruits and vegetables, and protein foods, during snacking occasions, as compared to mealtimes. These differences in intake between meals and snacks reflect what has been shown in the snacking literature to date.^2,7,12 Because children consume about a third of their total calories at snacking occasions, public health interventions seeking to improve the overall healthfulness of children’s dietary intake should encourage parents to see snacks as structured opportunities for nutrition and connecting with their children.

Less than half of snacks consumed by children were eaten at a table; instead, children ate snacks on the couch, scattered in different locations around the house, or someplace other than their house (e.g. the car, a restaurant, etc.) and nearly half of snacks were eaten while children were also engaged in screen time. The relaxed approach to the location of snacks might stem in part from the frequency of snacking occasions for children in this age range and the desire for parents to make it easier for them to provide their child with frequent snacks. These findings align with our previously conducted qualitative study, which found that parents more frequently allow their child to take on the responsibility of choosing where to eat their snacks and were flexible about the use of screens during snacking occasions and also extends the work of Davison et al,¹³ who found that parents had fewer rules for their children regarding snacking occasions, as compared to mealtimes. Overall, it seems that many parents provide less structure around what snacking occasions should look like for their children, resulting in more variability in location and use of screens.

Snacking occasions looked different for youth of different genders and weight statuses in several notable ways. For example, interesting differences in the consumption of certain foods eaten for snacks emerged between overweight and non-overweight youth. A greater percent of non-overweight youth (both boys and girls) reported the consumption of fruits and vegetables as snacks, as compared to overweight youth. These findings suggest that while an intervention targeting the promotion of healthful foods as snacks has the potential to improve the healthfulness of dietary intake for all youth, these messages might have particular relevance for youth at higher weight statuses. Further, notable contextual differences were observed; overall, boys and girls at higher weight status were more likely to eat snacks on couch. Further, among boys, those of higher weight status were less likely to eat foods prepared at home as snacks; the types of foods served as snacks were more frequently chosen because of child preference; snacks were more likely to be eaten in the presence of screens.

These differences in the structure of snacking occasions among children at higher weight statuses represent opportunities for clinicians and public health professionals who are seeking to improve the quality of foods served as snacks, as well as the context in which snacks are eaten to help to improve health outcomes among overweight youth. Primary care physicians are well aware of the rising rates of pediatric obesity in the United States, and healthy eating habits are a frequent topic of discussion during doctor’s visits. The current findings provide another way to discuss healthy eating strategies with families during well-child visits; physicians can highlight to parents the importance of paying attention not only to mealtime food choices but also to the foods served and eaten as snacks as a way to ensure healthy eating behavior. Families have diverse food and eating habits – having a variety of healthy eating suggestions and recommendations is key for providers who are advising them, particularly in the context of pediatric obesity.

Finally, in their 2017 systematic review on child snacking Blaine and colleagues called for the development of a detailed definition of snacking that addresses food type, context, and purpose, suggesting that such a definition is needed to advance this field of study. Indeed, a lack of clarity on how to best define snacking has made consistency in interpretation of findings across studies challenging and contributed, in part, to some of the disparate findings regarding the impact of snacking on dietary intake and weight status.^9,26 The EMA data from the current study suggest indicate that snacks, relative to meals, are more sugary and pre-prepared (food type), are more child-driven and less frequently pre-planned (purpose) and are less likely to be eaten at the table and more likely to be eaten outside of the home (context). These findings provide empirical articulation of the broader dimensions of snacking (food type, purpose, and context) outlined by Blaine and colleagues and contribute to the refinement of conceptual definitions of snacking for future research.

There are both strengths and limitations of this study. One strength is the use of EMA data, which allowed for the examination of several meal and snack characteristics across time and context. Also, the sample was highly diverse, including both racially/ethnically diverse participants and immigrants to the United States. There were also limitations to the study. One limitation is the overall small sample size (n=150), however using EMA data that was gathered multiple times per day over eight days allowed for examining more than 3500 meal and snack occasions, which improves our ability to capture both common and rare eating occasion features for analyses. Finally, EMA surveys were only completed by the primary caregiver. Including the secondary caregiver in future research may broaden study findings or reinforce findings already described for primary caregivers.

