Depressive symptoms and parenting alliance as mediators between food insecurity and child behavior

Abstract The Family Stress Model provides a versatile framework for family research, and while researchers have recently used it to explore constructs among Latinx families, few studies have used it in research about rural Latinx families. The FSM is used to examine the relations between food insecurity, maternal depressive symptoms, parenting alliance, and child behavior. Data from NC 1171, “Rural Families Speak about Health”, are analyzed in a study using path analysis for 99 Latinx participants. Results will show that maternal depressive symptoms were associated with a decrease in parenting alliance. Implications regarding food insecurity, depression, parenting, and child behavior in the context of rurality and culture are discussed.

especially in regard to mental health care (Myers, 2019).In order to rectify the significant knowledge gap, it is crucial to focus on rural minoritized families, in order to understand their experiences and develop effective support systems.
Food insecurity among the rural population is exacerbated by larger distances to food resources (Garasky et al., 2006) and Latinx families in rural communities commonly face additional barriers.The ability for Latinx immigrant families to access food and health resources is often constrained by contextual factors such as limited English language proficiency, and a lack of trained interpreters and culturally competent health-care providers, in addition to the common factors of lower incomes and limited formal education (Cristancho et al., 2008).Additionally, Pelto et al. (2020) documented fears of repercussions such as citizenship denial, having children removed from the home, repayment of the benefit, and that relatives could be deported, for receiving SNAP benefits.Furthermore, studies have revealed values and beliefs regarding food held by Latinx families in the US (Gomel & Zamora, 2007;Greder et al., 2012).Lack of understanding and appreciation of these values and beliefs can contribute to the creation of barriers to accessing food that Latinos believe are healthful.For example, Gomel and Zamora (2007) found in their study of Latino immigrant families that mothers believed canned, prepackaged and processed foods were not healthy.Greder et al. (2012) found Latino immigrant mothers shared this belief.They also valued feeding their children fresh produce and meat, as well as meals prepared from scratch.Thus, lack of understanding and appreciation of these values and beliefs can contribute to the creation of barriers to accessing food that Latinxs believe are healthful.
Stressors, such as food insecurity, that impact the family system may have a negative impact on child behaviors or mental health (Burke et al., 2016;Fiese et al., 2011;Greder et al., 2017;Poole Di Salvo et al., 2016;Slopen et al., 2010).Internalizing and externalizing child behavior problems are associated with household food insecurity (Perez-Escamilla & Vianna, 2012;Whitaker et al., 2006).Parenting stress, defined as the parent's "feelings and perceptions about caring for the child", is a mediator between household food insecurity and problem behaviors in children (Huang et al., 2010).Since the Latinx population experiences food insecurity at a higher rate, and rural Latinxs face unique barriers related to food and health access, more needs to be known about the relationship between food insecurity, depression, parenting, and child behavior among rural Latinxs.Exploration of the relationship between food insecurity, depression, parenting, and child behavior within rural Latinx families could guide the development of interventions that foster resilience and well-being.

Present study
The current study is guided by the Family Stress Model (FSM) (Conger & Conger, 2002) and its purpose is to test a variation of the FSM in an understudied population to determine if the association among food insecurity, depressive symptoms, parenting, and child behavior found in other groups exists for rural Latinx mothers and their children.In this study, household food insecurity is conceptualized as specific facet of financial distress.Parents' emotional state is conceptualized as depressive symptoms in mothers.Instead of directly measuring interparental conflict or withdrawal, the current study examines the quality of co-parenting between primary caregivers from the perspective of mothers.In a similar fashion to the FSM, child behavior is conceptualized as externalizing and internalizing.While controlling for financial distress, positive associations between food insecurity, maternal depressive symptoms and internalizing and externalizing child behaviors are hypothesized.Additionally, it is predicted that the relationship between household food insecurity and child behaviors will be mediated by maternal depressive symptoms and alliance among parents.

