Examining determinants and co-associations between fruit and vegetable and sugar-sweetened beverage monitoring practices among a sample of low-income Hispanic mothers

Abstract

The etiology of obesity is complex, and a growing body of research suggests that changing two or more behaviors at the same time will likely have a stronger impact than changing one behavior in isolation. A lack of fruit and vegetable (FV) intake and high sugar-sweetened beverage (SSB) consumption are two behaviors related to childhood obesity in the Hispanic community. The purpose of this study was to evaluate theory-based determinants of monitoring practices of FV and SSB among Hispanic mothers with young children using the Reasoned Action Approach. The study also examined the co-associations among the theoretical antecedents for both behaviors. Surveys were distributed in several community settings to urban Hispanic mothers (n=238) of 2–5-year-old children. Surveys assessed FV and SSB monitoring practices, attitudes, perceived norms, perceived behavioral control (PBC) and intentions. Two models were evaluated using structural equation modeling: attitudes, perceived norms and PBC of both behaviors were associated with intentions for both behaviors, and PBC and intentions were associated with monitoring practices of FV and SSB. Both models had adequate fit [FV (CFI=0.973; GFI=0.980; RMSEA=0.121); SSB (CFI=0.994; GFI=0.993; RMSEA=0.066)]. PBC for both FV and SSB monitoring was significantly associated with SSB monitoring, whereas FV intentions was only significantly associated with FV monitoring. Results from this study gives evidence that among Hispanic mothers, FV and SSB monitoring behaviors are interrelated. To promote SSB monitoring, more implementation strategies (i.e. skills-based interventions) are warranted, whereas to promote FV monitoring, motivational strategies are warranted.

Introduction

Childhood obesity is one of the most critical public health issues in our society, as early onset places children at greater risk of remaining obese into adulthood [1, 2]. Currently, over 30% of children between the ages of two and five years are overweight or obese, however Hispanic children of the same age group are about three times more likely to have obesity compared to white children [3, 4]. During the first years of life parental feeding styles and practices play a critical role in preventing obesity [5]. Although not always a direct proxy, parental feeding styles and practices centered around food monitoring behaviors are important given its association with dietary intake. For example, in a sample of mostly Caucasian preschoolers (2–5 years), maternal monitoring was positively related to fruit consumption [6], and in another sample of overweight and at risk for overweight children (5–10 years) parental monitoring was inversely associated with SSB consumption [7]. This has also been demonstrated in research with Hispanic families. For example, among low-income Hispanic families an indulgent parental feeding style is common, which is indicative of placing very little boundaries on what children consume (i.e. low demandingness) [8]. Subsequently, this type of feeding style has been associated with a higher body mass index among young children from Hispanic, African American, and Caucasian families [8–9]. Higher levels of maternal monitoring have also been associated with lower FV consumption among young Portuguese children (4 years) [10]. Finally, parental feeding practices in Mexican American families, such as having a positive involvement with their children (i.e. parental monitoring/limiting high-calorie foods, encouraging/compliment healthy eating, encouraging variety, asking the child what he/she wants to eat, and providing small food servings to children), pressuring to eat, restriction, and using food to control behaviors, have all been associated with the risk of childhood obesity [11].

It is well known that many factors influence parenting behaviors, such as dietary awareness, knowledge and the home environment, however it has been noted that the emphasis on research in this area has largely focused on how feeding practices and styles are associated with the dietary intake, and the weight status of children, whereas few studies seek to understand the theoretical determinants of parental feeding behaviors and practices [12, 13]. Having a theoretical explanation of parental feeding practices will help researchers and practitioners understand what specific determinants influence these behaviors, which will aid in the development of public health interventions. Results from a recent review showed that among 83 theories in the field of behavior change The Theory of Planned Behavior (TPB) is the 2^nd most utilized theory [14]. The Reasoned Action Approach (RAA) is the most recent adaptation of the TPB, and like the TPB, posits that behavior is primarily determined by one’s intentions (or readiness to engage in a particular behavior), as long as the individual has sufficient skills and the environment is conducive towards performing the behavior [15]. In turn, intentions are determined by one’s attitudes (feelings of favorableness towards a behavior), perceived norms (normative pressure to enact a behavior) and perceived behavioral control (PBC) (the perception of the degree to which a person is capable of and has control over performing a behavior) [15].

