Exploring a theory of change: Are increases in parental empowerment associated with healthier weight-related parenting practices?

Abstract

Introduction:

Parent health-related empowerment is defined as the process by which parents realize control over their life situation and take action to promote a healthier lifestyle. For decades, researchers have described the theoretical potential of empowerment in health promotion efforts, though few have empirically examined this hypothesized relationship. This study is one of the first to examine the relationship between parental empowerment and healthy weight parenting practices (i.e., food, physical activity, sleep, and media parenting), as a mechanism for early childhood health promotion in community settings.

Methods:

Low-income parents of preschool-aged children attending Head Start in Greater Boston between fall 2017 and spring 2019 were invited to complete a survey in the fall and spring of each academic school year (n=578 with two surveys and n=45 with four). Parental empowerment and healthy weight parenting practices were assessed using validated surveys. We used a multilevel difference-in-difference approach to estimate changes in healthy weight parenting practices score by changes in parental empowerment score.

Results:

Out of a possible score of four, the unadjusted mean (SD) score in fall was 3.20 (0.40) for empowerment and 3.01 (0.40) for parenting. An increase in parental empowerment was associated with an increase in healthier parenting practices (b=0.14; 95% CI=0.08, 0.20; p<0.0001).

Conclusions:

Parent empowerment may be an important target in interventions to prevent obesity in low-income children.

INTRODUCTION.

Most behaviors associated with healthy growth and weight – such as diet¹, physical activity², and sleep³ – are acquired during a child’s first few years of development^4–6. In this way, early childhood is considered a critical period for addressing risk factors related to healthy development^7–9. Without intervention, exposure to poor nutrition and behavioral risk factors in early childhood can carry adverse influences over the life course¹⁰. Those who develop overweight in their first few years are more likely to maintain overweight into adulthood, and experience heightened risk of related morbidity and mortality¹¹. There is also evidence to support the transmission of this risk from parent to child^12–16, through both biological and social mechanisms. As a means of breaking this cycle, family- and parent-centered health promotion strategies have gained attention in recent years¹⁷.

Substantial literature supports the linkage between parenting (i.e., knowledge¹⁸, practices^19,20, modeling²¹, and style²²) and healthy weight in early childhood^23,24; however, the mechanisms which underly sustainable parental behavior change are less understood. One proposed target of parent-centric health promotion interventions is that of parents’ health-related empowerment^25–27, defined as the process by which parents realize control over their life situation and take action to promote a healthier lifestyle^26,28–31. Conceptually grounded in Zimmerman’s (1992) Empowerment Theory³², the construct of parental empowerment encompasses critical awareness (i.e., recognizing that a range of factors including parenting practices, community factors and media affect children’s health behaviors)³², resource empowerment (i.e., knowing who to speak to and what questions to ask to get access to resources supporting child health behaviors)²⁷, and relational empowerment (i.e., leveraging personal relationships to promote child health)³³. Parents often express empowerment, for example, by developing an awareness for their unique socioecological context, a recognition of available resources, and an embrace of relationships beneficial for child health promotion²⁵. In this way, empowerment arises from the union of three domains central to positive parenting^26,27, including parent belief (e.g., self-efficacy) ³⁴, competency (e.g., knowledge about child health)^35,36, and behavior (e.g., actions taken to promote child health)^19,37. In relation to healthy parenting practices, increases in empowerment could manifest, for example, as acknowledging the link between parent and child health behavior, engaging with early childhood educators when in need of help or advice, or proactively sharing helpful information with other parents of young children.

There remains a limited understanding of the role of parental empowerment in the pathway from parenting to health outcomes in early childhood. Although a sizable body of literature has previously described the theoretical promise of empowerment in health promotion³⁸, few have empirically examined this hypothesized relationship^39–43. Of the work that has been done, a striking majority has focused on disease management^44–46, instead of prevention; few studies have examined the role of parental empowerment in preventing metabolic conditions from developing in the first place^40–43. Additionally, most of the empirical studies of empowerment in relation to health promotion have taken place in the context of low- and middle-income countries^47–51 which face different health outcomes of concern, such as stunting and wasting, compared to high-income countries. Therefore, across this body of literature, methods for operationalizing the construct of empowerment in the U.S. nutritional health context⁵¹ remains unclear.

