Skip to main content

Effects of a Brief Parenting Intervention In Shelters For Mothers And Their Children Experiencing Homelessness


This study was designed to examine the effectiveness of an evidence-based parenting program—Triple P Positive Parenting Program—in shelter settings for families experiencing homelessness. The intervention has not previously been evaluated in a shelter setting, where there is a critical need for evidence-based parenting programs. Using a within-group pre- and post-intervention with 3-month follow-up design, 39 mothers residing in a shelter with a child ages 2–6 years participated. Results of this preliminary study showed positive effects of Triple P Discussion Groups. There were significant improvements in mother-reported parenting practices and child behavior across time, but no change in child maltreatment risk as measured by the Brief Child Abuse Potential Inventory. Mothers rated satisfaction with the program high immediately after the group and again three weeks later. Results showed Triple P Discussion Groups are acceptable and have some positive effects for this vulnerable population in need of parenting support. We discuss implications of findings, limitations of the study (including a 33% attrition rate), and recommendations for further study.


  • Evaluated Triple P Discussion Groups delivered in shelters for homeless families.

  • Parenting improved; child problem behaviors reduced; no effects on abuse risk.

  • Mothers were highly satisfied with all aspects of the intervention.

During the 2017–2018 school year, more than 1.5 million children enrolled in public schools in the U.S. experienced a period of homelessness (US Department of Education 2020). This is a gross underestimate of the number of children who experience homelessness annually, as it excludes young children not yet in school and youth who have dropped out. Children who experience homelessness tend to also face other adverse life circumstances. Notably, there is a strong link between homelessness and child welfare involvement (Keeshin & Campbell, 2011; Park et al. 2004). Perlman and Fantuzzo (2010) found that 40% of children residing in emergency housing had experienced child maltreatment. Reducing risk of maltreatment for children experiencing homelessness should be a central goal for mental health and social services professionals. We aimed to accomplish two novel tasks: (1) to test the effects of Triple P Discussion Groups for families experiencing homelessness, and (2) to describe the acceptability of Discussion Groups for mothers residing in shelters.

Characteristics of Families Experiencing Homelessness

Children experiencing homelessness vary widely in their social, emotional, and behavioral functioning, and there is evidence of resilience among these children (e.g., Masten et al. 2014). They are, however, at elevated risk for a host of challenges in early childhood related to the experience of homelessness and the broader impact of poverty. In terms of mental health, children experiencing homelessness are at a greater risk than housed children in poverty of developing a behavior disorder (Bassuk et al. 2015; Yu et al. 2008). To illustrate, Lee et al. (2010) found young children residing in transitional housing had greater parent- and teacher-reported internalizing and externalizing problems than a matched sample identified by their teachers as exhibiting significant aggressive behavior. Children in transitional housing were rated a full standard deviation above the normative average on the measure of externalizing problems and 43% of those children were receiving special education services under the category of severe emotional disturbance. There is also evidence that some children demonstrate positive adjustment despite their experiences of homelessness (e.g., Obradovic, 2010), suggesting that protective factors may be buffering against negative outcomes. One widely studied protective factor for child who have experienced various forms of early adversity, including homelessness, is warm and responsive caregiving (Herbers et al. 2014; Smith, et al. 2015). However, providing responsive parenting while experiencing homelessness is challenging for a variety of reasons.

Many parents experiencing homelessness have faced risks that make them prone to engage in harsh parenting practices. Cutuli and Herbers (2014) outlined several individual-level risk factors associated with abusive parenting that sometimes characterize parents experiencing homelessness, including having less than a high school education, low levels of social support, extreme poverty, mental illness, their own experiences of child maltreatment, and recent experiences of domestic violence. Each of these characteristics, individually, predict harsh parenting and negative child outcomes. For example, parents experiencing depression are generally less responsive to their child’s needs; therefore, infants whose parents are depressed are more likely to develop insecure attachment with their mothers and internalizing behavior problems (Lim et al. 2011; Pelaez et al. 2008). Parents who are homeless and residing in shelters also experience stress related to their living conditions. They are “parenting in public” (Friedman, 2000) in chaotic and crowded conditions; they may face family separation because older boys are often not allowed to reside in shelters with their family; and parents must adhere to strict shelter guidelines about meal and bedtime, child supervision and disciplinary strategies that likely deviate from their typical patterns (Perlman et al., 2014). These cumulative stressors result in high risk of engaging in negative parenting practices and difficult parent–child relationships.

Despite the well-documented risk for maltreatment of children experiencing homelessness, there have been very few published evaluations of parenting programs in shelter settings. A review of literature on parenting programs in shelters revealed only 12 studies (Haskett et al. 2014). Very few of those studies included a control group, follow up assessments, or measures of treatment fidelity. Most of the interventions were not evidence-based and if they were, significant modifications were made for delivery of the intervention in shelters. On a positive note, feasibility and parents’ acceptability of the programs were promising (Haskett, et al. 2014).