The TPB/RAA are models that account for single/well-defined behaviors, such as condom use and cigarette smoking [16–19]. However, to date, only a few studies have used the theoretical determinants of the TPB/RAA for understanding parenting practices pertaining to obesity prevention. In one study, researchers evaluated both mothers and fathers of preschool-aged children for limiting screen time using the TPB [20], and in a related study by the same authors, maternal practices related to physical activity and screen time were evaluated [21]. In another study, the TPB constructs were used to predict SSB consumption among parents and their adolescents, and researchers evaluated the extent that encouraging drinks other than SSB and discouraging SSB from a parent had on the adolescent’s SSB intake [22]. Additionally, the Integrated Behavioral Model (a similar model to the TPB/RAA) was used to predict mother and fathers’ monitoring practices related to sugar-sweetened beverage (SSB) consumption of 7–12 year old children [23].

While these studies support the TPB as a robust framework for understanding parental behaviors it should be noted that they were done with primarily affluent, Caucasian parents [20–23]. Low-income, less acculturated Hispanic mothers’ parental perceptions regarding childhood obesity and parenting behaviors centered around health are different from their affluent, Caucasian peers. Several studies on low-income, foreign-born Hispanic mothers have demonstrated that they underestimate their overweight/obese child’s weight status [24, 25]. Further, low-income, less acculturated Hispanic mothers are less likely to perceive child obesity as an indicator of poor health [26]. Instead these mothers perceive non-clinical measures, such as good appetite and physical activity, as indicators of health and use culturally acceptable terms such as “thick”, “solid”, or “big-boned” to describe size [27]. Thus, it has been suggested that a different framework that considers the cultural weight preferences and health beliefs of low-income, less acculturated Hispanic mothers is needed when designing childhood obesity prevention programs [28].

Obesity prevention intentions are rarely implemented to change single/well-defined behaviors. Rather, they are often implemented to promote multiple behaviors that leads to the reduction of energy consumption, and increase energy expenditure. The same can also be true for promoting parental feeding practices. For example, the American Heart Association and the American Academy of Pediatrics encourages parents to promote foods that contribute to positive growth and development like fruits and vegetables (FV), while also limiting the consumption of foods that have little to no nutrient value such as SSB [29]. SSB intake is among the top contributors to the development of obesity in Hispanic communities [30]. Research shows for Mexican Americans, SSBs are the second largest overall calorie source (6.3% of total daily caloric intake), only surpassed by Mexican Mixed Dishes (11.4% of total daily caloric intake) [30], and results from another study showed only 9.2% of Hispanic children (6–11 years) met recommendations for having 0 daily servings of SSBs [31]. Low consumption of fruits and vegetables is also often noted as an important target for public health and community interventions focused on obesity prevention [32]. Besides obesity, fruit and vegetable intake has been linked to reducing the risk for type-2 diabetes [33], cardiovascular disease [34], and certain types of cancers (i.e. breast [35] and lung [36] cancers). Despite these benefits however, results from the aforementioned study showed that only 6.0% of Hispanic children (6–11 years) met recommendations for having ≥5 daily servings of fruits and vegetables [31].

Understanding these two parenting practices together (SSB and FV) stands to contribute more to obesity prevention efforts, than if one behavior was evaluated in isolation. This is the premise behind multiple health behavior change interventions (MHBC), which are theorized to have a greater impact than single-behavior change interventions [37]. Currently, research on MHBC is limited. However, a recent call was made for more theoretically driven research that explored how motivating factors of inter-related behaviors are cross-related [37]. Two behaviors that are increasingly important in the Hispanic community are FV and SSB consumption. Therefore, the purpose of this study was to evaluate theory-based determinants of these two parenting practices among low-income Hispanic mothers, and examine the co-associations among the theoretical antecedents for both behaviors.