Conclusions

The current study explored differences in contextual characteristics of mealtimes and snacking occasions within a population of low-income, racially/ethnically diverse school-aged children, and there were differences in snacking occasions for children of different weight statuses. Overall, findings suggest that children are more likely to eat pre-prepared foods at snacks, as compared to other meals. Further, snacking occasions are less structured than meals; children are more likely to eat snacks in locations other than at a table, as compared to meals and a greater prevalence of snacks are eaten in the presence of screens. Given that children consume about a third of their total daily calories at snacking occasions, clinical and public health interventions should seek to encourage parents to provide their children with healthful foods within a structured environment when they are eating between meals. Pediatric healthcare providers should consider incorporating this information into their health visits with families to empower them to more critically pay attention to what is served and eaten during snacking occasions. This information highlights the potential pitfalls of snacking and can help families and youth to build healthier snacking behavior in both the short and long-term.

Table 3.

Snacking Meal Characteristic Differences Stratified by Child Sex (N=1,080 Snacking Occasions; N=71 Girls and N=79 Boys)

Food Preparation	Girls (N=71)			Boys (N=79)
	N Occasions (Overall %)a	N Participant (Between %)b	Average Participant Frequency (Within %)c	N Occasions (Overall %)a	N Participant (Between %)b	Average Participant Frequency (Within %)c	P Valued
Who prepared the meal?
Parent/partner	283 (56%)	60 (85%)	63%	300 (52%)	64 (81%)	63%	0.65
A cook outside home (e.g. fast food, restaurant, another adult)	137 (27%)	44 (62%)	34%	132 (23%)	50 (63%)	39%	0.55
Child participant + parent/partner	87 (17%)	37 (52%)	38%	141 (25%)	41 (52%)	42%	0.20
How was the meal prepared?
Only homemade	197 (39%)	56 (79%)	48%	214 (37%)	63 (80%)	46%	0.45
Only pre-prepared	235 (46%)	56 (79%)	54%	280 (49%)	62 (78%)	54%	0.58
Only fast-food	52 (10%)	26 (37%)	27%	42 (7%)	26 (33%)	28%	0.67
Combination of homemade, pre-pared, fast-food	23 (5%)	18 (25%)	18%	37 (6%)	24 (30%)	28%	0.19
What were the primary decisions about what was served?
Child / family likes	129 (25%)	45 (63%)	40%	116 (20%)	44 (56%)	36%	0.14
Child asked for a specific food or meal	112 (22%)	39 (55%)	39%	116 (20%)	41 (52%)	35%	0.50
Desire to avoid conflict with child or a family fight	16 (3%)	8 (11%)	23%	17 (3%)	7 (9%)	34%	0.41
Healthy option	70 (14%)	31 (44%)	27%	73 (13%)	36 (46%)	30%	0.54
Planned meal	20 (4%)	15 (21%)	18%	28 (5%)	20 (25%)	17%	0.91
Using food available in the home	71 (14%)	28 (39%)	33%	64 (11%)	35 (44%)	23%	0.65
Quick and easy to make	49 (10%)	29 (41%)	24%	85 (15%)	39 (49%)	29%	0.005
Other	40 (8%)	22 (31%)	22%	74 (13%)	36 (46%)	30%	0.10
Social Dimensions of the Eating Occasion	Girls (N=71)			Boys (N=79)
	N Occasions (Overall %)a	N Participant (Between %)b	Average Participant Frequency (Within %)c	N Occasions (Overall %)a	N Participant (Between %)b	Average Participant Frequency (Within %)c	P Valued
How many children were present?
Child participant	93 (18%)	35 (49%)	37%	88 (15%)	36 (46%)	37%	0.68
Two children	174 (34%)	49 (69%)	50%	219 (38%)	53 (67%)	51%	0.85
Three children or more	240 (47%)	50 (70%)	59%	266 (46%)	55 (70%)	67%	0.99
Who were these children?
Child participant	93 (18%)	35 (49%)	37%	88 (15%)	36 (46%)	37%	0.68
Child participant + siblings(s)	345 (68%)	60 (85%)	75%	391 (68%)	67 (85%)	76%	0.60
Child participant + siblings(s)+others	69 (14%)	28 (39%)	33%	94 (16%)	39 (49%)	34%	0.24
Note for others: extended family members or non-family members
How many adults were present?
Parent participant	260 (51%)	59 (83%)	61%	357 (62%)	71 (90%)	66%	0.04
Parent participant + another adult(s)	247 (49%)	54 (76%)	58%	216 (38%)	57 (72%)	53%	0.04
Who were these adults?
Parent/primary caregiver	259 (51%)	59 (83%)	61%	357 (62%)	71 (90%)	66%	0.04
Parent + other family member(s)	229 (45%)	51 (72%)	56%	192 (34%)	52 (66%)	51%	0.02
Parent + other family member + non-family member	18 (4%)	12 (17%)	23%	24 (4%)	17 (22%)	21%	0.24
Note for other family member: second caregiver and grandparent
Features of the Eating Occasion	Girls (N=71)			Boys (N=79)
	N Occasions (Overall %)a	N Participant (Between %)b	Average Participant Frequency (Within %)c	N Occasions (Overall %)a	N Participant (Between %)b	Average Participant Frequency (Within %)c	P Valued
Where did the meal take place?
Around the table	256 (50%)	63 (89%)	58%	232 (40%)	64 (81%)	51%	0.02
On couch/chair in living area	105 (21%)	37 (52%)	38%	179 (31%)	54 (68%)	44%	0.01
Scattered throughout house or standing-up	77 (15%)	26 (37%)	32%	88 (15%)	34 (43%)	32%	0.61
Outside house (e.g. in the car, at a restaurant, other)	69 (14%)	29 (41%)	27%	74 (13%)	30 (38%)	32%	0.71
Note for other: at a park, school's picnic table/event, neighbor's home, relative's home, or friend's home.
What kinds of foods were eaten?
All Grains	158 (31%)	50 (70%)	45%	185 (32%)	57 (72%)	38%	0.25
Dairy	156 (31%)	50 (70%)	46%	175 (31%)	58 (73%)	41%	0.61
Fruit and Vegetables	264 (52%)	57 (80%)	54%	264 (46%)	63 (80%)	55%	0.12
Meat, Beans, and Nuts	87 (17%)	37 (52%)	31%	113 (20%)	44 (56%)	36%	0.27
Added Sugar (e.g., sugary drinks, cake, candy)	113 (22%)	45 (63%)	31%	146 (25%)	51 (65%)	37%	0.30
What was happening during the meal or snack?
Conversation only	224 (44%)	55 (77%)	54%	213 (37%)	61 (77%)	51%	0.19
Screentime activities only	105 (21%)	43 (61%)	38%	150 (26%)	49 (62%)	39%	0.10
Screentime + conversation	104 (21%)	32 (45%)	39%	108 (19%)	41 (52%)	35%	0.76
Conversation + other	14 (3%)	6 (8%)	20%	11 (2%)	6 (8%)	25%	0.07
No conversation + other	60 (12%)	25 (35%)	30%	91 (16%)	34 (43%)	32%	0.05
Note for screentime activities: Listening to headphones; playing videogames; watching TV and/or TV in background; using cell phone, computer, or tablet
Note for other: Reading/looking at a book or none of the above
What was the atmosphere like?
Relaxed	142 (28%)	49 (69%)	42%	200 (35%)	58 (73%)	45%	0.14
Enjoyable	242 (48%)	52 (73%)	58%	237 (41%)	63 (80%)	50%	0.17
Neutral	88 (17%)	29 (41%)	41%	92 (16%)	36 (46%)	36%	0.86
Rushed	20 (4%)	13 (18%)	17%	23 (4%)	13 (16%)	32%	0.87
Chaotic	8 (2%)	6 (8%)	17%	13 (2%)	5 (6%)	33%	0.34
Tense	7 (1%)	6 (8%)	19%	8 (1%)	7 (9%)	13%	0.73

