Marketing of unhealthy foods and beverages (‘junk food’) to children, which accounts for ∼90 % of child-targeted food advertising(1), influences dietary intake and preferences(Reference Sadeghirad, Duhaney and Motaghipisheh2) and remains an important public health challenge(3,Reference Radesky, Chassiakos and Ameenuddin4) . The disproportionate spending of junk food marketing dollars targeting Black and Hispanic youth further exacerbates health inequities in communities of colour(Reference Backholer, Gupta and Zorbas5,6) . Television is the primary outlet for food marketing; however, social and digital media platforms (e.g. Facebook, YouTube) are increasingly used to reach young consumers(Reference Alruwaily, Mangold and Greene7,Reference Fleming-Milici and Harris8) given their broad adoption and reach(9). Despite expert recommendations, current US food and beverage industry self-regulatory policy does not explicitly include social media when defining child-directed food marketing(10). Greater understanding of public opinion towards policies to restrict junk food advertising to children on social media is an important step towards understanding the message environment to inform potential interventions(Reference Jeffery, Forster and Schmid11–Reference Kumanyika14).
The current study describes the extent to which US adults are supportive, neutral or opposed to restricting junk food advertising to children on social media and explores associations of neutrality and opposition with socio-demographic and health-related characteristics.
Methods
In 2020–2021, we analysed data from the National Cancer Institute’s Health Information National Trends Survey (HINTS). HINTS is a cross-sectional, nationally representative survey most recently administered by mail between February and June 2020 among civilian, non-institutionalised US adults aged 18+ years (HINTS 5 Cycle 4; n 3865). Further methodology details are available elsewhere(15).
Measures
Social media policy
Respondents rated the extent they would support or oppose the following: ‘Junk food products, including candy, chips, soda, and flavored sports drinks, should not be advertised to children on social media’(16). Responses were dichotomised for logistic regression: ‘neutral or opposed’ (neither support nor oppose; oppose and strongly oppose) v. ‘supportive’ (support; strongly support).
Self-reported characteristics included(16):
Socio-demographic characteristics
Age; sex at birth; race/ethnicity (non-Hispanic White; non-Hispanic Black; Hispanic; non-Hispanic Other); education level (high school or less; technical, vocational or some college; college or more); children in the household (yes/no); visited a social networking site (e.g. Facebook, LinkedIn) in the prior 12 months (yes/no; labelled ‘social media use’).
Health-related characteristics
Perceived health status (poor/fair; good and very good/excellent); BMI weight status category calculated from self-reported height and weight (normal weight: 18·5–24·9 kg/m2; overweight: 25·0–29·9 kg/m2 and obese: ≥ 30·0 kg/m2); having one or more chronic conditions (yes/no; including diabetes or high blood sugar; high blood pressure or hypertension; heart condition; chronic lung disease, asthma, emphysema or chronic bronchitis; depression or anxiety disorder); believing at least one of the following statements has ‘a lot’ of influence on whether a person will develop cancer: being overweight or obese; gaining weight in adult life; eating too much red meat (yes/no; labelled ‘strong weight/diet-related cancer beliefs’).
Political viewpoint
Reported from a seven-point scale from very liberal to very conservative (categorised as liberal, moderate and conservative).
Statistical analysis
Analyses were conducted in RStudio (v1.3.1056) and applied sample jackknife replicate weights to provide estimates representative of the USA population. We used multivariable logistic regression with listwise deletion to examine the odds of being neutral or opposed to the social media advertising restriction – v. supportive – by socio-demographic and health-related characteristics, while controlling for political viewpoint, to identify targeted communication approaches for adults who are opposed or neutral and may not yet see the value in such a policy measure. We conducted a sensitivity analysis utilising multinomial logistic regression to examine relative risks of being ‘opposed v. supportive’ and ‘neutral v. supportive’ by the same characteristics.
Results
The analytic sample included n 2852 respondents (mean age 46·9 (se 0·4) years; 50·0 % male; 65·8 % non-Hispanic White) with complete information on social media policy opinion, socio-demographic characteristics, health-related characteristics and political viewpoint. The largest proportion of adults were neutral about restricting junk food advertising to children on social media (40·6 %; Table 1). Support (46·1 %) was more common than opposition (13·4 %). When dichotomised, a slight majority of adults were either neutral or opposed to the social media policy (53·9 %).
Prevalence of adults’ opinion on restricting junk food advertising to children on social media, Health Information National Trends Survey (HINTS), United States, 2020
Of the n 3865 total HINTS 5 Cycle 4 respondents, n 3769 (97·5 %) reported opinion on restricting junk food advertising to children on social media (weighted percentages: 6·5 % strongly oppose; 7·7 % oppose; 40·6 % neither support nor oppose; 22·3 % support; 22·9 % strongly support); and n 2852 (73·8 %) had complete information on characteristics reported in Table 2.
Table 2 includes estimates from the multivariable logistic regression.
Correlates of adults’ opinion on restricting junk food advertising to children on social media, Health Information National Trends Survey (HINTS), United States, 2020
* Mean (se) and percentages are weighted to reflect US population estimates.
† The Non-Hispanic Other category includes the following self-reported races: American Indian or Alaska Native, Asian Indian, Chinese, Filipino, other Asian, other Pacific Islander and multiple races selected.
‡ BMI calculated from self-reported height and weight; n 61 respondents with BMI < 18·5 kg/m2 (underweight) were excluded.