Methods

Data Collection

Surveys were administered in person, via one-on-one interviews with a bilingual researcher. Given the length of the survey, and the inexperience mothers had participating in research such as this, mothers were interviewed so everything on the survey could be read to them and the researchers could confirm they understood each element. This also helped mothers finish the survey in a timely manner, and helped when they had questions about the survey content. The research team included Native Spanish-language speakers who have lived in the US for more than 8 years, and were fluent in English. All interviewers were trained by the principal investigator (PI) of this study (K.L.) who is bilingual and bicultural, and has extensive research experience working with Spanish-speaking families in the US. The PI trained members of the research team on appropriated methods to collect information from Spanish-speaking participants with low educational attainment and low health literacy. This included going one-by-one through each page of the survey to assure the team understood the survey questions, and the intent of what the questions were supposed to measure. During the time members of the research team were able to ask questions about potential barriers they may encounter, or what to do if mothers did not understand what was being asked of them. The PI also shared her experiences working with Spanish-speaking families, and gave advice for best practices. Finally, the PI arranged meetings with the research team and leaders from local community agencies to discuss the best ways to approach mothers in the field to recruit them for participating in the study, and efficiently administering the survey.

Hispanic mothers were recruited from numerous urban settings in the community (i.e. churches, community agencies, preschools, and daycares) in five zip codes in the southwest part of Oklahoma City where a majority of the Hispanic population resides. Recruitment strategies included the use of media (TV, radio, and newspaper advertisements), fliers, and a lay community health worker visit to recruitment sites. A pre-screening was conducted, and to meet inclusion criteria participants had to self-identify as Hispanic, be living with their 2-to-5-year-old child, and the family was low income per Supplemental Nutrition Assistance Program eligibility criteria. Also, as part of the pre-screening process mothers reported that their preschooler did not have any special dietary needs or developmental disabilities.

Instrumentation

The Reasoned Action Approach (RAA) constructs were oriented towards two maternal monitoring behaviors and defined as: Making sure my preschooler eats half of his/her plate filled with fruits and vegetables, at least 5 days a week (Spanish translation as it appeared on the instrument—Asegurarme de que mi hijo(a) coma la mitad de su plato de frutas y verduras por lo menos 5 días a la semana) and Make sure my preschooler does not drink sugary beverages (Kool-Aid®, Capri-Sun®, Soda, Tampico®)” (Spanish translation as it appeared on the instrument - Asegurarme de que mi hijo(a) no beba bebidas azucaradas (Kool-Aid®, Capri-Sun®, Soda, Tampico®). These two behaviors were chosen based on a community needs assessment that was implemented with Hispanic mothers of preschoolers in Oklahoma City [38]. During the assessment, Hispanic mothers (n=55) participated in 9 focus groups, in which they identified 38 different behaviors they believed contributed to obesity in children. The behaviors were then rated as “Low”, “Medium”, or “High” on its level of importance and changeability by the study investigators, and the two behaviors used in this study were deemed as most changeable and important.

Each construct of the RAA was evaluated for each behavior (intentions, attitudes, perceived norms, and perceived behavioral control (PBC)). For both behaviors, three items evaluated intentions (i.e. I intend to <“behavior” was either Make sure my preschooler does not drink sugary beverages or Make sure my preschooler eats half of his/her plate filled with fruits and vegetables at least 5 days a week>); 6 items evaluated attitudes (i.e. <Doing the “behavior”> will help my preschooler… examples in included […be healthy] or […not become overweight]); 12 items evaluated perceived norms (i.e. My husband/partner thinks that I should <do the “behavior”>; and 9 items evaluated PBC (i.e. How much control do you have for making sure your preschooler <does the “behavior”>). All scales were normalized to −2 to +2 by taking the mean of the items in the scale [i.e. indicating a strong negative attitude (−2) to a strong positive attitude (+2)]. For each monitoring behavior mothers responded to 1-item: On average, how many days per week do you make sure your preschooler eats half of his/her plate filled with of fruits and vegetables? < En promedio, ¿cuántos días por semana usted se asegura que su hijo(a) coma la mitad de su plato de frutas y verduras?>; and On average, how many days per week do you give your preschooler sugary beverages? <En promedio, ¿cuántos días por semana usted se asegura que su hijo(a) nunca tome bebidas azucaradas? (i.e. Kool-Aid, Capri-Sun, Soda, Tampico)>. Responses were from 0 days (días) to 7 days (días) per week.