Household food insecurity, depression, parenting, and child behavior
Household food insecurity is associated with depressive symptoms among rural mothers (Browder et al., 2013;Greder et al., 2017;Huddleston-Casas et al., 2009;Lent et al., 2009;Nagata et al., 2019) with higher food insecurity associated with higher levels of depressive symptoms.Greder et al. (2017) found differences between older and younger children when looking at rural mothers' depressive symptoms as a mediator between food insecurity and internalizing and externalizing child behaviors.Depressive symptoms partially mediated the relationship between food insecurity and externalizing behaviors among younger children, while it fully mediated the relationship between depressive symptoms and both types of behaviors among older children.Browder et al. (2013) found that rural Latina immigrant mothers whose depressive symptoms were consistently high were more food insecure than those whose depressive symptoms were consistently low.Huddleston-Casas et al. (2009) sought to examine the relationship between household food insecurity and depressive symptoms in rural mothers over time.They found the relationship to be reciprocal, indicating a concurrent causal relationship.Lent et al. (2009) also examined household food insecurity and depression over time in rural mothers.Depressive symptoms were associated with remaining food insecure over time, and there were mediating mechanisms in this relationship.Specifically, depression limited the ability for the depressed individual to work, depression in one family member limited the ability of another family member to work, and child depression made it difficult to find childcare.Olson et al. (2004) found maternal depressive symptoms of rural mothers to be a risk factor for household food insecurity.In an urban sample of Latinx mothers, Nagata et al. (2019) found that food insecurity was related to maternal clinical depression and pervasive developmental problems in children.Gee and Asim (2019) found in a nationally representative sample that parental aggravation mediated the relationship between adult food insecurity and child behavior.In another nationally representative sample, Johnson and Markowitz (2018) found that earlier exposure to food insecurity was more strongly associated with negative social emotional outcomes in kindergarten.In sum, the literature shows that household food insecurity and depressive symptoms are related, and both factors impact children in the household.However, only one study examined rural Latinx mothers, and did not include measures related to child outcomes.Therefore, more research is needed to determine if rural Latinx mothers' depressive symptoms and household food insecurity are related to child outcomes.
Overall, rural children with mental, behavioral, and developmental disorders are more likely to have a parent with lower mental health status and their parents more commonly face financial difficulties that impact food and housing when compared to their urban counterparts (Robinson et al., 2017).Among urban preschool children, household food insecurity was associated with child behavior problems (Whitaker et al., 2006).In their review of how household food insecurity affects child development, Perez-Escamilla and Vianna (2012) found strong evidence of household food insecurity as a strong stressor with potential direct and indirect effects on child behavior.Research about the relationship between food insecurity and child behavior has also focused on potential mediators.Huang et al. (2010) found parenting stress mediated the relationship between food insecurity and child behaviors.However, short-term household food insecurity was not as impactful as family background and parental characteristics on child behaviors.
Other researchers examined how maternal depression impacts child behaviors.Maternal depression may place children at risk for internalizing and externalizing behavioral problems (Coyne & Thompson, 2011;Goodman et al., 2011).Goodman et al. (2011) meta-analysis indicated statistically significant, yet small effect sizes, for relationships between maternal depression and children's internalizing and externalizing behaviors.There were some differences in effect sizes due to gender and child age.Future research is needed to explore combinations of risk factors faced by mothers and children that lead to behavioral problems in children of mothers with depressive symptoms (Goodman et al., 2011).

Parenting alliance and co-parenting
Co-parenting is conceptualized as the process of teamwork and cooperation between two people raising a child or children (Feinberg, 2003) and parenting alliance is the degree to which co-parents support each other in the parenting role (Abidin, 1992).The interaction effect of parenting alliance and positive parent-child activities was related to internalizing child behaviors in a sample of families of children referred to a community mental health center (Gullan et al., 2014).Analysis of data from the Fragile Families and Child Wellbeing study found that union instability and externalizing child behaviors were mediated by co-parenting support (Karberg & Cabrera, 2020).In a sample of Italian families, coparenting mediated marital dissatisfaction and children's behavior (Camisasca et al., 2019).Amato et al. (2011) found in a cluster analysis that children with parents who had cooperative co-parents had fewer behavioral problems when compared to the other types of parenting post divorce.Similarly, in a sample of divorced Portuguese parents, children of parents who were in the identified profile of cooperative coparenting had lower levels of internalizing and externalizing behaviors (Lamela et al., 2016).Researchers have also discovered gender moderated the association between coparenting quality and behavioral outcomes among 2-year-old children of African American and Latinx adolescent mothers, with lower levels of externalizing and internalizing behaviors in girls and higher levels of externalizing behaviors and no difference in internalizing behaviors in boys associated with coparenting that is higher in quality (Mack & Gee, 2018).