Triple P—Positive Parenting Program

Triple P is a parenting program based on behavioral principles and social learning theory. It was designed to prevent or reduce social, behavioral and emotional problems of children by building the knowledge, skills, and self-confidence of parents (Sanders, 1999). Triple P is provided in a tiered service model ranging from prevention services (Level 1) to intensive parenting interventions for high-risk parents (Levels 4 and 5). Level 3 Discussion Groups are considered a low-moderate intensity parenting intervention. There are four separate groups designed to deliver training with a focus on improving parent–child interactions and applying positive parenting skills to specific problem situations (e.g., disobedience, fighting & aggression).

Decades of research point to positive effects of Triple P on multiple outcomes, across diverse populations (e.g., Abate et al., 2020; Sanders, 1999). According to a meta-analysis, Triple P leads to moderate positive effects (all Levels combined) on measures of parent (d = 0.58) and child (d = 0.51) behavioral observations, parenting practices (d = 0.58) and child social, emotional and behavioral problems (d = 0.47) (Sanders, et al. 2014). Prinz et al. (2009) conducted the first population-level evaluation of the suite of Triple P services in the U.S. They randomly assigned 18 counties in SC to deliver Triple P or services-as-usual and found significantly less growth of substantiated child maltreatment cases (d = 1.09), fewer out of home placements (d = 1.22) and fewer injuries related to child maltreatment (d = 1.14) in counties with Triple P available. A more recent evaluation of population-level implementation of Triple in 34 of 100 NC counties was conducted, and results indicated small but meaningful effects on child maltreatment indices (Schilling et al. 2020). The aim of our study was to determine if Triple P Discussion Groups were effective at decreasing the risk of maltreatment specifically for children in families experiencing homelessness. Following is a brief review of the effects of Triple P Discussion groups.

Effects of Discussion Groups

Several controlled studies on effects of Discussion Groups yielded moderate to strong effect sizes for parenting and child behavior that were maintained at follow up (Joachim et al. 2010; Morawska et al. 2011; Tully & Hunt, 2017). Interestingly, Dittman, Farruggia, Kepown and Sanders (2016) also found significant effects of Discussion Groups on child problem behaviors and parenting practices, despite not using the follow-up calls to individual parents that are included in standard delivery of Discussion Groups. In contrast to these positive results, findings were disappointing in a study of Discussion Groups for parents of adolescents, with no effects for parenting or parent-youth relationship quality (Dittman et al. 2020). Most prior studies have examined the effects of attendance at a single Discussion Group (most often on the topic of disobedience), but Palmer et al. (Palmer et al. 2019) compared effects of attendance at a single group with effects of attending all four Discussion Group topics. Results showed added benefits of the “sufficient exemplar” training. In sum, literature suggests that Discussion Groups can help improve positive parenting practices and decrease child behavior problems, especially for parents of younger children. However, we note that many of these studies were conducted in Australia or New Zealand with predominantly middle class, college educated parents.

Mejia, Calam, and Sanders (2015) evaluated effects of Discussion Groups with low-income parents in Panama who rated their children above the mean score on the ECBI Intensity scale. Investigators found moderate to strong treatment effects in terms of child problem behaviors and parenting behavior 2 weeks after intervention; at 3 and 6-month follow ups, even stronger effects were found. This study provides some evidence that Discussion Groups can be effective, without modifications, for low-income parents experiencing above-average intensity of child problem behaviors. To date, no studies have been conducted on the use of Discussion Groups for families experiencing homelessness.

Triple P in Shelters

Two publications describe delivery of Triple P in a shelter or shelter-like setting. Glazemakers and Deboutte (2013) investigated the effects of Level 4 groups for 10 parents with intellectual deficits in a residential program for parents who had experienced domestic violence. This non-controlled, small study conducted in Belgium suggested Triple P Level 4 was feasible in the shelter setting, but the authors found inconsistent outcomes across measures of child social-emotional adjustment, parental mental health, and parenting behavior. In the second study, Wessels and Ward (2016) measured acceptability of Triple P among women residing in shelters for women impacted by domestic violence in South Africa. Mothers rated common Triple P parenting strategies on acceptability, usefulness, likelihood of use, and current use. Results showed that mothers found Triple P acceptable but that time constraints and living in a shelter were considered likely barriers to implementing new parenting strategies.

For several reasons, Triple P is a recommended intervention for parents residing in shelters (Cutuli & Herbers, 2014; Haskett et al. 2014a; Haskett et al. 2014b). In addition to the strong evidence base in support of Triple P, it is adaptable across contexts and populations. Discussion Groups have the potential to provide social support, which is often limited for parents experiencing homelessness (Zugazaga, 2008). The groups allow for individualization through take-home workbooks and follow-up phone calls to individual parents. Groups are only 2 h in length, which is important given the highly transient nature of the population. Holtrop et al. (2015) interviewed 40 parents living in transitional housing to determine the topics, approaches, and considerations they desired in parenting supports. The most frequently requested topic was methods to improve child compliance. There are four Discussion Groups, each covering a different topic; the content in the Dealing with Disobedience group most closely aligns with improving child compliance. In terms of approach considerations, Holtrop et al. (2015) found that parents most wanted to discuss how past experiences affected parenting and how to parent in the context of transitional housing. There is nothing specifically built into Discussion Groups that addresses past experiences or parenting in transitional housing, but there is ample time for parents to discuss these issues during the group. Regarding preferred activities during groups, parents most wanted to learn from other parents and to watch videos. Discussion Groups heavily incorporate both activities. For all these reasons, Triple P Discussion Groups seem particularly appropriate, and potentially effective, for parents residing in shelters for families experiencing homelessness.