Using multiple indicators, the instrument was deemed valid (via face and content validity by a panel of experts and construct validity using confirmatory factor analysis [SSB constructs model fit (Comparative Fit Index (CFI) = 0.94, Tucker–Lewis Index (TLI) = 0.924, and Root Mean Square Error of Approximation (RMSEA) = 0.066); FV constructs model fit (CFI=0.972, TLI=0.964, and RMSEA=0.046)]) and reliable (via Cronbach’s alpha scores [SSB constructs scores (0.72 to 0.94); FV constructs scores (0.80 to 0.91)]), besides the autonomy scales (SSB=0.41; FV=0.59). Further details about the Reasoned Action Approach instrument (i.e. items for all of the scales, scoring and psychometric values) can be found elsewhere [38].

Acculturation was also assessed as a background variable using three questions: number of years of residence in the U.S., place of birth, and the language participants generally read and speak in the home (response options: only Spanish, Spanish better than English, Both equally, English better than Spanish, and only English) [39].

All of the instruments used for this study were initially created in English and translated into Spanish. The protocol for translating the instruments from English to Spanish followed Marin’s double translation methodology [40]. Double translation is also referred to as back translation. This method requires the use of at least two bilingual translators who work independently. First the English-speaking researcher starts with an English-language version of the instrument, and then one member translates the English version to the target language (in this case, Spanish). The second member then blindly translates the new version in the target language (Spanish) back into the original language (English). The two English versions are then checked for discrepancies. Any inconsistencies between the two versions are then identified and discussed with the translators until a consensus is reached [41]

Heights (m) and weights (kg) for all participants (mothers and preschoolers) were measured in triplicate per standardized procedures [42]. Height was measured using a stadiometer, and weight was measured using a digital scale. Body mass index (BMI) for the mothers was determined using the standard formula (i.e. kg/m²) [43]. BMI-for-age percentile for the preschoolers was determined using the Children’s BMI Tool for Schools [44].

Statistical Analysis Plan

Given how the RAA constructs have a systematic path (see Figure 1), structural equation modelling was employed to evaluate two models (SPSS AMOS version 22.0). In the first model attitudes, perceived norms and PBC related to FV and SSB monitoring were associated with intentions for both behaviors, and then intentions and PBC for both behaviors were associated with monitoring practices related to SSB monitoring. The second model was the same, except FV monitoring was used in place of SSB monitoring. Appropriate indices were used to evaluate model fit [i.e. Comparative Fit Index (CFI ≥0.95 deemed acceptable), Goodness of Fit Index (GFI ≥0.95 deemed acceptable), and Root Mean Square Error of Approximation (RMSEA ≤0.08 deemed acceptable)] [45]. This study was approved by the sponsoring institution’s Institutional Review Board.

Figure 1.

Two hypothesized models for a) Fruits and Vegetable monitoring practices and b) Sugar Sweetened Beverage monitoring practices

Results

Overall, 238 Hispanic mothers of preschoolers participated in this study. The mean age of the mothers was 33.1 years (±6.4), and for their children it was 3.8 years (±1.1). Obesity rates were high as almost half of the mothers were obese (41%) and slightly over a quarter of their children were obese (26%). Most of the mothers (92%) were born outside of the US and a majority had been in the US at least 11 years (61%). Additionally, almost all of the mothers (80%) spoke only Spanish or Spanish better than English at home. Over half of the mothers in this study were unemployed, and 73% attained a high-school education or less.