Sample
Data for this HSIRB approved study are comprised of a subset of participants from in-person computer assisted interviews conducted between 2010 and 2013 with 444 mothers who lived in a rural community in one of the 13 states (California, Hawaii, Illinois, Iowa, Kentucky, Massachusetts, Nebraska, New Hampshire, North Carolina, South Dakota, Tennessee, Texas, and Washington) that participated in Rural Families Speak about Health (RFSH), a project designed to examine how rural low-income families' physical and mental health are impacted by multiple contextual factors (Mammen & Sano, 2014).For the current study, mothers were 18 years of age or older, had at least one child 18 months to 12 years of age, lived in households with incomes at or below 185% of the federal poverty level, and identified as Latino/Latina.If the mother had more than one child, a focal child was randomly selected, using a random number table.Mothers responded to all child-related measures based on the focal child.Nearly all of the mothers lived in rural counties that were designated as having an urban influence code (UIC) ranging from 6 to 10.The urban influence code classifies US counties into categories according to population size, urbanization, and access to larger communities.Higher UIC values indicate greater rurality (i.e.counties with less urban influence) with values ranging from 1 to 12 (Economic Research Service ERS, USDA, 2007).
The current study analyzed data from Wave 1 from RFSH and includes 99 Latina mothers who resided in California, Hawaii, Iowa, Massachusetts, North Carolina, Nebraska, Tennessee or Washington.Mothers whose randomly selected focal child was under 18 months of age (n = 10) were excluded from the present study because the measure that assessed child behaviors used in the study (i.e.Child Behavior Checklist, Achenbach & Rescorla, 2000) contained items relevant to children 18 months of age or older.An a-priori sample size calculation indicated that this sample size was sufficient for analyses with four predictors, statistical power of 0.8, and anticipated effect sizes (f 2 ) of 0.29.

Recruitment
Mixed Purposive Sampling (MPS) (Mammen & Sano, 2012), a nonprobability sampling method that combines strengths of both purposive sampling and chain-referral sampling, was employed to recruit participants for RFSH.In other words, initial participants were purposively sampled and were asked to refer others who met the inclusion criteria.Once recruited, participants were interviewed in person in their homes or in a private conference room in a public building.The research team in each state identified a person to serve as the interviewer.Interviewers were either graduate students, faculty members, University Cooperative Extension specialists, or a family-serving professional who received training in interview techniques.The interviewer read the questions and typed participants' responses onto a computer template.The interview protocol included demographic questions and questions from standardized instruments pertaining to the physical and mental health of mothers and children.In each household, a focal child was randomly selected and mothers responded to questions about children with this child's information.Participants were offered gift cards after completing the interview process.Based on the preference of the mothers, interviews were conducted in English or Spanish.

Household food insecurity (predictor variable)
Household food insecurity was measured using the Six-Item Short Form of the USDA Household Food Security Module and the associated Six-Item Food Security Scale (Bickel et al., 2000).Household food security scores were based on the number of affirmative response questions such as: 1) "In the last 12 months, the food that (I/we) bought just didn't last, and (I/we) didn't have money to get more."Response options included often true, sometimes true, never true, don't know or refused, with often true and sometimes true considered affirmative responses; and 2) "In the last 12 months, did you or other adults in your household ever cut the size of your meals or skip meals because there wasn't enough money for food?" with options including yes, no, or don't know.Affirmative responses were coded with a score of one, with higher scores indicating higher levels of household food insecurity.The alpha for the sample was .75(n = 99).Overall food security scores were created by summing the score for each question.Food security categories were created based on the overall score.The range for the overall score is 0-6, and the range for the categories is 1-3.

