The Relationship Between Media Food Marketing Influence and Unhealthy Food Intake in Parent–Adolescent Dyads: An Actor–Partner Interdependence Model

Abstract

Media food marketing (MFM) may greatly influence adolescents’ and parents’ dietary behaviors through direct and mutual influences, yet the interplay of these dynamics is unexplored. This study investigated the impact of parents’ and adolescents’ trust in MFM on their own and each other’s unhealthy food consumption (actor and partner effects). Parent–adolescent dyadic data (n = 1656 dyads) collected from the 2014 Family Life, Activity, Sun, Health and Eating study were analyzed. Actor–Partner Interdependence Models were estimated to assess for actor and partner effects of MFM trust/influence on unhealthy food intake (i.e., detrimental food, junk food, sugar-sweetened beverages, fast and convenience foods) while accounting for dyadic interdependence. Covariates included parents’ and adolescents’ age and sex, and parents’ health literacy, food insecurity, and food label reading habits. Parents’ and adolescents’ higher MFM trust/influence were related to their higher unhealthy food intake (actor effects). Parents’ higher MFM trust/influence was also associated with unhealthy food intake in adolescents (partner effect). Partner effects from adolescents to parents were found for sugar-sweetened beverages only. Parents’ and adolescents’ MFM trust/influence was related to adolescents’ unhealthy food intake, highlighting the need to address it in both parents and adolescents to reduce adolescents’ unhealthy dietary intake.

1. Introduction

Media food marketing (MFM) is an effective strategy used by the food industry to influence food consumption behaviors in youth [1,2]. Most MFM is for unhealthy foods [3,4,5]. Children and adolescents are exposed to this marketing via advertisements and embedded content in movies, television shows, advergames, and videos streamed on popular platforms such as YouTube [5,6,7]. Exposure to this content affects children and adolescents’ food preferences, knowledge, and eating behaviors [1,3,7]. Adolescents are additionally exposed to MFM via advertisements and embedded content on social media, which likely shape their dietary perceptions and choices [8].

Most of the research on the influence of MFM on eating behaviors focuses on children and adults [9]. However, adolescents’ key developmental characteristics and increased spending power increase their vulnerability to MFM strategies [10]. In adolescence, rapid neurobiological changes increase adolescents’ vulnerability to external influence [11]. For example, adolescents’ reward systems and appetite cues are fully developed and hypersensitive while their inhibitory control is still underdeveloped, making them more prone to acting impulsively and responding to stimuli that promise reward even when they are aware of marketers’ attempts to persuade them [12]. Adolescence is also a period of increased autonomy and independent health decision-making [13], making them an easy target for marketers. Relatedly, in adolescence, the role of peer influence is more pronounced [14,15], with media being considered a “super peer” [16,17]. MFM strategies capitalize on these vulnerabilities by employing tactics to appeal to peer approval and status [11] and activate the reward system in the brain [10,12]. While there is an increase in autonomy and peer influence during adolescence, parents still influence adolescents’ eating behaviors via the home food environment and norms for and modeling of eating behaviors [18,19,20]. Parents also pass on brand loyalty and trust to their adolescents via their purchases and responses to MFM [21]. Thus, the complex interplay of MFM influences on parents’ and adolescents’ dietary intake warrants dyadic exploration.

An important goal of MFM to youth is to develop lifelong consumers by building trust and loyalty toward their brands [10,12]. McCullough and Gooddall [22] proposed that repeated exposure to advertising helps individuals ascribe more positive opinions about the foods advertised and reinforces positive experiences with the foods, which results in favorable intentions to purchase and consume the products and brand loyalty. Further, the more individuals are exposed to a brand, the higher the likelihood of brand recognition and familiarity, which builds trust [12,23]. While past studies have focused on the exposure of food advertisements [24,25,26], few studies have explored how trust in and perceptions of food advertising impact adolescents’ eating behaviors.

Current Study

Parents’ and adolescents’ eating behaviors are related [19] and influenced by the media [27,28]. Furthermore, adolescents’ media trust [29] and influence [7,30] are related to their food intake. The extent to which these relationships are interdependent has not yet been explored. Given that parents have primary responsibility for the family’s home food environment [18] and influence children’s brand recognition of food marketed in the media [21], the objective of this study is to identify patterns of interdependence between parents’ and adolescents’ MFM trust/influence and unhealthy food intake. Specifically, we explored actor and partner effects of MFM trust/influence on unhealthy food intake. Our primary hypotheses included: (1) adolescents’ and parents’ MFM trust/influence would be positively related to their own unhealthy food intake (actor effect), (2) adolescents’ MFM trust/influence would be positively related to their parents’ unhealthy food intake (partner effect), and (3) parents’ MFM trust/influence would be positively related to their adolescents’ unhealthy food intake (partner effect).