Mean and standard deviations for the RAA constructs can be found on Table 1. Results showed that mothers had generally positive intentions, attitudes and PBC for monitoring FV, but neutral perceived norms. Mothers also had generally positive intentions to monitor SSB, but mostly neutral attitudes, perceived norms and PBC. Table 1 also showed that mothers generally had significantly higher monitoring practices (p<0.001; d=0.53), stronger intentions (p<0.001; d=0.30), more favorable attitudes (p<0.001; d=1.61), and higher levels of PBC (p<0.001; d=1.47) for FV compared to SSB. Results from Table 2 also shows that there was moderate association between intentions (r=0.372; p<0.001), attitudes (r=0.172; p<0.01), and PBC (r=0.242; p<0.001), for FV and SSB monitoring, but there was a strong association between PN (r=0.895; p<0.001).

Table 1:

Differences between Reasoned Action Approach constructs for Fruit and Vegetable and Sugar Sweetened Beverage monitoring practices

Variables	Possible Range	Fruits & Vegetables m (SD)	SSB m (SD)	t-statistic	p-value	effect size Cohen’s d
Parental Monitoring (in days)	0–7	4.86 (1.98)	3.68 (2.46)	7.104	0.001	0.53
Intentions	−2 to +2	1.54 (0.72)	1.26 (1.12)	3.963	0.001	0.30
Attitudes toward Behavior	−2 to +2	1.54 (0.87)	−0.23 (1.33)	18.658	0.001	1.61
Perceived Norms	−2 to +2	−0.55 (1.01)	−0.55 (1.04)	0.057	0.863	--
Perceived Behavioral Control	−2 to +2	1.31 (0.66)	0.32 (0.69)	18.447	0.001	1.47

Table 2.

Pairwise correlation (r) among Reasoned Action Approach constructs for Fruit and Vegetable and Sugar Sweetened Beverage monitoring

	1.	2.	3.	4.	5.	6.	7.	8.
1. FV Behavioral Intentions	-	.204**	.213**	.502***	.372***	−.023	.202**	.261***
2. FV Attitudes towards Behavior		-	.153*	.307***	.418***	.172**	.160*	.283***
3. FV Perceived Norms			-	.298***	.128*	.046	.895***	.192*
4. FV Perceived Behavioral Control				-	.233***	.020	.255***	.242***
5. SSB Behavioral Intentions					-	.162*	.191**	.318***
6. SSB Attitudes towards Behavior						-	.074	.317***
7. SSB Perceived Norms							-	.157*
8. SSB Perceived Behavioral Control								-

Notes:

^***p≤.001;

^**p≤.01;

^*p≤.05.

FV (Fruits and vegetables); SSB (Sugar-sweetened beverages)

Figure 1 shows the hypothesized path models. Figure 2 shows the results for the final path model for FV, and Figure 3 shows the results for the final path model for SSB. All non-significant paths were excluded in the final model to aid in interpretation.

Figure 2.

Final model for Fruits and Vegetable monitoring practices

Note: Paths and standardized regression weights significant at ***p<.001; **p<0.01; *p<0.05.

FV (Fruit and Vegetable); SSB (Sugar Sweetened Beverages)

Figure 3.

Final model for Sugar Sweetened Beverage Monitoring Practices

Note: Paths and standardized regression weights significant at ***p<.001; **p<0.01; *p<0.05.

FV (Fruit and Vegetable); SSB (Sugar Sweetened Beverages)

Final Path Model for FV

Overall, the model fit was mostly adequate for predicting FV monitoring practices (CFI=.973, GFI=.980, RMSEA=.121). FV attitudes (b=.407, standard error [SE]=.078, p=.001; β=.319), FV perceived norms (b=−.344, SE=.142, p=.016; β=−.312), SSB perceived norms (b=.390, SE=.136, p=.004; β=.364), and SSB PBC (b=.324, SE=.102, p=.001; β=.200) were significant predictors of SSB intentions, and collectively predicted 25.1% of the variance of SSB intentions (R²=.251); FV PBC (b=.484, SE=.067, p=.001; β=.441), and SSB PBC (b=.176, SE=.064, p=.006; β=.169), were significant predictors of FV intentions, and collectively predicted 28.5% of the variance of FV intentions (R²=.285); and FV intentions (b=.503, SE=.211, p=.017; β=.183) was the only significant predictor of FV monitoring practices, accounting for 5.5% of the variance (R²=.055).