Maternal depression (mediating variable)
Symptoms for depression among mothers were measured using a shortened form (Andersen et al., 1994) of the Center for Epidemiological Studies Depression Scale (CES-D; Radloff, 1977).Mothers reported how often they felt or behaved in the last 2 weeks in response to 10 statements such as I had trouble keeping my mind on what I was doing and I felt that everything I did was an effort.Response options included: rarely or none of the time (less than 1 day = 0), some or a little of the time (1-2 days = 1), occasionally or a moderate amount of time (3-4 days = 2), and all of the time (5-7 days = 3).Responses were summed and total scores ranged between 0 and 30.A score of 10 or higher indicates clinically significant depressive symptomology.The alpha for this sample is .74(n = 98).

Parenting alliance (mediating variable)
The Parenting Alliance Measure (PAM) measures the strength of alliance between primary caregivers in raising their children (Abidin & Konold, 1999).Mothers responded to 20 statements using a Likert scale ranging from 1 to 5, with 5 indicating the highest level of agreement.Examples of statements include talking to the other primary caregiver about our child is something I look forward to; the other primary caregiver pays a great deal of attention to the child; the other primary caregiver and I agree on what our child should and should not be permitted to do; I feel close to the other primary caregiver when I see him or her play with the child.The alpha for this sample is .91(n = 81).Scores were created by summing the score from each answer.

Child behavior (outcome variable)
The Achenbach System of Empirically Based Assessment (ASEBA) contains two scales designed to evaluate children's internalizing and externalizing behaviors: the Child Behavior Checklist 18 months−5 years (CBCL 1 ½−5) and the Child Behavior Checklist 6-18 years (CBCL 6-18) (Achenbach & Rescorla, 2000).The checklists contain questions pertaining to the child's emotions and behavior in the past 6 months.Mothers were asked to think about the behavior of a randomly selected focal child and report whether each statement was "0 = not true," "1 = somewhat or sometimes true," or "2 = very true or often true".The internalizing subscale provides a rating for the extent to which the child has exhibited symptoms of anxiety, depression, or withdrawal.The externalizing subscale provides a rating of the extent to which the child has exhibited symptoms of aggression, hyperactivity, or noncompliance.Because this measure is a checklist, the reliability measure of alpha is not applicable.The score for 6-18-year-old checklist is the sum of the 113 items, and the range is from 0 to 226.The score for 18 months−5-year-old checklist is the sum of 100 items, and the range is from 0 to 200.

Financial distress (control variable)
A modified version of the Personal Financial Wellness Scale ™ (PFW ™ Scale) (Prawitz et al., 2006); was used to measure mothers' perceptions of their financial situation.. Mothers responded to seven questions (i.e.How often do you worry about your financial situation?How often do you find yourself just getting by financially and living paycheck to paycheck?) using a Likert scale (1 = Never, 2 = Occasionally, 3 = Sometimes, 4 = Frequently, 5 = Very frequently).A higher score indicates more financial distress.The alpha for this sample is .73(n = 98).After two items were reversed scored, so that higher scores indicated more distress, the items were summed to create an overall financial distress score.The range of possible scores is 5-35.

Analysis
Descriptive statistics and correlations of all study variables and demographics were calculated using SPSS.The conceptual model examined relations between household food insecurity, mothers' depressive symptoms, parenting alliance, and internalizing and externalizing child behaviors while controlling for financial distress.The model detailed in Figure 1 was analyzed in Mplus as a two-group model using bias controlled bootstrap iterations and maximum likelihood estimation.To test for indirect effects, 1,000 bias corrected bootstrap iterations were performed.Bootstrapping procedure guards against a non-normal distribution of the mediating effect (Shrout & Bolger, 2002) and the risk of spuriousness.The groups are based on the age ranges defined in the Child Behavior Checklist.The younger group is comprised of children aged 18 months to 71 months, while the older group is comprised of children 72 months to 155 months.Financial distress was controlled for by entering it into the model first.To test for the mediating effect of depressive symptoms and parenting alliance, indirect effects were tested, as well as 95% confidence intervals for the specific indirect effects.