Using a difference-in-difference approach, the current study is among the first to assess the relationship between the process of increasing parental empowerment²⁵ and acquiring healthier weight parenting practices⁵² (i.e., change in parental empowerment predicting change in parenting practices). Informed by empowerment theories^29,31,32, we hypothesized that, among parents of preschool-aged children from low-income households, increased parental empowerment is associated with increases in healthier parenting practices related to child weight and growth.

METHODS.

Research setting.

Communities for Healthy Living (CHL) is a family-centered intervention (NCT03334669; R01DK108200)⁵³ designed to prevent the development of obesity in pre-school-aged children enrolled in Head Start. Between 2016–2019, CHL was implemented and evaluated in a cluster-randomized controlled trial using a stepped wedge design⁵³ across 16 Head Start programs in the Greater Boston area. Collectively, these programs serve over 1650 children and their families each year⁵³. A comprehensive description of CHL, including its theoretical background, the intervention design, and the families served by the participating programs are reported elsewhere ⁵³. The study was initially approved by Institutional Review Board at the Harvard T.H. Chan School of Public Health and then again by the Institutional Review Board at Boston College when the study team relocated (June 19, 2019).

Study design and data collection.

The current study is observational in design; longitudinal data collected across four consecutive academic semesters (2017–2019) were used to create and examine change scores for both parental empowerment and healthy weight parenting practices. The survey included nine questions on healthy weight parenting practices including food, physical activity, media and sleep parenting practices (i.e., the Obesity Parenting for Intervention (OPTION) scale) and fifteen questions on parental empowerment (i.e., the Parental Empowerment through Awareness, Relationships, and Resources (PEARR²⁵) scale). Change in parenting practices and parental empowerment – from fall to spring of an academic school year – were examined through the current study.

Procedures

In the fall of the 2017–2018 and 2018–2019 academic years, all parents of children ages 3 to 5 years enrolled in Head Start programs participating in CHL (23 individual sites, nested within 16 programs), including intervention and control sites, were invited to complete a self-reported health behavior survey (n=2545 families; Supp Figure 1). Paper copies of the survey, a consent form, brief instructions, and a return envelope were sent home with each child; online survey completion was also offered. All participants provided informed, written consent before completing the survey. All materials were available in English, Spanish, and Chinese. Where possible, research staff met with parents at their childcare centers to explain the study and distribute the survey materials. Parents who completed the survey in fall were re-contacted in spring and asked to complete the survey a second time for the same child. In the context of this analysis, fall measures in 2017 and 2018 are referred to as “baseline” data (collected from parents near the start of the school year) and spring measures as “follow-up” data (collected from parents near the end of the school year).

Study sample.

The target sample included all unique family units with at least one child enrolled at the start of the 2017 or 2018 school year; for each family, the parent served as the unit of analysis. We define “parent” as the primary caregiver within each family. For inclusion in the analytic sample, parents were required to have at least one child enrolled in Head Start for a full academic school year (i.e., first enrolled between September 1 and October 15 and disenrolled for summer break no earlier than May 15, with no longer than a thirty day leave of absence during the school year).

Each parent was also required, for inclusion in our difference-in-difference analysis, to have change scores for parental empowerment and parenting practices, which were calculated as the difference between their baseline and follow-up scores during an academic school year. In other words, each family was required to have completed at least two surveys within either the 2017 or 2018 school year (i.e., fall of 2017 and spring of 2018, or fall of 2018 and spring of 2019). Those with only one survey completed were not included in the analysis, for lack of change scores. Of the 2545 parents invited to complete the survey, 893 (35.0%) completed at least one survey; 1643 (64.6%) parents did not complete a survey (Supp Figure 1). Of those who completed at least one survey, 623 parents completed a baseline and follow-up survey in at least one of the two school years (233 (37.4%) in the 2017 school year, 345 (59.7%) in the 2018 school year, and 45 (7.2%) in both school years).

Demographic and socioeconomic variables.

Data on general demographics and socioeconomic factors were extracted from Head Start enrollment records and merged with survey data via unique parent identifiers. Parent demographic variables included age (continuous in years), sex (male, female), and race/ethnicity (non-Hispanic (NH) Asian, NH Black (or African American), Hispanic/Latino (any race), NH white, and NH other (including biracial and multiracial)). Socioeconomic variables included marital status (married, other), educational attainment (less than high school, high school graduate, more than high school) and employment status (unemployed, employed, other).