Given that age and gender are associated with eating behaviors in adolescents [31,32] and adults [33], these variables were included as covariates. Several critical factors affecting food availability were also considered as covariates. Specifically, reading food labels reduces the amount of unhealthy foods people purchase [34] and likely affects how much one is influenced by MFM [35]. However, health literacy, one’s ability to access, comprehend, assess, and apply health information in everyday health decision-making [36], is required to read food labels accurately. Lastly, knowing what to eat and translating that knowledge into action relies on having the resources to do so, specifically food security [37]. Therefore, for our secondary hypotheses, we hypothesized that higher health literacy, reading food labels, and food security among parents would be related to lower unhealthy food intake for parents and adolescents.

2. Methods

2.1. Study Design and Data Collection Procedures

Data for this study were sourced from the 2014 Family Life, Activity, Sun, Health and Eating (FLASHE) study, which is a project of the National Cancer Institute (NCI) within the National Institute of Health [38]. FLASHE, designed as a cross-sectional national study of a non-probability US sample, aimed to investigate the interplay between psychosocial, generational, and environmental factors concerning cancer prevention behaviors. Dyads were asked to complete two web-based surveys, one primarily focused on diet and another on other health behaviors including physical activity, sun health, sleep, and tobacco use. FLASHE enrolled 1945 parent–adolescent dyads, and at least one member of 1711 dyads completed at least one survey. The study protocol was approved by the Westat, Inc. and NCI Special Studies Institutional Review Boards.

Data collection occurred between April and October 2014, and recruitment of parent–adolescent dyads was facilitated through the Ipsos Consumer Opinion Panel. Eligibility criteria required participating parents to be over 18 years old, cohabit with at least one child aged 12–17 years for at least half of the time, and express willingness to be contacted for study participation. One adolescent from each household was randomly selected to participate in the study. To ensure representation, balanced sampling techniques were employed to align the sample with key demographic parameters such as gender, household income, and race/ethnicity, effectively mirroring the diversity of the United States population. Four diet and physical activity surveys were completed by the parent–adolescent dyads, and upon their completion, the dyads received monetary compensation (USD 5 to USD 10 per survey). In the current study, the analyses specifically focus on data derived from the surveys related to the dietary habits of parents and adolescents. Dyads were excluded if one (n = 100) or both (n = 134) dyad members did not complete the diet survey. The total analytic sample included 1656 dyads. For additional details of the study’s methodology, please refer to the detailed report of FLASHE methods [39].

2.2. Measures

2.2.1. Independent Variables

MFM trust and influence. Parents’ and adolescents’ trust in food media and MFM’s influence on their behaviors were assessed with three items developed and cognitively tested for the FLASHE survey. Parent and adolescent respondents were prompted to think about messages they saw or heard about foods and drinks on television, magazines, radio, internet, or billboards. They then indicated their level of agreement with the following statements: when I see advertisements for foods or drinks, (1) I want to try the advertised foods or drinks, (2) I think the advertised foods or drinks will taste good, and (3) I trust the messages advertised. Responses were on a 5-point Likert scale ranging from strongly disagree to strongly agree. The three items had good internal consistency (Cronbach’s alpha = 0.83 and 0.84 for parents and adolescents, respectively), thus, a mean score was computed and used for respondents’ “MFM trust/influence.” Higher scores indicated a higher trust in and influence of MFM messages.