§ Chronic conditions included: diabetes or high blood sugar (17·5 %); high blood pressure or hypertension (34·6 %); a heart condition such as heart attack, angina or congestive heart failure (7·6 %); chronic lung disease, asthma, emphysema or chronic bronchitis (12·0 %) and depression or anxiety disorder (23·6 %).
‖ Adults with strong weight/diet-related cancer beliefs indicated that at least one of the following statements has ‘a lot’ of influence on whether or not a person will develop cancer: being overweight or obese (34·0 % ‘a lot’), gaining weight in adult life (22·7 % ‘a lot’), and/or eating too much red meat (22·9 % ‘a lot’).
Unweighted analytic sample size: n 2852 of 3865 total HINTS 5 Cycle 4 respondents (73·8 %). OR and 95 % CI are reported from multivariable logistic regression adjusted for characteristics included in the table.
Socio-demographic characteristics
Non-Hispanic Black adults and adults in the non-Hispanic other category had 2 and 1·7 times the odds, respectively, of being neutral or opposed to the social media policy than non-Hispanic White adults. Compared with adults with a college degree or higher, adults with lower education levels had 1·7 to 2·6 times the odds of being neutral or opposed to the policy. Sex at birth, living with children in the household and social media use were not statistically significantly.
Health-related characteristics
Compared with adults in the normal weight status category, adults categorised as overweight had 1·4 times the odds of being neutral or opposed to the social media policy. Having strong (v. weaker) weight/diet-related cancer beliefs was associated with 53 % lower odds of being neutral or opposed to the policy. Perceived health status and having chronic condition(s) were NS.
Political viewpoint
Adults with a moderate or conservative (v. liberal) political viewpoint had 1·5 and 1·7 times the odds, respectively, of being neutral or opposed to the policy.
Multinomial logistic regression also yielded statistically significant estimates for race/ethnicity, education, BMI category, weight/diet-related cancer beliefs and political viewpoint (see online supplemental Table S1), although some associations were significant in one outcome comparison (opposed v. supportive or neutral v. supportive). For example, given other variables held constant, adults categorised as obese were 1·8 times more likely than adults in the normal weight status category to oppose the social media policy than support it (the neutral v. supportive association was not statistically significant). Estimates for overweight v. normal weight were not statistically significant but similar in magnitude to that of the logistic regression.
Discussion
Our analysis of 2020 HINTS data suggests that slightly more than half of US adults (53·9 %) were either neutral towards or opposed to restricting junk food advertising to children on social media. This may indicate a lack of awareness of the issue or its importance for child health. We observed significant differences in opinion by race/ethnicity, education, BMI category, weight/diet-related cancer beliefs and political viewpoint. Non-Hispanic Black adults, and those without a college degree, with overweight, and reporting a moderate or conservative political viewpoint had greater odds of being neutral or opposed to the social media policy. As might be expected, adults who believed weight status and diet are strongly associated with cancer had greater odds of policy support.
Although not specific to social media, Fleming-Milici and colleagues examined US parents’ support for reducing marketing of unhealthy foods to children in the broader food environment. In contrast to our findings, results from the 2009 to 2012 study suggested greatest policy support among women and Hispanic and Black parents(Reference Fleming-Milici, Harris and Liu12). Variation in study population, time, messaging environment focus and policy support measurement, challenge direct comparisons to the current analysis among all adults. However, both studies highlight the role that political viewpoint and differences by race/ethnicity may play in supporting policies to restrict marketing of unhealthy foods to children.
Our results suggest that beliefs and understanding of the science related to weight, diet and cancer risk are potentially modifiable targets for communication strategies to increase policy support and political will(Reference Richmond, Kotelchuck, Holland, Detels and Knox17). Targeting communication efforts to particular contexts will be important, especially given observed differences in opinion by BMI category, race/ethnicity and education.
Prior research suggests that support for public health policies is strongest for those aiming to protect children(Reference Jeffery, Forster and Schmid11). However, when controlling for other factors, living with child(ren) was not significantly associated with policy opinion in the current study. This may be due to limitations in the questionnaire, which did not include items on whether the respondent was a parent(16). Further, age of child(ren) in the household was not assessed, nor was child screen time, household social media behaviours and household eating behaviours (e.g. diet quality, food decision making) – potentially important characteristics for future research that could be associated with policy opinion(Reference Fleming-Milici and Harris8,Reference Fleming-Milici, Harris and Liu12) . Future exploratory research may also seek to understand adults’ rationale for their policy opinion.
Conclusions
Targeted communication interventions that increase public awareness of the links between weight, diet and cancer – particularly for adults with higher BMI, non-Hispanic Black adults and those with lower education – may increase support for restricting junk food advertising to children on social media, for which a high proportion of US adults have a neutral stance. Such restrictions could improve children’s food environments to prevent diet-related diseases.
Acknowledgements
Acknowledgements: None. Financial support: This research received no specific grant from any funding agency, commercial or not-for-profit sectors. The opinions expressed by the authors are their own and this material should not be interpreted as representing the official viewpoint of the US Department of Health and Human Services, the National Institutes of Health or the National Cancer Institute. Conflict of interest: There are no conflicts of interest. Authorship: All authors conceptualised the secondary analysis. A.R.K. conducted statistical analyses and wrote the first draft of the manuscript. All authors reviewed the manuscript and provided critically important intellectual content on subsequent drafts. Ethics of human subject participation: The current study used de-identified publicly available data. Institutional review board approval was not required.
Supplementary material
For supplementary material accompanying this paper visit https://doi.org/10.1017/S1368980021003359