Current Study

The need for parenting interventions in shelters is apparent based on literature reviewed herein, but investigations of effects of evidence-based programs in shelters is lacking. We aimed to build the evidence base on promoting positive parenting and reducing risk for maltreatment among vulnerable families without homes by investigating effects of Triple P Discussion Groups. We expected that parents would report significantly less intense child behavior problems, less negative parenting practices, and lower risk of maltreatment at post-treatment than pre-treatment. Treatment effects were expected to be maintained over time. Whether the parents were satisfied with the intervention immediately following the group and three weeks after the group were exploratory research questions.



Participants were 39 mothers recruited from a single shelter. The shelter predominantly served low-income, single African American mothers with children under the age of 8 years and had no special eligibility requirements for admission other than homelessness. To be eligible to participate in the study, mothers had to have a child between the ages of 2–6 years and speak English. Participants were predominantly African American (84.6%), had an average age of 29.31 years, had never been married (79.5%), and were not employed at the time (61.5%). About 33% completed less than high school, 30% completed high school or GED equivalent, and 33% completed some college courses. Most reported having one (38.5%) or two (23.1%) children residing with them in the shelter. Parents with multiple children completed measures on their child between the ages of 2–6 years with the most significant behavioral issues. The majority of target children (69%) were female and the average age was 3.67 years at pre-intervention.

Eight mothers did not attend the intervention group, making them ineligible to complete the remaining time points of data collection. We could not locate three participants for post-intervention data collection, and an additional two participants who completed pre- and post-intervention assessments were unable to be located to complete follow-up assessments. Thus, 26 participants completed all study components, for an attrition rate of 33.33% which is consistent with attrition in other studies of Discussion Groups (Mejia, Calam & Sanders, 2015). An a priori power analysis was conducted with the effect size set to 0.45 and power set to 0.80; 21 participants were needed, so our sample size of 26 was adequate to detect effects.


Four Discussion Groups were held across 8 months with 14, 9, 7, and 9 mothers in the groups. In coordination with shelter staff, the first author recruited participants face-to-face in common areas of the shelters and via posters. Mothers were asked if they wanted to attend a Discussion Group about parenting in about 1 week. They were told about the free dinner, childcare, and monetary incentives for completing assessments. Mothers were incentivized to participate via payment following completion of the measures. They were paid $20 for completing the pre-intervention (T1) measures, $30 post-intervention (T2) measures, and $40 for follow-up (T3) measures. T1 data were collected the week before the group took place. T2 data collection occurred approximately three weeks after the group. This short-term, post intervention follow up is standard in evaluations of Discussion Groups that have found promising effects (Chung, et al,, 2015; Dittman et al. 2016; Joachim et al. 2010; Mejia et al. 2015; Morawska et al. 2014, 2011). T3 data collection took place 3 months after the group. To reduce the likelihood of attrition due to the transient nature of this population, the follow up period was shorter than other evaluations of Discussion Groups. We collected data in person and provided an overview of the consent form at each time point before completing the measures. Procedures were approved by the university IRB.


Level 3 Discussion Group is a 2 h skills training intervention. Groups cover one of four topics and providers can select the one(s) most appropriate for their participants: dealing with disobedience, managing fighting and aggression, developing good bedtime routines, and hassle-free shopping for children. For the current study, only Dealing with Disobedience was delivered. At the beginning of the group, parents were given a workbook to aide their participation and subsequent implementation of the strategies with their children. The structure of the group included a standardized power point presentation with embedded example videos and allowed ample time for discussion of the strategies presented. Over the 2 weeks following the intervention, the provider attempted to complete two phone calls with each mother to discuss any barriers they were experiencing in using the strategies and to provide praise and corrective feedback as needed. There is a semi-structured format used for these phone calls. Notably, fewer than 25% of mothers participated in either of the two phone calls during this time period. Therefore, this discussion often took place in person prior to the post-intervention measures (T2) being completed.

Triple P is a manualized intervention that requires providers to complete training and pass an accreditation exam. The first author was an accredited provider and conducted Discussion Groups for this study. Groups were held at the shelter in the evenings, and dinner and childcare were provided to reduce barriers to participation. Standard Triple P adherence checklists were used to complete a fidelity check on implementation. Such checklists are not provided for phone call follow-ups, but a checklist developed by the researcher was used to guide the calls.