2.2.2. Dependent Variables

Intake of unhealthy foods was assessed using a 27-item dietary screener adapted from the Dietary Screener Questionnaire [40] and the dietary questionnaire used in the National Youth Physical Activity and Nutrition Study [41]. Participants responded to items on a 6-point ordinal scale, and these responses were recoded into daily intake frequency as follows: I did not eat or drink the item = 0/day, 1–3 times/week = 0.286/day, 4–6 times/week = 0.714/day, 1 time/day = 1/day, 2 times/day = 2/day, and 3 or more times/day= 3/day [42]. The FLASHE study developed several food group categories by grouping items and summing their recoded daily intake frequencies. For this paper, we used four categories: junk food, sugar-sweetened beverages (SSB), fast and convenient foods, and detrimental foods. For junk food, candy/chocolate, cookies/cake, potato chips, fried potatoes, and frozen desserts were included. For SSB, soda, energy drinks, sweetened fruit drinks, and sports drinks were included. For fast/convenient foods, fried potatoes, fried chicken, pizza, tacos, burgers, and heat-and-serve were included. The detrimental food variable included all the aforementioned foods, as well as processed meat and sugary cereals. To account for potential overestimation and extreme outliers, daily intake frequencies for each food item were top-coded by FLASHE study investigators if they exceeded the threshold z-score of |≥3.29|. Values with z-scores exceeding this limit were top-coded by removing them and substituting them with the nearest value that did not have a z-score exceeding 3.29 [40].

2.2.3. Covariates

Within-dyad covariates. Parents and adolescents reported their age in years, and the FLASHE study’s public use dataset provided parents’ ages in four categories (18–34, 35–44, 45–59, 60+). Parents and adolescents self-identified as male or female.

Between-dyad covariates. Parental health literacy was assessed with the validated Single-item Literacy Screener [43]. Parents responded to “How often do you need to have someone help you read written material from your doctor or pharmacy?” The responses were dichotomized into “adequate health literacy” (responses: never, rarely) and “inadequate health literacy” (responses: sometimes, often, always). Parents also completed a single-item measure on reading food labels developed by the FLASHE study team. They were asked, “How often do you read the detailed food labels or nutrition facts?” Response options were on a 5-point Likert scale ranging from never to always. The responses were dichotomized into “often/always read food labels” and “never/rarely/sometimes read food labels.” Household food insecurity was assessed using a validated two-item food insecurity risk screener [44] in which parents were asked to indicate whether the following statements were never true, sometimes true, or often true in their household in the past 12 months: “we worried whether our food would run out before we got money to buy more” and “the food that we bought just didn’t last, and we didn’t have money to get more.” Respondents were categorized as food insecure (responded often true on one or both) or food secure (all other responses). This screener has a sensitivity of 97% and a specificity of 83% in identifying households with young children at risk of food insecurity [44].

2.3. Statistical Analyses

Descriptive statistics and Pearson correlations between parents’ and adolescents’ MFM trust/influence and unhealthy food intake were computed in SAS 9.4 TS1M8 [45]. Cohen’s d was computed to illustrate the size of the differences between dyad members on the unhealthy diet and MFM trust/influence variables. To test our hypotheses, we used parent–adolescent dyadic data and the Actor–Partner Interdependence Model [46], a statistical technique that accounts for dyadic interdependence. APIMs were computed using APIM_SEM, a free web-based application that estimates APIMs via structural equation modeling using maximum likelihood estimation [47]. The lavaan program in R [48] is embedded in APIM_SEM. Actor and partner effects of the parent–adolescent dyads’ MFM trust/influence on their unhealthy food intake were estimated within the models (see Figure 1). APIM_SEM estimates two types of standardized Beta. Beta(o) uses the overall standard errors across participants allowing for between-group comparisons (i.e., parents vs. adolescents). Beta(s) uses separate standard errors for parents and adolescents allowing for within-dyad comparisons (i.e., differences between parent–adolescent pairs) [47]. Only Beta(o) is reported in the results table as it allows for interpretations across dyad members. To control for collinearity between independent variables and between the shared unexplained variance of the outcome variables, covariances between the independent variables and between the residual errors of the outcome variables were included in model estimations. Regarding covariates, a between-dyad covariate has one measurement per dyad as it is a shared ‘trait’ and is regressed on both dyad member outcomes as it is expected to affect both dyads, while a within-dyad covariate has a separate measurement for each dyad member as it is unique to the member and is regressed on each dyad member’s respective outcomes. The APIM included three between-dyad covariates measured only in parents (health literacy, reading food labels, food insecurity) and two-within dyad covariates (age, sex). To determine the dyadic pattern of interdependence that is most likely to describe the parent–adolescent relationship regarding MFM trust/influence and unhealthy food intake, the k parameter, the ratio of the partner effect to actor effect, was computed [46]. Values of k can be infinite; however, values closer to −1, 1, and 0 have meaningful interpretations. Values closer to −1 are considered contrast patterns whereby the actor and partner have similar effect sizes but opposite signs. Values of 1 are considered couples-effect patterns whereby the actor and partner have similar effect sizes in similar directions. Values are considered actor-only patterns when only the actor causal variable and not the partner causal variable affect the actor outcome variable. These values are only interpreted if the absolute standardized actor effect is >0.1 and statistically significant [46]. To accurately interpret k, confidence intervals for k are needed, and these were calculated via Monte Carlo bootstrapping sampling [46,47]. Missing data were handled via full information maximum likelihood. Dyads with residuals of the fitted model greater than four standard deviations were removed, and models were re-estimated.