Final Path Model for SSB

The model fit was also adequate for predicting SSB monitoring practices (CFI=.994, GFI=.993, RMSEA=.066). The paths for predicting FV and SSB intentions were the same as previous model, however for predicting SSB monitoring practices, SSB PBC (b=−1.042, SE=.236, p=.001; β=−.292) and FV PBC (b=.556, SE=.268, p=.038; β=.148) were the significant predictors, accounting for 10.8% of the variance (R²=.108).

Discussion

Feeding practices that are established in early childhood can persist into adolescence and adulthood, and can significantly contribute to the development or prevention of obesity in childhood [46, 47]. The purpose of this study was to add to the literature by understanding theory-based determinants of two monitoring practices of low-income Hispanic mothers (FV and SSB consumption), and understand how are they are inter-related. Bringing this information to light will help to make implications for future public health interventions that target childhood obesity in the Hispanic community. According to the final path models, 10.8% of the variance of maternal monitoring was explained for SSB monitoring and 5.5% of FV monitoring was explained. Although these results are promising, they also suggest that the RAA may have limited utility for these behaviors in this population. Meta-analyses of prospective TPB [48] and RAA [49] studies report these models typically explain 19.3% to 30.9% of the variance of health behaviors. It should be noted however, that these were for mostly personal behaviors (i.e. abstinence, cancer detection, of physical activity), and not complex behaviors like parental monitoring practices.

Previous studies have shown promise that the TPB may be an effective model for predicting parenting practices related to obesity prevention. In a previous study with Australian parents, monitoring practices related to screen time (i.e. my child’s screen time is limited to <1 h per day in the next week’) were evaluated, and results showed that intentions and PBC predicted 46% of the variance of screen time monitoring practices [20]. In another study in the United States with mothers and fathers, intentions (along with perceived skill and environment) predicted 28.5% of the variance of SSB monitoring practices [23]. These studies cannot be compared to our study in full however, because the samples from these studies were from primarily affluent, Caucasian samples of both mothers and fathers, and our sample was based on low-income Hispanic mothers, and fathers were excluded. As previously noted, low-income, less acculturated Hispanic mothers have different health beliefs and cultural weight preferences compared to their peers [24, 25]. For example, low-income, foreign-born Hispanic mothers tend to underestimate the weight of their overweight/obese child, do not perceive obesity as a poor health indicator, and use the terms “thick”, “solid”, and “big boned” to describe their children’s size [26, 27]. Further, it is important to note that even though not all of the RAA constructs were significantly related with FV and SSB monitoring behaviors, it does not discount the usefulness of the model in its entirety. As Fishbein and Ajzen [15] have explained, the relative importance for theory-based determinants of behaviors can vary from group to group, and across different cultures. For example, in a review of 30 studies using the TPB, when predicting behavioral intentions for different behaviors, attitudes was the strongest predictor for 16 of the behaviors (ex. walking for pleasure), whereas subjective norms and PBC were the strongest predictor for the other 14 behaviors [50]. Therefore, future work is needed to understand additional predictors of maternal feeding practices in this population, that go beyond intentions and PBC, and could be added to this model such as acculturation or social support.

One of the theories about how multiple health behavior change (MHBC) interventions function is that overlap likely exists in theoretical antecedents of ‘like behaviors’, and therefore, changes in one antecedent will likely create changes to others [51]. In the context of childhood obesity prevention interventions where multiple behaviors are being promoted such as FV and SSB monitoring, this may translate as changes in attitudes to monitor FV could simultaneously create changes in attitudes to monitor SSB. For example, mothers who change their attitudes about monitoring FV through a public health intervention may also change their attitudes about monitoring SSB, even though the intervention did not directly target SSB monitoring. This change is theoretically created because mothers may group FV and SSB monitoring behaviors under the category of obesity prevention behaviors for my child. According to Table 2, intentions (r=0.372; p<0.001), attitudes (r=0.172; p<0.01), and PBC (r=.242; p<0.001) between monitoring FV and SSB, although significant, did not have strong associations. This indicated that even though these two behaviors both contribute to the prevention of obesity, Hispanic mothers’ beliefs about monitoring these foods are not highly related. This likely shows that the sources of these beliefs are different among mothers, and thus different intervention strategies and behavior change techniques may be needed to change these behaviors, given the small amount of overlap they contain. This was not found for perceived norms however, which showed a considerable overlap between behaviors (r=0.895; p<0.001). This demonstrates the main tenet of multiple health behavior change interventions, in that norms about obesogenic behaviors likely overlap, that if mothers feel pressure to monitor one obesogenic behavior, they may also innately feel pressure to monitor other behaviors they consider obesogenic. This has yet to be empirically tested however, and would be an interesting area to conduct future research. Furthermore, it would also be vital to understand what other behaviors mothers may consider obesogenic.