Parenting practices

The Parenting Scale (Arnold et al. 1993; Rhoades & O’Leary, 2007) was used to assess self-reported parenting behavior at every time point. The 30-item questionnaire requires parents to rate their parenting style on a 7-point Likert scale in relation to two possible responses to various parenting scenarios. For example, parents are given the scenario “when my child misbehaves…” then parents choose between “I usually get into a long argument with my child” (anchored at 1) or “I don’t get into an argument” (anchored at 7). Three mean subscale scores (Laxness, Over reactivity and Hostility) are generated and a Total score is generated by summing all items. Higher scores indicate more dysfunctional parenting practices. The Total score has strong psychometric properties and can adequately discriminate between clinical and non-clinical populations (Arnold et al. 1993). The PS has been widely utilized in the evaluation of Triple P and is appropriate for use with high-risk populations (Steele et al. 2005). For this study, internal consistency of the Total score was α = 0.67 at T1, α = 0.80 at T2, and α = 0.90 at T3.

Child problem behaviors

The Eyberg Child Behavior Inventory (ECBI; Eyberg & Ross, 1978) is a 36-item parent-report measure of behavior problems of children ages 2–16 years. The ECBI was used to evaluate child problem behaviors at every time point. For each item, parents were asked to rate how often the child exhibited that behavior on a Likert scale of 1–7 (1 = never; 7 = always) (Intensity scale) and if they considered that behavior to be a problem by circling “yes” or “no” (Problems scale). Raw scores from the Intensity scale were used as an outcome measure, similar to other studies of Discussion Groups (e.g., Chung et al. 2015; Dittman et al. 2016; Mejia et al. 2015; Morawska et al. 2014). The ECBI is psychometrically sound and internal consistency of the Intensity score for this study was excellent at T1 (α = 0.96), T2 (α = 0.95) and T3 (α = 0.94).

Child abuse risk

Parents completed the 34-item Brief Child Abuse Potential Inventory (BCAP; Ondersma et al. 2005) to assess maltreatment risk at every time point. Parents answered each item by circling if they agreed or disagreed with the statement. The 25-item Risk total score, a primary outcome variable in this study, was summed according to Ondersma et al. (2005), with a higher score indicating greater potential for abuse. Previous studies have found a conservative clinical cut off score of 12 on the Risk scale resulted in 0.91 sensitivity and 0.93 specificity (Ondesrma et al. 2005). The BCAP Risk scale has strong internal consistency (0.89), significantly overlaps with the full CAP Abuse Risk score (r = 0.96) and is associated with measures of parenting behavior, depression, substance abuse, anti-social behavior, and other risks for maltreatment (Kelley, et al. 2015; Ondersma et al. 2005; Walker & Davies, 2010). For this study, the internal consistency of the Risk score was acceptable for T1 (K-R 20 = 0.79), T2 (K-R 20 = 0.79) and T3 (K-R 20 = 0.84). Reliability and utility of the Lie and Random Responding subscales with high-risk samples has been less than adequate in past studies (Walker & Davies, 2012; 2010) and was therefore not used in the current investigation.

Client satisfaction

Triple P provides two measures of parents’ satisfaction with services. First, parents completed a 10-item Discussion Group Satisfaction Questionnaire (DG-SQ) immediately following the group. Parents use a 7-point Likert scale to provide feedback about how helpful they believe the information will be and how well the provider facilitated the group. The Client Satisfaction Questionnaire (CSQ) was used to assess satisfaction with the Discussion Group at T2. It included 12, 7-point Likert scale items and one free response item. Likert scale item scores are summed to yield a Total satisfaction score. The CSQ is regularly administered by Triple P providers following the completion of service, although there is minimal support for psychometric properties (Sanders, et al. 2014). For this study, the internal consistency was strong for the DG-SQ Total score (α = 0.89) and CSQ Total score (α = 0.98).


Participants reported their marital status, age, highest level of educational attainment, race, sex, current residence, number of children in their care, current and previous six months of mental health services received for themselves or their target child, target child sex and age on a demographic form at each time point. At T2 and T3 parents were asked, “Have any major life changes occurred since the last time we met?” If they answered yes, they were asked to describe what change(s) occurred. These questions were asked to understand potential threats to the internal validity of the study. At T2 about 40% of parents reported a major life change had occurred since last seeing the examiner. Some major changes reported included having a child removed and returned to them via CPS involvement, job or childcare changes, and death in the family. Notably, no two participants reported the same major life change at T2. At T3 about 86% of participants reported experiencing a major life change. At that point, the most frequently reported changes were leaving the shelter (48%) and childcare changes (30%).


Preliminary Analyses

Correlations among all the dependent variables were computed. No correlation coefficients exceeded 0.80 (all were below 0.40) which would have warranted a separate ANOVA for the correlated dependent variables due to concerns about multicollinearity. To determine if participants who dropped out of the study were significantly different from participants who completed the study, t-tests or chi-square analyses were completed. No significant mean differences were found between participants who completed all three time points and those who did not on demographic characteristics or on two out of three of the primary outcome variables at pre-assessment (ECBI and BCAP). There was a significant difference between participants who completed all three time points (M = 106.73, SD = 21.61) and those who did not (M = 91.08, SD = 13.66) on the PS (t [37] = −2.38, p = 0.023).