Predictor: Change in parental empowerment.

The Parental Empowerment through Awareness, Relationships, and Resources (PEARR) survey includes 15 questions on the development of context-specific critical awareness, the recognition of available resources, and the embrace of relationships beneficial for the promotion of their child’s health²⁵. Our team developed this scale in response to the absence of clear literature on the measurement of parent health-related empowerment in the setting of childhood obesity prevention; emphasis in question design was placed on simplicity of wording for the sake of clarity and comprehension across a wide range of literacy levels. Items were drawn and adapted from previous literature – representing diverse languages and contexts^54–59 – on resource empowerment^33,57,60,61, critical awareness or consciousness^62,63, and relational empowerment^33,57,59,60. A previous paper²⁵ reports on the process to develop the scale along with its factor structure and internal consistency²⁵. Briefly, researchers used principal component analysis and exploratory factor analysis to test the factorial structure of the scale²⁵. Using confirmatory factor analysis, researchers confirmed the structure of the final subscales, which demonstrated strong internal consistency (α= 0.83–0.90)²⁵.

For each item (previously published²⁵, Supp Table 1), such as “I know how to find programs, services, or other resources in my community”, parents were asked to rate their level of agreement on a scale of one to four (strongly disagree to strongly agree). As other researchers have done in similar studies of caregiver empowerment^{44,56,57,64–66}, we created a single composite parental empowerment score by averaging the responses across all items (range=1–4, items=15). A higher score indicates higher empowerment. The items in this score demonstrated adequate internal consistency (Cronbach’s alpha= 0.88; inter-item correlations range from 0.13 to 0.72).

Outcome: Change in healthy weight parenting.

As has been previously reported⁵², our team developed the Obesity Parenting for Intervention (OPTION) survey to assess healthy weight parenting practices (i.e., actions and behaviors)– as opposed to parenting style (i.e., attitudes and emotions)⁶⁷ – across the domains of food, physical activity, and sleep. The goal was to compile a brief measure tapping all dimensions of healthy weight parenting that could be used in intervention studies; items (n=9) were drawn from validated scales -- such as the Comprehensive Feeding Practices Questionnaire^68–70, the Activity Support Scale for Multiple Groups^71,72, the Sleep Parenting Scale for Infants⁷³ -- which have been applied in diverse family contexts. As previously published⁵², confirmatory factor analysis supported adequate factorial validity of the OPTION scale, and multi-group structural equation modeling supported adequate measurement invariance of the scale by survey language (English and Spanish) and respondent sex (male and female) ⁵². In line with the scoring methods of previous studies with similar scales^74–77, we chose to average item responses across domains to create a single healthy parenting score. Previous literature supports the co-occurrence or clustering of multiple health behaviors ^76–79 and, as we observed in our data, responses across the different domains are often correlated (inter-item correlations range from −0.14 to 0.59; Cronbach’s alpha= 0.62)⁵².

Statistical analysis.

To examine differences in demographic and socioeconomic characteristics of those included in the analytic sample versus those eligible, but not included, we conducted chi-squared tests. Factors that differed significantly between the two samples were included parent age, race, ethnicity, level of spoken English proficiency (Table 1).

Table 1.

Characteristics of Head Start parents in their first fall enrollment, 2017–2019.