3. Results

3.1. Sample Characteristics

See

Table 1. Descriptive statistics for parent–adolescent dyads.

	Descriptives				Correlation
	Adolescent	Parent
Variables	Mean ± SD/N(%)	Mean ± SD/N(%)	t a	Cohen’s d (95% CI)	r(adolescent)(parent) b
Parent age in years
18–34		188 (11.4)
35–44		713 (43.1)
45–59		694 (41.9)
≥60		47 (2.8)
Adolescent age in years
12	219 (13.2)
13	326 (19.7)
14	276 (16.7)
15	288 (17.4)
16	326 (19.7)
17	202 (12.2)
Sex
Male	810 (48.9)	421 (25.4)
Female	823 (49.7)	1221 (74.4)
Parent health literacy
Inadequate literacy		38 (2.3)
Adequate literacy		1596 (96.4)
Parent food label reading frequency
Never/rarely/sometimes read food labels		815 (49.2)
Often/Always read food labels		829 (50.1)
Food insecurity
Never or sometimes food insecure		1467 (88.6)
Often food insecure		174 (10.5)
Detrimental food intake (score range 0–9) c	4.85 ± 2.91	3.54 ± 2.50	18.30 *	0.51 (0.45, 0.57)	0.56 *
Junk food intake (score range 0–3) d	1.85 ± 1.21	1.40 ± 0.99	15.31 *	0.40 (0.34, 0.45)	0.48 *
Fast and convenience food intake (score range 0–3) d	1.49 ± 1.02	1.09 ± 0.80	17.28 *	0.45 (0.39, 0.50)	0.52 *
Sugar-sweetened beverages intake (score range 0–3) d	1.27 ± 1.23	0.91 ± 1.15	10.39 *	0.27 (0.22, 0.32)	0.38 *
MFM trust/influence (score range 1–5) e	3.36 ± 0.90	3.07 ± 0.88	12.56 *	0.31 (0.26, 0.36)	0.44 *

MFM = media food marketing. ^a t statistic from paired sample t-test to compare parent and adolescent mean scores; ^b correlations between parents and adolescent scores; ^c Detrimental food intake scores ranged from 0 to 9 with higher scores indicating a higher intake. ^d Junk food, fast and convenience food, and sugar-sweetened beverages intake scores ranged from 0 to 3 with higher scores indicating a higher intake. ^e Media food marketing trust/influence scores ranged from 1 to 5 with higher scores indicating a higher trust in and influence of food advertising messages on participant. * p < 0.001.

for descriptive statistics. The majority of parents were ≥35 years (~88%) and female (74.4%). Adolescents’ ages ranged from 12 to 17 years (Mean = 14.5, SD = 1.6), and almost an equal proportion were male (48.9%) and female (49.7%). The parent (69.1%) and adolescent (63.7%) samples were majority non-Hispanic White. Most parents in our sample had adequate health literacy (96.4%) and were food-secure (88.6%). Approximately 51% of parents often or always read the detailed food labels or nutrition facts. Compared to parents, adolescents’ MFM trust/influence and detrimental food, junk food, fast/convenience foods, and SSB intake were significantly higher.

3.2. APIM Results

APIM results are presented in

Table 2. Results of the Actor–Partner Interdependence Models predicting parents’ and adolescents’ unhealthy food intake from their media food marketing trust/influence.