Table 1 Results of paired samples t-tests

Tests of Treatment Effects

Primary analyses were conducted twice, once using listwise deletion methods which resulted in a sample of 21, and again using the last observation carried forward (LOCF) method to impute missing data, which resulted in a sample of 37 (missing items on some measures resulted in invalid total scores, so those scores were not included in analyses). LOCF provides a conservative estimate of treatment effects and has an equally low risk of Type 1 error as other data imputation methods (Overall et al. 2009). To test primary hypotheses, a one-way within-subjects repeated measures multivariate analysis of variance (RM-MANOVA) was used. Results using listwise deletion methods revealed a significant within-subjects effect of time (Wilks’ Lambda = 0.45, F [6, 76] = 3.22, p = 0.007; multivariate partial eta squared = 0.20). Results using LOCF were similar, with a significant within-subjects effect of time (Wilks’ Lambda = 0.64, F [6, 31] = 2.91, p = 0.023; multivariate partial eta squared = 0.36). Because both approaches yielded similar findings, all following univariate analyses were conducted using listwise deletion for the most conservative estimate of treatment effects. Paired samples t-tests (Table 1) were used to determine treatment effects on individual dependent variables. Results from paired samples t-tests revealed statistically significant change on the PS occurred between T1 and T2 (t [27] = 2.97, p = 0.006) and statistically significant change on the ECBI occurred between T1 and T3 (t [27] = 2.36, p = 0.02). No statistically significant changes were found between any time points for the BCAP.


The mean score and standard deviation on the Discussion Group Satisfaction Questionnaire (DG-SQ; M = 6.30; SD = 0.71) and Client Satisfaction Questionnaire (CSQ; M = 5.71; SD = 1.10) were very high immediately following the group and 3 weeks later (Tables 2 and 3). Additionally, a paired samples t-test revealed no significant difference between parents’ immediate satisfaction (measured by DG-SQ total scale score) and their satisfaction three weeks later (measured by CSQ total scale score; t [20] = 1.32, p = 0.20).

Table 2 Discussion group—Satisfaction survey descriptives
Table 3 Client satisfaction questionnaire descriptives


Despite high rates of child welfare involvement among families experiencing homelessness (Keeshin & Campbell, 2011; Park et al. 2004), there is little research on the effectiveness or acceptability of evidence-based parenting programs for these families. This study is the first to evaluate the use of Triple P Discussion Groups delivered in shelters for families experiencing homelessness. Overall, findings provide preliminary evidence that this brief group intervention is effective at improving parenting practices and child behaviors for these families. Mothers reported significant improvement in their parenting behavior 3 weeks after the intervention but did not report significant changes in their child’s behavior until 3 months after the group. This pattern was not reported in prior studies of Discussion Groups; however, given the mechanism by which child behavior change is hypothesized to occur (i.e., social learning model of parent–child interaction), improved parent behavior preceding changes in child behavior is not surprising (Forehand et al. 2014), especially in our high-risk sample compared to rates of change of families in prior studies involving samples of parents with higher income.

Despite positive changes in parenting practices and child behavior following the intervention, there were no changes in parents’ child abuse risk as measured by the BCAP. Lack of change in risk for abusive parenting could be explained by several factors. First, the BCAP is a more distal measure of treatment effectiveness than our other measures because it does not directly assess parenting behavior or child behavior—the two main targets of Triple P. Although scores on the BCAP have been correlated with parenting behavior in prior studies, BCAP scores are most closely correlated with parents’ psychological well-being (Dawe et al. 2017), and evidence is mixed regarding effects of Triple P Discussion Groups on parents’ psychological well-being. That is, of the three studies on Discussion Groups that measured parental psychological well-being, only one found significant effects (Mejia, et al. 2015). Second, the mean risk score of this sample was high (M = 11.82) and just below the conservative clinical cutoff of 12. In fact, 61% of the sample scored at or above the cutoff at pre-intervention. That percentage decreased to 41% at post-treatment then increased again to 52% at the 3-month follow-up. In other studies that have utilized the CAP Inventory as a predictor of treatment outcomes, sample mean scores prior to intervention have not approached the risk score cut off (Costello & Moreland 2015). Perhaps mothers in this study were at a substantially higher risk than other samples before intervention, and might have needed more intensive intervention to experience reductions in risk for abuse.

Importantly, mothers were highly satisfied with the program initially and again three weeks later, suggesting the program, unaltered, was acceptable to this unique population of parents. High satisfaction aligns with investigations of other levels of Triple P in shelter environments (Haskett et al. 2018; Wessels & Ward, 2016). Given high acceptability, Triple P Discussion Groups should be considered for wide use in shelters.