	Included a (n=623)	Excluded b (n=1922)	Total c (n=2545)	P-valued
First enrolled
Fall 2017	58.1	63.3	62.0
Fall 2018	41.9	36.7	38.0
Sex				0.6
Male	2.9	2.6	2.6
Female	96.3	97.1	96.9
Missing	0.8	0.3	0.4
Age (years)				<0.0001
17–29	26.7	38.1	35.3
30–35	32.9	30.5	31.1
36+	38.0	30.9	32.6
Missing	2.4	0.5	0.9
Race / Ethnicity				<0.0001
NH Asian	14.8	7.4	9.2
NH Black	31.3	37.2	35.7
Hispanic/Latino	40.6	43.4	42.7
NH Othere	2.1	2.2	2.2
NH White	10.1	7.4	8.1
Missing	1.1	2.5	2.1
Primary language				0.1
English	29.4	32.8	31.9
Spanish	35.3	35.9	35.8
Other	34.5	30.7	31.6
Missing	0.8	0.7	0.7
Spoken English proficiency				<0.0001
Not at all or not well	37.1	27.4	29.8
Well or very well	61.6	70.8	68.6
Missing	1.3	1.8	1.7
Number of parents in the house				<0.0001
1	54.7	69.1	65.6
2	45.3	30.9	34.4
Educational attainment				0.3
< High School	21.2	24.7	23.9
High School	37.4	37.8	37.7
> High School	37.2	35.9	36.2
Missing	4.2	1.6	2.2
Employment status				0.07
Unemployed	24.7	28.4	27.5
Employed	26.3	22.8	23.7
Otherf	44.8	48.4	47.5
Missing	4.2	0.4	1.3

Shown is the percent.

Note: Due to missing data, some categories may not sum to 100% of the study sample.

^aThose included in the analysis had at least one year of change data available (i.e., both fall and spring measures from the same year).

^bThose excluded from the analysis did not have change data available.

^cThe target sample includes all family units with at least one child enrolled at Head Start at the start of the 2017 or 2018 school year.

^dShown are the p-values of the chi-squared tests run to examine whether the frequency of demographic and socioeconomic were significantly different in the analytic sample versus those eligible, but not included.

^eThe category “other” includes parents who identified as biracial/multiracial, as well as American Indian, Alaska Native, Native Hawaiian or Pacific.

^fThe category “other” includes those who reported being in a job training program, being in school full-time (without parttime employment), “not applicable”, or “other.”

We tabulated the unadjusted mean scores for parental empowerment and healthy parenting, at baseline and follow-up, in Table 2. We also reported the mean change (follow-up minus baseline scores) in empowerment and parenting scores, first among all participants and secondarily stratified by change in empowerment: (1) those who reported increased empowerment and (2) decreased empowerment score.

Table 2.

Mean parental empowerment and healthy weight parenting scores, stratified by those who increased or decreased in empowerment.

	Total parent-years	Report empowerment d	Report disempowerment d
	n=668 c	n=324	n=270
Parental empowerment a
Baseline (fall)	3.20 (0.40)	3.03 (0.35)	3.40 (0.37)
Follow-up (spring)	3.23 (0.41)	3.39 (0.38)	3.05 (0.39)
Change b	0.03 (0.45)	0.36 (0.30)	−0.35 (0.34)
Healthy weight parenting a
Baseline (fall)	3.01 (0.40)	2.96 (0.40)	3.06 (0.39)
Follow-up (spring)	3.03 (0.38)	3.04 (0.40)	3.00 (0.37)
Change b	0.02 (0.36)	0.09 (0.37)	−0.06 (0.35)

Shown is the mean (standard deviation).

^aParental empowerment and parenting scales ranges from 1 to 4, strongly agree to strongly disagree.

^bChange is calculated as the difference in empowerment or parenting score from baseline to follow-up measurement (i.e., fall to spring) in the same academic school year.

^cThis includes 623 unique parents, 578 of which have one year of change data and 45 have two years of data, which results in 668 parent-years.

^dEmpowerment is defined as an increase in parental empowerment score from baseline to follow-up (fall to spring) in a single school year; disempowerment is defined as a decrease in parental empowerment score from baseline to follow-up (fall to spring) in a single school year.

Through a difference-in-difference approach⁸⁰, we assessed the relationship between change in parental empowerment and change in healthy weight parenting practices over an academic school year. Both change scores – that of parental empowerment and healthy weight parenting practices – were modeled as continuous variables. Mixed models with random intercepts were used to account for the nesting of parent-years (n= 668) within Head Start programs (n=16); we did not incorporate a third level to account for time, as only 45 parents (of 623 parents in our sample) contributed data from more than one year.

To account for potential selection bias, we applied inverse probability weights⁸¹, which were calculated as the probability of selection into our analytic sample based on factors identified as differing significantly between our analytic sample and those of all Head Start families eligible for survey completion (i.e., parent age, race/ethnicity, and level of spoken English proficiency).