	Detrimental Foods Intake			Sugar-Sweetened Beverages Intake			Junk Food Intake			Fast and Convenience Foods Intake
	B (95% CI)	β(o)	r a	B (95% CI)	β(o)	r a	B (95% CI)	β(o)	r a	B (95% CI)	β(o)	r a
Parents
Intercept	3.46 *** (2.59, 4.33)			1.27 *** (0.82, 1.72)			1.22 *** (0.85, 1.59)			1.13 *** (0.88, 1.38)
Actor	0.55 *** (0.43, 0.67)	0.10	0.24	0.28 *** (0.21, 0.34)	0.18	0.23	0.16 *** (0.11, 0.22)	0.15	0.16	0.14 *** (0.10, 0.17)	0.29	0.21
Partner	0.09 (−0.02, 0.21)	0.04	0.05	0.07 * (0.01, 0.13)	0.05	0.06	0.01 (−0.04, 0.06)	0.01	0.02	0.03 (−0.00, 0.07)	0.05	0.05
k	0.17 (−0.06, 0.41)			0.25 (−0.01, 0.50)			0.08 (−0.24, 0.41)			0.24 (−0.03, 0.52)
Age	−0.27 *** (−0.39, −0.14)			−0.20 *** (−0.27, −0.13)			−0.01 (−0.07, 0.04)			−0.07 *** (−0.11, −0.04)
Female	−0.67 *** (−0.87, −0.47)			−0.37 *** (−0.48, −0.26)			−0.06 (−0.15, 0.03)			−0.18 *** (−0.23, −0.12)
Adolescents
Intercept	3.61 *** (2.58, 4.64)			0.60 * (0.10, 1.10)			1.51 *** (1.06, 1.96)			1.20 *** (0.88, 1.52)
Actor	0.40 *** (0.26, 0.54)	0.17	0.15	0.18 *** (0.11, 0.25)	0.14	0.13	0.10 ** (0.03, 0.16)	0.09	0.08	0.13 *** (0.08, 0.17)	0.17	0.15
Partner	0.27 *** (0.12, 0.41)	0.11	0.11	0.16 *** (0.09, 0.24)	0.13	0.12	0.09 ** (0.02, 0.15)	0.09	0.07	0.06 ** (0.02, 0.11)	0.09	0.08
k	0.67 (0.15, 1.19)			0.91 (0.25, 1.56)			0.91 (−0.16, 1.98)			0.49 (0.04, 0.95)
Age	−0.02 (−0.09, 0.05)			0.03 (−0.01, 0.06)			−0.05 ** (−0.08, −0.02)			0.01 (−0.01, 0.03)
Female	−0.46 *** (−0.69, −0.24)			−0.25 *** (−0.36, −0.14)			−0.03 (−0.12, 0.06)			−0.12 *** (−0.18, −0.05)
Covariate b
Parent HL→PUFI	−0.68 (−1.35, −0.01)			−0.60 *** (−0.93, −0.26)			−0.28 (−0.01, 0.56)			−0.28 * (−0.48, −0.09)
Parent HL→ AUFI	−0.74 (−1.57, 0.09)			−0.38 (−0.76, 0.01)			−0.11 (−0.48, 0.25)			−0.41 ** (−0.67, −0.15)
Food insecurity → PUFI	0.27 (−0.04, 0.58)			0.31 *** (0.15, 0.48)			−0.05 (−0.18, 0.09)			0.08 (−0.01, 0.17)
Food insecurity → AUFI	0.03 (−0.34, 0.40)			0.08 (−0.11, 0.26)			−0.03 (−0.19, 0.13)			0.07 (−0.05, 0.18)
Parent food label reading → PUFI	−0.73 *** (−0.92, −0.54)			−0.31 *** (−0.41, −0.21)			−0.19 ** (−0.27, −0.11)			−0.15 *** (−0.21, −0.10)
Parent food label reading → AUFI	−0.37 ** (−0.60, −0.14)			−0.16 * (−0.27, −0.04)			−0.06 (−0.16, 0.04)			−0.07 (−0.14, 0.00)
Covariance
Adolescent MFM↔ Parent MFM	0.34 *** (0.30, 0.38)			0.35 *** (0.31, 0.40)			0.34 *** (0.30, 0.38)			0.34 *** (0.30, 0.38)
Residual error AUFI↔PUFI	1.57 *** (1.34, 1.79)			0.38 *** (0.32, 0.44)			0.33 *** (0.28, 0.37)			0.13 *** (0.11, 0.15)

AUFI = adolescent unhealthy food intake; HL = health literacy; MFM = media food marketing trust/influence; PUFI = parent unhealthy food intake. ^a pairwise partial correlation (correlation between specific media food marketing trust/influence and unhealthy food intake after controlling for other predictors in the model); ^b between-dyad covariate. * p < 0.05, ** p < 0.01, *** p < 0.001.

.