Fidelity to the Triple P Discussion Group model during the groups was 100%, but there were some topics and adverse reactions that are important to mention because they highlight providers’ ability to adapt the intervention to meet the unique needs of this sample while maintaining treatment fidelity. At the start of the first group, a parent asked why the provider wanted to offer Triple P to the parents; she wondered whether the women had been targeted for intervention because they were “all bad parents” and in need of parenting intervention. From that point forward, the provider began each group by normalizing parenting struggles and explicitly stating that residing at the shelter did not make participants a “bad parent.” Additionally, each group had a negative reaction to the recommended time out routine shown in a video in which a child was put in the bathroom with the door closed. Given frequent involvement of Child Protective Services in the lives of these families, mothers’ reactions to using a closed room for time out was understandable, thus other more acceptable ways to utilize time out were discussed by the parents. Finally, topics that were unique to living in shelters were discussed spontaneously at all four groups, including novel places to conduct time out, maintaining safety of children in an unsafe environment (e.g., bunk beds, concrete floors), shared experiences of homelessness and how homelessness had impacted their children, stress of parenting in public, and use of group positive behavior charts for use by all families in communal sleeping spaces. Facilitators providing Discussion Groups in shelters could consider incorporating these elements. The Triple P Discussion Group model allows ample flexibility to address the unique concerns and ideas of parents residing in shelters while maintaining fidelity to the intervention.

Wide variability of pre-intervention functioning of the mothers and children highlights heterogeneity within the sample and the resilience of many of these parents and children. Despite evidence of individual differences in functioning, it is important to note that mothers involved in this study demonstrated overall risk for negative parenting in terms of scores on measures of child behavior problems and abuse potential. Specifically, almost half of the mothers (47%) met or exceeded the clinical cut off indicating serious child behavior challenges and 61% met or exceeded the clinical cut off indicating elevated risk for child abuse. The Triple P system—designed to be a population level intervention—promotes minimal sufficiency such that the minimal level of parenting support and time is delivered for positive effects. Though providers may be inclined to deliver a higher level of service for parents experiencing homelessness, this study suggests that Level 3 Triple P Discussion Groups may be sufficient for some families. In fact, given the variability of baseline scores across the three outcome measures, it is likely that a lower level of intensity might have been sufficient for some parents who attended the group while there were others who could have benefited from more intensive Triple P. Future studies that are more highly powered should examine moderators of treatment effects, as there are likely subgroups of parents residing in shelters for whom Discussion Groups might be more or less effective. In summary, these promising preliminary findings and high satisfaction ratings of Triple P Discussion Groups suggests that Discussion Groups are a potentially effective and acceptable way to promote positive parenting practices and reduce child behavior challenges for sheltered families experiencing homelessness with children ages 2–6 years of age.

Limitations and Future Directions

Our study provides preliminary evidence to support the use of Triple P Discussion Groups in shelters, but it is not without limitations. The most serious limitation is the lack of a control group, which would have allowed for more conclusive results regarding treatment effectiveness. Without a treatment-as-usual or wait-list control group, it is difficult to rule out confounding variables. Another limitation of this study is that the primary investigator also served as the intervention provider. Ideally, an accredited Triple P provider from the community or shelter would have facilitated the groups to increase external validity and increase sustainability of Triple P delivery in the shelter program.

A multi-method multi-informant assessment would have provided the strongest assessment of outcomes, but that was not feasible due to limited time and funding. Future studies could use a second rater of child problem behaviors (e.g., such as shelter case manager) and observational measures of parent–child interactions and child behaviors to assess changes more comprehensively across time. Additionally, internal consistency of the Parenting Scale was marginally acceptable at pre-treatment (α = 0.67) but was reasonable at later time points. In the first group of parents assessed, several mothers seemed confused by the rating scales. In subsequent assessment sessions, more instructions were provided to the mothers individually to ensure understanding. We learned that investigators must carefully administer measures to this highly stressed group of parents who reside in crowded, chaotic environments.

In terms of threats to internal validity of the study, there was a 33% attrition rate among participants and intent-to-treat analyses were not feasible due to the within-group design of the study. At pre-treatment, mothers who were lost to attrition endorsed significantly lower dysfunctional parenting practices as measured by the Parenting Scale. It is unclear if their lower endorsement of dysfunctional practices or potential perceived lack of needing parenting support contributed to their lower attendance. Most parents who dropped out reported work conflicts, shelter demands, and changing residence as reasons they could not attend the group. It is noteworthy that many parents who left the shelter were still able to be located and completed assessment at follow-up (n = 11). Future larger, controlled studies could better account for drop-out, which was less than anticipated given the transient nature of the population. As an additional threat to internal validity, many mothers reported major life changes had occurred between pre- and post-intervention and follow-up. Without a control group, it is difficult to attribute changes to the intervention alone as other major life events were occurring at the same time for these families (e.g., job/childcare changes, moving residences, etc.). However, it is impressive that parents reported significant changes in parenting practices and child problem behaviors after Discussion Groups despite stressful life changes occurring during that time.