Results
Demographics are displayed in Tables 1 and 2. The demographics for the sample as a whole (Table 1) and the sample divided by child age groups are provided (Table 2).
Descriptive statistics are displayed in Tables 3 and 4. The descriptive statistics for study variables are provided for the whole sample and the sample divided by child age.For the whole sample, 29.3% of the households were identified as food insecure (n = 29).Seventeen percent of mothers were at or above the cut-score (10) for depressive symptomatology (n = 17).For the whole sample, the average parenting alliance score was 91.71 (SD = 10.85).For the whole sample, the average financial distress score was 24.70 (SD = 3.57).For internalizing behaviors, the mean of the sum scores for younger children was 7.99 (SD = 6.71) and for older children 7.67 (SD = 5.36).For externalizing behaviors, the mean of the sum scores for younger children was 11.60 (SD = 8.99) and for older children 8.10 (SD = 7.58).
Correlations of study variables are found in Tables 5 and 6.Correlations for the study sample as a whole, as well as divided by child age, are provided.For the sample as a whole, internalizing and externalizing behaviors of children in the younger group as well as the older group were significantly correlated (r = .84,p < .01;r = .37,p < .05).Additionally, depressive symptoms were significantly correlated with parenting alliance and financial distress (r = −.32,p < .01;r = .27,p < .01).As  The model fit the data in an acceptable manner.The χ 2 (8) = 10.80,p < .21.The comparative fit index (CFI) was .95, the Tucker Lewis Index (TLI) was .82,and the root mean square error of approximation (RMSEA) was .08 with a 90% CI [.000, .20].The model results are detailed in Table 7.While none of the parameter estimates were significant for the younger age group, the association between maternal depressive symptoms and parenting alliance trended toward significance (β = −.27,p = .12).For the older age group, depressive symptoms were significantly related to parenting alliance (β = −.41,p = .007).Additionally, among the older child group, the association between household food insecurity and parenting alliance (β = .20,p = .12)and the association between household food insecurity and externalizing behavior (β = .33,p = .15)trended toward significance.Due to only one significant result, the change in the chi-square value between models with individual pathways constrained and a baseline model in which all pathways were unconstrained was not examined to determine if any paths varied significantly between younger and older children.

Discussion
Rural Latinx mothers commonly experience many stressors, including those of household food insecurity and depressive symptoms that can affect child behavior (Coyne & Thompson, 2011;Goodman et al., 2011;Perez-Escamilla & Vianna, 2012;Whitaker et al., 2006).The degree of alliance between primary caregivers is important to examine when looking at child behavior outcomes.The Family Stress Model (Conger & Conger, 2002) is an appropriate framework to apply to this study because the economic stressor of household food insecurity is examined while controlling for financial distress.Maternal depressive symptoms and parenting alliance were examined as potential mediators between household food insecurity and child behavior.
The purpose of this study was to determine if maternal depressive symptoms and parenting alliance served as mediators in the relationship between household food insecurity and child internalizing and externalizing behaviors among rural Latinx households.While these two variables did not serve as mediators in the relationship, an increase in maternal depressive symptoms was significantly related to a decrease in parenting alliance among mothers who had older children (6-12 years old) while controlling for financial distress.Additionally, among mothers who had older children, there were two associations that trended toward significance.First, the association between household food insecurity and parenting alliance trended toward significance, showing the possibility that an increase in household food insecurity was associated with an increase in parenting alliance while controlling for financial distress.While the direction of this association may seem contrary to expectations, it is possible that mothers experiencing increased food insecurity form a stronger alliance with their coparent/caregiver to protect against the stress of household food insecurity.It may be that when facing the stressor of household food insecurity, levels of cohesion increase between the primary caregivers, creating a stronger alliance (Olson, 2000).The second association with a trend toward significance was between household food insecurity and child externalizing behavior.This possible association is consistent with the literature (Perez-Escamilla & Vianna, 2012;Slopen et al., 2010;Whitaker et al., 2006).While this direct relation is not explained by mechanisms related to family processes, it may be more physiological in nature, with the physical consequences of household food insecurity directly increasing externalizing behaviors among children.While none of the associations for the group of  younger children were significant, one did trend toward significance.Similar to the significant association for the group of older children, there was a trend that showed the possibility that an increase in maternal depressive symptoms was related to a decrease in parenting alliance.Maternal depressive symptoms are a stressor that reduces cooperation between parents and this is reflected in lower levels of parenting alliance.
While acknowledging that trends toward significance are indeed not significant, it is interesting to note the possibility that household food insecurity and depressive symptoms are stressors that affect parenting alliance in opposite ways.While commonly one would expect stressors to have a negative association with parenting alliance, in this study, only depressive symptoms had this association.