3.2.1. Detrimental Food Intake (DFI)

The analytic dataset included 1609 dyads (47 outliers removed). The model results supported the actor effects hypotheses, partially supported the partner effect hypotheses, and supported the interdependence of the dyads’ MFM trust/influence and adolescents’ DFI. The model explained ~7% and ~16% of the variance in adolescents’ and parents’ DFI, respectively. After controlling for all other predictor variables in the model, adolescents’ and parents’ DFI were positively related (partial-intraclass correlation [partial-ICC] = 0.40, p < 0.001). The actor effects for adolescents and parents were significant, suggesting that, after controlling for partner effects and covariates in the model, adolescents’ and parents’ MFM trust/influence were related to higher DFI for adolescents and parents, respectively. The partner effect from adolescents to parents was not significant. However, the partner effect from parents to adolescents was significant and positive, suggesting that, after controlling for actor effects and covariates in the model, parents’ MFM trust/influence was positively related to their adolescents’ DFI. Interdependence patterns were consistent with the actor-only model as the confidence interval for k included 0 (k = 0.17; 95% CI [−0.06, 0.40]). Specifically, parents’ DFI was better explained by their MFM trust/influence. Interdependence patterns were consistent with the couple model for adolescents as the confidence interval for k included 1 (k = 0.67; 95% CI [0.15, 1.19]). Specifically, adolescents’ DFI was better explained by their and their parents’ MFM trust/influence.

Adequate health literacy in parents was related to lower DFI for parents. Parents often/always reading food labels was related to lower DFI for both dyad members. Older parents and parents and adolescents who were female had significantly lower DFI.

3.2.2. SSB Intake

The analytic dataset included 1642 dyads (14 outliers removed). The model results supported the actor and partner effects hypotheses and the interdependence of the dyads’ MFM trust/influence and adolescents’ SSB intake. The model explained ~7% and ~16% of the variance in adolescents’ and parents’ SSB intake, respectively. After controlling for all other predictor variables in the model, adolescents’ and parents’ SSB intake were positively related (partial-ICC = 0.33, p < 0.001). The actor effect for adolescents and parents were positive and significant, suggesting that, after controlling for partner effects and covariates in the model, higher MFM trust/influence in parents and adolescents were associated with greater SSB intake in parents and adolescents, respectively. The partner effects from adolescents to parents and from parents to adolescents were positive and significant. This suggests that, after controlling for actor effects and covariates in the model, parents’ and adolescents’ MFM trust/influence were positively related to their adolescents’ and parents’ SSB intake, respectively. Interdependence patterns were consistent with the actor-only model for parents (k = 0.25; 95% CI [−0.01, 0.50]), suggesting that parents’ MFM trust/influence better explained their SSB intake. For adolescents, the couple model (k = 0.91, 95% CI [0.25, 1.56]) was plausible, suggesting that adolescents’ SSB intake was better explained by their and their parents’ MFM trust/influence.

Adequate health literacy in parents was associated with their lower SSB intake, and parents often/always reading food labels was associated with lower SSB intake in both parents and adolescents. Higher food insecurity was related to higher parental SSB intake. Older parents and parents and adolescents who were female had lower SSB intake.

3.2.3. Junk Food Intake

The analytic dataset included 1602 dyads (54 outliers removed). The model results supported the actor effects hypotheses and partially supported the partner effect hypotheses while the interdependence of the relationships was less clear. The model explained ~3% and ~5% of the variance in adolescents’ and parents’ junk food intake, respectively. After controlling for all other predictor variables in the model, adolescents’ and parents’ junk food intake were positively related (partial-ICC = 0.41, p < 0.001). The actor effect for adolescents and parents were positive and significant, suggesting that, after controlling for partner effects and covariates in the model, higher adolescents’ and parents’ MFM trust/influence were, respectively, related to their higher junk food intake. The partner effect from parents to adolescents was significant and positive, suggesting that, after controlling for actor effects and covariates in the model, higher parents’ MFM trust/influence was related to higher junk food intake by their adolescents. Interdependence patterns were consistent with the actor-only model (k = 0.08; 95% CI [−0.24, 0.41]), suggesting that parents’ junk food intake was better explained by their MFM trust/influence. For adolescents, interdependent patterns were not fully identifiable as the confidence interval contained both 0 and 1, suggesting that both the actor-only and couple models could be plausible.

Parents who often/always read food labels had lower junk food intake. Older adolescents had lower junk food intake than younger adolescents.