In terms of fidelity to the Triple P model, adherence was 100% in the groups, but completion of the follow-up phone calls for individual support was very low. Less than 25% of mothers completed even one of the two planned follow-up phone calls, and no mothers completed both phone calls in the 2 weeks following the group. Many mothers expressed not having cell phone minutes to talk on the phone, but they maintained the ability to text via wireless connection for scheduling post-intervention and follow-up assessments. Future studies could conduct follow-ups via text to determine if that brief form of communication can be effective. Alternatively, future studies could omit follow-up calls considering this study found effects without completion of phone calls for most parents. In fact, Dittman et al. (2016) found strong effects without inclusion of the individual phone calls with a middle to high income sample.

Generalization of findings is restricted to mothers residing in shelters. Fathers were not available for participation because they were not allowed to reside at the shelter. Future research is necessary to determine whether this intervention is effective for fathers who are homelessness. Results from a recent study indicate that mothers might benefit more than fathers from participation in brief Triple P interventions (Palmer et al. 2019). Future investigations also should examine the impact of Triple P groups for homeless parents not residing in shelters. Most homeless families live doubled-up or in low-rate motels (U.S. Department of Education 2020). Reaching and serving those parents might be even more challenging than serving families in shelters, but we must make a concerted effort to serve them.

Funding and agency buy-in are necessary to complete a fully randomized experimental design of Triple P Discussion Groups in shelters. Our study provides preliminary evidence needed to seek funding needed to complete more rigorous studies in the future. Additionally, this study provides evidence that extensive adaptations to evidence-based programs may not be necessary to serve vulnerable and underserved populations in ways that are viewed as highly acceptable and effective. Triple P Discussion Groups appear to provide enough flexibility within fidelity to meet the unique needs of mothers experiencing homelessness and show strong promise for being both acceptable and effective.


  1. Abate, A., Marek, R., Venta, A., Taylor, L., & Velez, L. (2020). The effectiveness of a home-based delivery of Triple P in high-risk families in rural areas. Journal of Child and Family Studies, 29, 997–1007.

    Article  Google Scholar 

  2. Arnold, D. S., O’Leary, S. G., Wolff, L. S., & Acker, M. M. (1993). The Parenting Scale: a measure of dysfunctional parenting in discipline situations. Psychological Assessment, 5, 137–144.

    Article  Google Scholar 

  3. Bassuk, E. L., Richard, M. K., & Tsertsvadze, A. (2015). The prevalence of mental illness in homeless children: a systematic review and meta-analysis. Journal of the American Academy of Child and Adolescent Psychiatry, 54, 86–96.

    Article  PubMed  Google Scholar 

  4. Chung, S., Leung, C., & Sanders, M. (2015). Triple P—Positive Parenting Program: the effectiveness of group Triple P and brief parent discussion group in school settings in Hong Kong. Journal of Children’s Services, 10, 1–5.

    Article  Google Scholar 

  5. Costello, A. H., Moreland, A. D., Jobe-Shields, L., Hanson, R. F., & Dumas, J. E. (2015). Change in child abuse potential as a predictor of post-assessment child disruptive behaviors after participation in PACE. Journal of Child and Family Studies, 24, 2989–2998.

    Article  Google Scholar 

  6. Cutuli, J. J., & Herbers, J. E. (2014). Promoting resilience for children who experience family homelessness: opportunities to encourage developmental competence. Cityscape, 16(1), 113–139.

  7. Dawe, S., Talpin, S., & Mattick, R. P. (2017). Psychometric investigation of the Brief Child Abuse Potential Inventory in mothers on opioid substitution therapy. Journal of Family Violence, 32, 341–348.

    Article  Google Scholar 

  8. Dittman, C. K., Burke, K., & Hodges, J. (2020). Brief parenting support for parents of teenagers dealing with family conflict: a randomized controlled trial. Child & Youth Care Forum, 49, 799–816.

    Article  Google Scholar 

  9. Dittman, C. K., Farruggia, S. P., Keown, L. J., & Sanders, M. R. (2016). Dealing with disobedience: an evaluation of a brief parenting intervention for young children showing noncompliant behavior problems. Child Psychiatry and Human Development, 47, 102–112.

    Article  PubMed  Google Scholar 

  10. Eyberg, S. M., & Ross, A. W. (1978). Assessment of child problem behaviors: the validation of a new inventory. Journal of Clinical Child Psychology, 7, 113–116.

    Article  Google Scholar 

  11. Forehand, R., Lafko, N., Parent, J., & Burt, B. B. (2014). Is parenting the mediator of change in behavioral parent training for externalizing problems of youth? Clinical Psychology Review, 34, 608–619.

    Article  PubMed  PubMed Central  Google Scholar 

  12. Friedman, D. H. (2000). Parenting in public: Family shelter and public assistance. Columbia University Press, New York, NY.

  13. Glazemakers, I., & Deboutte, D. (2013). Modifying the “Positive Parenting Program” for parents with intellectual disabilities. Journal of Intellectual Disability Research, 57, 616–626.