To put our results into practical recommendations for public health interventions, Steinmetz and colleagues [52] recommends differentiating motivational processes, or those that help individuals create favorable intentions towards performing a behavior, and implementational processes, or those that help people act upon their intentions by enhancing PBC. Based on the final models, it appears that to promote FV monitoring, interventions should focus on motivational processes to enhance FV intentions. To promote SSB monitoring, interventions should focus on methods to enhance implementational processes by promoting SSB and FV PBC. This again demonstrates the difference between how public health practitioners should consider promoting FV and SSB monitoring among Hispanic mothers. According to a recent meta-analysis on TPB/RAA behavior change interventions, the most utilized implementational methods focus on increasing skills, planning, goal-setting, and self-monitoring, and the most utilized motivational methods were providing information (i.e. advantages/disadvantages of monitoring child food intake), persuasion and social encouragement [52]. For TPB/RAA based interventions, public group-based programs also tend to have a strong effect than private, one-on-one interventions [52].

This study has a number of strengths. First, our survey used a strong theoretical framework that has been well-established in social and behavioral science. We then extended the theory to multiple health behavior change research as an area of potential growth. We also recruited a difficult to reach target population, given the potential language-barrier, and concerns for privacy this population contains. To overcome these barriers we used trained bi-lingual researchers, and reached out to local community agencies to assist in recruitment and build trust among community members. Finally, we administered our survey via using interviews rather than self-directed means. This was important because the mothers in our study were able to ask questions while taking the survey, which improved our data quality. This study is not without limitations however. First, data were based on self-report; therefore participant responses may have been biased, untruthful or incorrect, or participants may have misinterpreted questions and provided erroneous answers that could introduce error in our measurement models. Not much can be done however, since the RAA variables are psychosocial in nature. Second, the outcome variable of interest was based on parental monitoring, rather than actual child intake of FV and SSB, and it is unknown what the relationship was between these variables (monitoring and intake) in our sample. Third, a convenience sample of Hispanic mothers participated in this study, thus our results may not be generalizable to other racial and ethnic groups. Forth, the behaviors of interest evaluated in this study (FV and SSB monitoring) were only evaluated using one item. In the future, it would be more robust to evaluate these variables in more comprehensive ways, such as a 24-hour recall or 7-day food records [14-new]. Fifth, the psychometrics were not tested in the translated measures. Finally, this study utilized a cross-sectional research design, thus making it premature to make any causal inferences between the variables under investigation.

Conclusion

During early childhood parents transmit cultural and family eating traditions and practices that teach children what, when, and how much of specific foods and beverages they should consume [5, 47]. It is therefore useful to understand significant theoretical determinants of parental feeding practices, because such information can guide the development of public health interventions. Our results show preliminary work that suggests significant theory-based determinants of FV and SSB monitoring practices of low-income Hispanic mothers, and how the determinants are inter-related. However, results from the final models show that there is room for improvement by introducing additional factors that are not included in the RAA. Clearly, more work is needed in this area to predict, and ultimately change, monitoring practices related to obesity prevention among Hispanic mothers. In addition, what was not evaluated in this article was the translation of maternal monitoring practices to child consumption patterns. While ultimately ‘success’ is defined as whether children eat fruits and vegetables, or refrain from consuming SSBs, what was evaluated in this research is important because it shows the mechanisms that are involved in this process.