3.2.4. Fast/Convenience Food Intake (FFI)

The analytic dataset included 1592 dyads (64 outliers removed). The model results supported the actor effects hypotheses and partially supported the partner effect hypotheses, while the interdependence of the relationships was less clear. The model explained ~6% and ~12% of the variance in adolescents’ and parents’ FFI, respectively. After controlling for all other predictor variables in the model, adolescents’ and parents’ FFI were positively related (partial-ICC = 0.36, p < 0.001). The actor effects for adolescents and parents were positive and significant, suggesting that after controlling for partner effects and covariates in the model, higher adolescents’ and parents’ MFM trust/influence were, respectively, related to their higher FFI. The partner effect from parents to adolescents was significant and positive, suggesting that, after controlling for actor effects and covariates in the model, higher parents’ MFM trust/influence was related to FFI for their adolescents. The interdependence pattern for parents was the actor-only model (k = 0.24; 95% CI [−0.03, 0.52]), suggesting that parents’ FFI was better explained by their MFM trust/influence. The interdependent patterns for adolescents were not plausible as the confidence interval included both 0 and 1.

Adequate health literacy in parents was related to lower FFI for both dyad members. Parents often/always reading food labels was related to their lower FFI. Older parents and parents and adolescents who were female had lower FFI.

4. Discussion

This study examined the interdependence of parents’ and adolescents’ MFM trust/influence and their unhealthy food intake. Using a national sample of parent–adolescent dyads for the FLASHE study [39], this study found that parents’ and adolescents’ unhealthy food intake were positively correlated, suggesting congruence in their dietary intake. As hypothesized, higher MFM trust/influence among parents and adolescents was related to their higher unhealthy food intake (actor effect). Higher parental MFM trust/influence was related to higher unhealthy food intake in adolescents (partner effect). Furthermore, a consistent pattern emerged across food groups: only parents’ MFM trust/influence was associated with their own unhealthy food intake. For adolescents, MFM trust/influence of both parents and adolescents more effectively explained their DFI and SSB intake. However, the results were not definitive about whether adolescents’ junk food and FFI were related solely to their MFM trust/influence or both their and their parents’ MFM trust/influence.

MFM aims to develop lifelong customers by building trust and brand loyalty [10,12]. Most MFM to children and adolescents focuses on unhealthy foods [3,4,5]. Therefore, it was expected that parents and adolescents with higher MFM trust/influence in this study would have higher intake of unhealthy foods. Notably, the correlation between parents’ and adolescents’ MFM trust/influence and the correlation between parents’ and adolescents’ unhealthy food intake are likely driven by social determinants of health factors implicated in both variables, such as media exposure, cultural influences, and neighborhood characteristics. Policies addressing MFM should consider the intersection of MFM and social determinants of health (e.g., MFM for unhealthy foods in neighborhoods that are food deserts).

The cross-sectional nature of the study precludes us from determining whether the relationship between MFM trust/influence and unhealthy food intake in adolescents is a developing phenomenon in adolescence or a spillover effect from childhood. However, an unarguable distinction between MFM to children vs. adolescents is the new ways in which food companies reach adolescents and expand their influence. The data analyzed in our study were collected in 2014, and since then, food companies’ direct access to adolescents has intensified due to adolescents’ increased access to the internet and smart devices [49], social media platforms, and exposure to influencers. Food companies use more subtle tactics to influence adolescents, such as social media ads and influencers, product placement in video games, and ads on video streaming apps [1,50]. These tactics prey on vulnerabilities typical in adolescent development, including their search for identity [51], wanting to feel a sense of belonging and relevance [52], and peer approval [11]. Despite the ubiquity of exposure to MFM in the lives of adolescents, there is little attention on regulating how food products are marketed to adolescents via minimally regulated platforms. Our findings suggest that targeting MFM directed at adolescents is necessary and we propose that doing so through policies regulating marketing to adolescents may reduce unhealthy food intake in this population.

Adolescents’ DFI and SSB intake were better explained by both parents’ and adolescents’ MFM trust/influence. These findings suggest that in addition to directly contributing to adolescents’ unhealthy food intake, MFM may also continue to indirectly influence adolescents’ unhealthy behavior via its effect on parents. There are several possible hypotheses that may explain parents’ high MFM trust/influence being associated with adolescents’ unhealthy food intake. First, parents’ MFM trust/influence may be associated with home food availability [18,20], such that parents’ responsiveness to MFM may mean more unhealthy food options available in the home. Second, parents with higher MFM trust/influence may be less inclined to regulate their adolescents’ diet, critically engage in conversations about unhealthy foods and MFM, or model healthy food intake. Third, parents with higher MFM trust/influence may have held these attitudes since their adolescents were children. Thus, their adolescents may have experienced a home food environment, parent modeling of unhealthy food intake, and brand loyalty and trust for unhealthy foods prior to adolescence. Relatedly, adolescents with higher MFM trust/influence may have developed these attitudes due to exposure to unhealthy foods in the home (i.e., reinforcement of MFM messages), exposure to unhealthy foods in their neighborhoods, and witnessing or learning from parents’ MFM trust/influence, in addition to succumbing to MFM tactics targeting them directly.