To account for potential confounding, we also controlled for the number of parents in the household, parent sex, race/ethnicity, employment status, and CHL intervention status, which previous research has linked with both empowerment⁸² and parenting practices⁸³. Although our difference-in-difference approach does account for time-invariant confounding, inclusion of these potential confounders in the model may offer increased precision to our estimates.

As sensitivity analyses, models were rerun only among parents in the programs assigned to CHL’s control arm. This was done to confirm that change in empowerment was associated with parenting behavior change among those who had no exposure to the intervention. We conducted all analyses in SAS software version 9.4 (Cary, NC); we defined statistical significance at the alpha = 0.05 level. Researchers can access the data, data dictionary, code, and output used to produce this manuscript via Open Science Framework.

Missing data.

Of the total 668 parent-years, 636 (95.2%) had no missing data. Of those missing key confounder data (n=32, 4.8%), 16 were missing responses for one item, 11 for two, and five for three to four. With fewer than 5% of our sample missing, we have chosen to run a complete case analysis^84–86. As a sensitivity analysis, we used multiple imputation to account for missing data; after using the MI procedure in SAS 9.4 to impute five datasets, using all variables in the analytic models as predictors, we pooled the results using Rubin’s rule via the MIANALYZE procedure in SAS 9.4⁸⁷ (Model 4).

RESULTS.

Sample characteristics.

Demographic characteristics of Head Start parents included in the analytic sample (n=623) were compared with the remainder of the Head Start population across these two years (n=1922). The analytic subsample more likely to be living with a coparent in the same household, male, and older; they were also more likely to specify a primary language other than English and report low English fluency, compared to the eligible but not included parents. Additionally, NH Black, and Hispanic/Latino parents were less likely to be included in the analytic sample, in comparison to NH White parents.

Most of the sample completed the survey in English (69.1%), while 23.8% of the sample completed the survey in Spanish and 7.1% completed it in Chinese (data not shown). Approximately one third (32.9%) of the parents included in the analysis were between the ages of 30–35 years old (Table 1). Our sample was racially and ethnically diverse, with 40.6% identifying as Hispanic or Latino (any race), 31.3% as NH Black or African American, 14.8% as NH Asian, 10.1% as NH white, and 2.1% other (includes those who identified as biracial/multiracial, as well as Native American, Alaska Native, Native Hawaiian or Pacific Islander). Fewer than one in two parents reported having two parents in the household (45.3%). Nearly two of every three parents (58.6%) reported their highest educational attainment as high school or less. A quarter of parents reported unemployment (24.7%).

Change in healthy weight parenting by change in parental empowerment.

The unadjusted mean (SD) baseline empowerment score across all parents was 3.20 (0.40); Table 2). To compare the magnitude of change among those who did exhibit a change in parenting or empowerment, we tabulated the mean change score among those who increased and those who decreased parental empowerment score. Among those who reported increased parental empowerment score (i.e., reported empowerment; n=324, 48.5%), the mean change was 0.36 (SD=0.30); this same group exhibited an unadjusted mean change in healthy weight parenting of 0.09 (0.37). The mean healthy weight parenting practice score in baseline for this group was 2.96 (0.40). Among those who reported disempowerment (i.e., decreased parental empowerment score from baseline to follow-up (fall to spring); n=270, 40.4%), the mean change in parental empowerment score was −0.35 (0.34); the unadjusted mean change in healthy weight parenting in the same group was −0.06 (SD=0.35).

Table 3 reports results of the primary analyses examining the association of change in empowerment with change in health-related parenting. Three models were run to examine this association: an unadjusted model (Model 1), an unadjusted model with inverse probability weights based on the number of adults in the household, parent race/ethnicity, and level of English proficiency (Model 2), and a model with inverse probability weights (same as Model 2) and adjustment for number of parents in the household, parent sex, race/ethnicity, employment status, and CHL intervention status (Model 3). A fourth model was run, as a sensitivity analysis accounting for missing data, with adjustment for the same variables and imputed values to account for missing data (Model 4). All models produced remarkably similar results, which are presented in Table 3. A one-unit increase in empowerment score was statistically significantly associated with a mean 0.14-unit increase in parenting score (95% CI: 0.090, 0.217; Table 3, Model 3). In all models, this association was found to be statistically significant, though small (partial eta-square=0.03).

Table 3.