Limitations and strengths
While this study contributes to the knowledge about the experiences of rural Latinx mothers and their children, there are limitations as well.In order to gain access to a population containing potentially undocumented people, a nonprobability sample was used.Although this makes the study population non-representative of the rural Latinx population as a whole and limits generalizability, it allows researchers to learn more about a population that is difficult to access.Next, the study design is based on self-reported data from mothers, which introduces bias due to social desirability as well as common-method bias.Additionally, each construct is measured only by one scale, which reduces convergent validity.Another limitation related to measures is that inconsistent or harsh parenting was not measured, which is a part of the original FSM.The data are cross sectional and no causal or predictive claims can be suggested.Finally, the sample size was small, which increases the likelihood of Type II error, that is, there may have been more significant results, but the power to detect significant associations was limited by the sample size.Future studies should consider a random sampling strategy, the inclusion of viewpoints of other family members, the inclusion of observational data, measurement of constructs in multiple ways, longitudinal design, and an increased sample size.However, despite these limitations, the study contributes to the literature by including parenting alliance in an application of the FSM and examining a population of rural Latinx families.This study has strengths.A seldom studied population is examined within the framework of the well-known and extensively tested Family Stress Model (Conger & Conger, 2002), using measures that are reliable and valid, and financial distress is controlled for in an attempt to isolate household food insecurity as a stressor separate, but related to, financial distress.

Conclusions
In this study, it was clear that an increase in depressive symptoms was associated with a decrease in parenting alliance among rural Latinx mothers who had children six to 12 years of age.Professionals such as couple and family therapists and family life educators can work together to assess the stressors faced by families of rural Latinx mothers, with a particular focus on depressive symptoms and on how well primary caregivers operate as a team.These can be measured easily during an office visit with a mental health professional.
This study did not find associations between food insecurity and depressive symptoms or depressive symptoms and child behaviors for either age group.Therefore, it is important for future studies to determine if there is a protective mechanism at work among rural Latinx families that buffers them from the associations found in other populations.As with all families, it is important that rural Latinx families have access to resources that can positively benefit them and their families, so they can interact in an optimal manner.In order for primary caregivers to have a strong alliance, the opportunity to gain or maintain positive mental health is important.
Positive mental health could be promoted, and depressive symptoms decreased, through increased access to mental health resources that are culturally relevant and professionals who are prepared to provide services to a predominantly Spanish-speaking population.Another potential mode of accessing resources to promote positive mental health is telehealth, a method to deliver mental health services for people in rural areas (Brownlee et al., 2010).Telehealth has been shown to reduce depressive symptoms for rural, Mandarin-speaking immigrants who received treatment from a Mandarin-speaking therapist (Zheng & Gray, 2014).To increase opportunities for positive impact on mental health, it is important for these strategies to be enacted simultaneously and be specifically tailored to specific population groups of Latinx immigrants.Increased access and use of mental health services may decrease depressive symptoms and thereby increase alliance between primary caregivers as they care for their children.
Future research should further examine the role parenting alliance plays in applications of the FSM among rural Latinx mothers.Parenting alliance may be affected negatively or positively depending on the type of stressor.Household stressors such as food insecurity may operate differently than depressive symptoms in their impact on parenting alliance.Findings from this study suggested that the association between household food insecurity and parenting alliance should be further explored, including examining potential mechanisms that mediate the relationship.

Table 1 . Demographics for the total sample n = 99
detailed in Table6, for children in the older group, parenting alliance was negatively correlated with depressive symptoms (r = −.34,p < .05).Household food insecurity was negatively correlated with externalizing behaviors for this group as well (r = −.31,p < .05).For children in the younger group, depressive symptoms were negatively correlated with parenting alliance (r = −.30,p < .05).

Table 6 . Pearson correlations of study variables for older group of children (above the diag- onal) and younger group of children (below the diagonal)
*Correlation is significant at the 0.05 level (2-tailed).