Both parent’s health literacy and food label reading were negatively related to unhealthy food intake for either parents only or both dyad members. Parents’ health literacy might affect the type of health information they seek or understand and may affect their ability to critically analyze media messages, thus affecting their decision-making regarding food intake and food availability in the home, both of which likely affect adolescents’ unhealthy food intake. Health literacy skills predict the ability to accurately read food labels or seek out information on food labels [53]. Furthermore, individuals who read food labels may do so because they may be interested in making more informed decisions about the foods they eat and may not be as trusting of or easily influenced by MFM. Parents who read food labels may be more discerning in the foods they consume, food parenting practices for their adolescents, and food purchasing behaviors [54]. Regarding food insecurity, some studies show no relationship between household food insecurity and adolescents’ sugary food intake [55], while other studies show that food-insecure children have higher intake of soda [56] and lower intake of fruits and vegetables [57]. Nonetheless, food insecurity is a factor in parents’ unhealthy food intake and may indirectly influence adolescents’ diet through parental pathways (e.g., modeling, home food availability).

5. Strengths and Limitations

The current study is the first to explore the dyadic relationships between parents’ and adolescents’ MFM trust/influence and their unhealthy food intake. By including actor–partner effects in the study design, the study captures the interplay of mutual influences of family dietary behaviors within the family. Another key strength of this study is the use of robust methods to assess both actor and partner effects via the APIM, effectively accounting for their interdependence and allowing for a nuanced assessment of these effects. Additionally, the use of a large national sample enhanced the generalizability of our results. Finally, another notable strength is the study’s significant translational potential, bridging the gap between research and policy. The findings provide evidence to support policies to regulate MFM to both adolescents and parents. Specifically, the direct effect of MFM on adolescents suggests that there should be policies regulating how unhealthy foods are marketed to this population and the indirect effect of MFM from parents to adolescents suggests that universal policies to reduce MFM of unhealthy food would benefit multiple generations.

This study is not without limitations. The FLASHE data are 11 years old and much has changed regarding MFM since the data were collected. However, the implications of the findings still stand today given the lack of attention to MFM trust/influence on adolescents’ dietary intake and the increase in the avenues for MFM to influence adolescents’ eating behaviors since these data were collected. This study is cross-sectional; therefore, causation and temporality cannot be determined. There may be a feedback loop between unhealthy food intake and MFM trust/influence such that intake of unhealthy food may provoke trust in MFM. Relatedly, the focus on one parent–adolescent dyad per household does not allow for examining the cumulative effect and/or interdependence of several family members’ behavior and food-related family dynamics in relation to MFM trust/influence. The study implications are limited by measurement. Future studies should focus on a more precise evaluation, particularly assessing trust in MFM related to branded, junk, and fast food, to better understand the specific aspects of MFM trust/influence that are most significant in shaping adolescents’ unhealthy dietary habits. Other measurement considerations include parents’ and adolescents’ digital literacy, media exposure and correlates, cultural context and background, as well as marketing factors such as brand loyalty and marketing tactics. Shared household factors (e.g., home media environment, neighborhood food environment, food availability) and adolescents’ out of home food environment including peer influence and the school environment likely affect the actor effects and interdependence of parent–adolescent dyads’ MFM trust/influence and unhealthy eating. Future studies should account for these and other relevant variables. Lastly, typical self-report measure biases such as recall and social desirability might have impacted the accuracy of the data collected.

6. Conclusions

This study examined the relationship between MFM trust/influence and unhealthy food intake in a national sample of parent–adolescent dyads. The interdependence patterns suggest that parents’ and adolescents’ MFM trust/influence best explained adolescents’ unhealthy food intake, while only parents’ MFM trust/influence explained parents’ unhealthy food intake. We conclude that MFM trust/influence has important dyadic relationships with adolescents’ unhealthy food intake, and it is important to address MFM trust/influence in both dyad members to mitigate the harmful effects of MFM on adolescents’ diet.