Estimated association between empowerment and changes in healthy weight parenting score.

	Estimate (SE)	95% CI	p - value
Model 1 A	0.15 (0.03)	0.09, 0.21	<0.0001***
Model 2 B	0.15 (0.03)	0.09, 0.21	<0.0001***
Model 3 C	0.14 (0.03)	0.08, 0.20	<0.0001***
Model 4 D	0.15 (0.03)	0.09, 0.21	<0.0001***

Stars indicate significance

^*p<0.05

^**p<0.01

^***p<0.001.

^AResults from an unadjusted and unweighted linear mixed effects regression model.

^BApplied inverse probability weights to account for the number of parents in the household, parent age, race/ethnicity, employment status, and spoken English proficiency (based on Head Start intake records).

^CControlled for number of parents in the household, parent sex, race/ethnicity, employment status, and CHL intervention status. Applied inverse probability weights to account for the number of parents in the household, parent age, race/ethnicity, employment status, and spoken English proficiency (based on Head Start intake records).

^DSensitivity analysis: Controlled for number of parents in the household, parent sex, race/ethnicity, employment status, and CHL intervention status and used pooled data from five imputed datasets to address missing data.

Sensitivity analyses.

When the primary analyses were rerun among just parent-years in CHL’s control arm (n=305; compared to the entire analytic sample, n= 668), we observed no meaningful difference in the magnitude or direction of association (data not shown).

DISCUSSION AND CONCLUSION.

Among an ethnically diverse group of parents from low-income households with young children, increases in parental empowerment were associated with increases in parenting practices supporting health weight and growth. Specifically, we observed a one-unit increase in empowerment score to be associated with a statistically significant, though small, mean increase of 0.14 units in healthy weight parenting score.

This small effect size of 0.14 translates to a parent reporting higher agreement with one to two healthy weight parenting practices at follow-up, compared to baseline; for example, a parent may respond with a higher level of agreement to (1) exercising regularly and (2) taking their child places where they can be active. Previous systematic reviews on healthy weight parenting practices suggest that minor changes – such as the modification of one to two practices, as we observed in this study – may be linked with clinically meaningful benefits to child health and behavior outcomes^88,89.

A wealth of literature exists on the relationship between healthy parenting practices in relation to child health and nutrition^90–93. Consequently, the family-focused model to obesity prevention has become the gold standard in childhood obesity prevention^94–100; however, under-engagement of parents remains a frequently cited barrier to intervention success^101–103. Our results underscore the usefulness of empowerment theory as a foundation from which to construct and implement healthy weight promotion efforts in family and early childhood education contexts. Strategies to increase parent empowerment find a natural home at Head Start, which serves to foster parent engagement¹⁰⁴ and synergize meaningful connections between parents and early childhood professionals¹⁰⁵. If, as our results suggest, the process of parental empowerment is predictive of healthier parenting practices, then interventionists in early childhood education settings like Head Start should consider centering parental empowerment within parenting programming efforts. For example, parental empowerment – i.e., a caregiver’s sense of control and ability to act towards healthy lifestyle practices²⁵ – could be promoted as a means of bringing about healthy parenting practices, healthy habit formation^106,107 and reduced metabolic health risk^108,109 in young children and families⁹².

Our observed linkage between parental empowerment and healthy parenting practices adds to a growing body of evidence identifying empowerment as a key driver of family health promotion^110–116. While empowerment is most often studied as an intervention outcome^117–121, we present one of the first quantitative, longitudinal assessments of empowerment as a predictor of behavior among U.S. parents with low income. In line with previous studies¹²², which are largely limited to low- and middle-income nations^{110–112,123}, our findings support empowerment as a predictor of behaviors for the promotion of healthy growth and development in the context of an ethnically diverse low-income U.S. sample.

Despite the prevalent mention of “empowerment” in the health promotion literature^124–127, it continues to lack clear conceptualization^128–130, as do “parenting practices,” which are often incorrectly conflated with “parenting styles”^131,132. This paper offers one of the first clear frameworks for the operationalization and interpretation of parental empowerment²⁵, as well as a clear, working definition of empowerment to contribute to the health promotion literature among a sample of diverse parents from low-income urban households in the U.S.