^a.Overall tabulations count the frequency of response at the sample population level. For example, of the girls' 507 snacking occasions, 56% (N=283) were prepared by the parent or partner.

^b.Between tabulations compute the fraction of the participants in the sex stratified sample (N=71 girls and N=79 boys) who reported a response at least once for the respective snacking occasion. For example, 60 participants with girls (80%) had at least one snacking occasion in which the parent or partner prepared the snack, which means that 11 participants with girls never had a snack prepared by a parent/partner over the course of the study. Only 6 participants with girls had at least one snack that they considered to be a tense atmosphere.

^c.Within-participant statistics report the fraction of the time that participants reported the same snacking characteristic over multiple occasions. For example, the most frequent response about the snacking atmosphere among participants with girls was "enjoyable" (48%) and 73% of the stratified sample of girls (N=52) reported the snacking characteristic at least once. Of the 52 participant girls who had an “enjoyable” snacking occasion, they reported the same response about other snacking occasions 58% of the time -- almost 3 in every 5 snacks.

^d.Response values were dummy coded to test if the probability of the snacking response differed between boys and girls. Meal occasions were restricted to snacks and generalized estimating equation models were fitted to account for repeated observations and adjusted for parent sex, child age and race/ethnicity, household structure (e.g., single parent with/without other adults and multiple parents with/without other adults) and reported annual household income.