Our study offers several strengths. First, we evaluated longitudinal data in a large sample of ethnically, racially, and linguistically diverse parents in a major urban area of the Northeastern U.S. Our sample is also representative of the families enrolling in Head Start across our 16 programs, and we included data from approximately 25% of all eligible families. The key demographic and socioeconomic differences between parents who completed two surveys within a year, and those who were eligible but did not, were number of parents in the household, parent age, race, ethnicity, and spoken English proficiency, and employment status (Table 1); those included in the analytic sample were more likely to have two parents in the household, be older, identify as NH White or NH Asian, report low spoken English proficiency (i.e., speak English “not at all” or “not well”, and report being employed). The diversity of our sample also offers important insights for on-the-ground health promotion practice. Our study was one of the first to use a brief empowerment scale with strong psychometric properties²⁵ in the context of healthy child growth and weight; our survey measures could be easily reproduced for use in other community settings¹²⁸. Analytically, our difference-in-difference modelling approach also allowed for the control of unmeasured time-invariant confounding, strengthening the interpretations of our findings.

Several limitations should also be noted. First, there are demographic and socioeconomic differences between those who completed surveys and those who were eligible but did not (Table 1). To address potential selection bias, we applied inverse probability weights (Models 2–4), which accounted for factors linked with probability of survey completion. Additionally, some data are missing for covariates. However, this represented only 4.8% of our sample and, following multiple imputation to account for missing data, the model results did not change.

Our parenting and empowerment measures also rely on self-report, which may introduce measurement error or conscious bias¹³³; however, previous research in this field supports the convergence of data on parental behavior collected via self-report and observational methods¹³⁴. Further, the validity and reliability of the scales in this study have been previously examined, and the items themselves are drawn from validated scales -- such as the Comprehensive Feeding Practices Questionnaire^68–70, the Activity Support Scale for Multiple Groups^71,72, the Sleep Parenting Scale for Infants⁷³, and the Family Empowerment Scale^54–57 -- which has been applied in diverse family contexts.

A second limitation is that most parent empowerment and parenting practices scores were at the high end of the scale; a ceiling effect may have limited the ability to detect changes in these scores, as has been noted in studies using similar measures¹³⁵. In addition, this analysis was conducted in the context of a family-based randomized control trial targeting empowerment as a mechanism of behavior change; although we controlled for intervention status and reran the analyses among the control programs, there could be unaccounted contamination, transmitted from the intervention to control. While the factor structure of the empowerment scale has been confirmed in prior measurement work, the validity of the scale in languages other than English⁵² (i.e., measurement invariance by language) has not been assessed; we have yet to confirm whether the items included in the empowerment scale maintain consistency, both conceptually and statistically, across diverse groups of respondents^136,137. In contrast, the parenting scale has demonstrated measurement invariance across parent sex and language (Spanish and English)⁵².

With both empowerment¹³⁸ and parenting practices^139–142 often being highly culture-specific, future research should focus on the examination of the validity and reliability of these scores in other contexts, samples, and/or languages. Similarly, the role of sex – of both the child and parent – warrants further exploration within the relationships of a parent’s actualization of empowerment¹⁴³ and parenting decisions^{135,144–146}, as well as a child’s receipt^147,148 of such behaviors, as these are often defined by gender-specific norms. Specifically, more focus should be placed on fathers who are strikingly underrepresented^149,150 in both parental empowerment and parenting literature. The operationalization of parental empowerment and parenting practices may be specific to cultural context ^128,151; therefore, future research on the translation of these measures to diverse spaces is necessary.

In conclusion, parental empowerment is predictive of healthy weight parenting in a large sample of ethnically, racially, and linguistically diverse parents with low income. This is the first study to examine the relationship between change in empowerment and healthy weight parenting over time, and further establishes the theoretical underpinnings of empowerment as a target in family-based behavior change work. Given that healthier weight-related parenting practices are linked with healthier lifestyle behaviors (i.e., diet, physical activity) and body weight in children^152–160, results from this study support parent empowerment as a predictor of healthy parenting practices for the prevention of obesity in low-income children^57,59,60,63.

HIGHLIGHTS.

• Increased parental empowerment was linked with healthy weight parenting practices.

• Results establish theoretical underpinnings of empowerment in behavior change.

• Parental empowerment may be an important target in child obesity interventions.