    Article  PubMed  Google Scholar 

  14. Haskett, M. E., Armstrong, J. M., Neal, S. C., & Aldianto, K. (2018). Perceptions of Triple P- Positive Parenting Program Seminars among parents experiencing homelessness. Journal of Child and Family Studies, 27, 1957–1967.

    Article  Google Scholar 

  15. Haskett, M. E., Loehman, J., & Burkhart, K. (2014). Parenting interventions in shelter settings: a qualitative systematic review of the literature. Child & Family Social Work, 21, 272–282.

    Article  Google Scholar 

  16. Haskett, M. E., Perlman, S., & Cowan, B. A. (2014). Supporting families experiencing homelessness. Springer, New York, NY.

  17. Herbers, J., Cutuli, J. J., Supkoff, L. M., Narayan, A. J., & Masten, A. S. (2014). Parenting and coregulation: adaptive systems for competence in children experiencing homelessness. American Journal of Orthopsychiatry, 84, 420–430.

    Article  Google Scholar 

  18. Holtrop, K., Chaviano, C. L., Scott, J. C., & McNeil Smith, S. (2015). Identifying relevant components to include in a parenting intervention for homeless families in transitional housing: Using parent input to inform adaptation efforts. American Journal of Orthopsychiatry, 85, 600–611.

    Article  Google Scholar 

  19. Joachim, S., Sanders, M. R., & Turner, K. M. T. (2010). Reducing preschoolers’ disruptive behavior in public with a brief parent discussion group. Child Psychiatry and Human Development, 41, 47–60.

    Article  PubMed  Google Scholar 

  20. Keeshin, B. R., & Campbell, K. (2011). Screening homeless youth for histories of abuse: prevalence, enduring effects, and interest in treatment. Child Abuse & Neglect, 35, 401–407.

    Article  Google Scholar 

  21. Kelley, M. L., Lawrence, H. R., Milletich, R. J., Hollis, B. F., & Henson, J. M. (2015). Modeling risk for child abuse and harsh parenting in families with depressed and substance-abusing parents. Child Abuse and Neglect, 43, 42–52.

    Article  PubMed  Google Scholar 

  22. Lee, S. S., August, G. J., Gewirtz, A. H., Klimes-Dougan, B., Bloomquist, M. L., & Realmuto, G. M. (2010). Identifying unmet mental health needs in children of formerly homeless mothers living in a supportive housing community sector of care. Journal of Abnormal Child Psychology, 38, 421–432.

    Article  PubMed  Google Scholar 

  23. Lim, J., Wood, B. L., Miller, D., & Simmens, S. J. (2011). Effects of paternal and maternal depressive symptoms on child internalizing symptoms and asthma disease activity: mediation by interparental negativity and parenting. Journal of Family Psychology, 25, 137–146.

    Article  PubMed  PubMed Central  Google Scholar 

  24. Masten, A. S., Cutuli, J. J., Herbers, J. E., Hinz, E., Obradovic, J., & Wenzel, A. J. (2014). Academic risk and resilience in the context of homelessness. Child Development Perspectives, 8, 201–206.

    Article  PubMed  PubMed Central  Google Scholar 

  25. Mejia, A., Calam, R., & Sanders, M. R. (2015). A pilot randomized controlled trial of a brief parenting intervention in low-resource settings in Panama. Prevention Science, 16, 707–717.

    Article  PubMed  Google Scholar 

  26. Morawska, A., Adamson, M., Hinchliffe, K., & Adams, T. (2014). Hassle Free Mealtimes Triple P: a randomised controlled trial of a brief parenting group for childhood mealtime difficulties. Behaviour Research and Therapy, 53, 1–9.

    Article  PubMed  Google Scholar 

  27. Morawska, A., Haslam, D., Milne, D., & Sanders, M. R. (2011). Evaluation of a brief parenting discussion group for parents of young children. Journal of Developmental and Behavioral Pediatrics, 32, 136–145.

    Article  PubMed  Google Scholar 

  28. Obradovic, J. (2010). Effortful control and adaptive functioning of homeless children: variable-focused and person-focused analyses. Journal of Applied Developmental Psychology, 31, 109–117.

    Article  PubMed  PubMed Central  Google Scholar 

  29. Ondersma, S., Chaffin, M., Mullins, S., & LeBreton, J. (2005). A brief form of the Child Abuse Potential Inventory: development and validation. Journal of Clinical Child and Adolescent Psychology, 34, 301–311.

    Article  PubMed  Google Scholar 

  30. Overall, J. E., Tonidandel, S., & Starbuck, R. R. (2009). Last-observation-carried-forward (LOCF) and tests for difference in mean rates of change in controlled repeated measurements designs with dropouts. Social Science Research, 38, 492–503.

    Article  Google Scholar 

  31. Palmer, M. L., Keown, L. J., Sanders, M. R., & Henderson, M. (2019). Enhancing outcomes of low-intensity parenting groups through sufficient exemplar training: a randomized control trial. Child Psychiatry & Human Development, 50, 384–399.

    Article  Google Scholar 

  32. Park, J. M., Metraux, S., Brodbar, G., & Culhane, D. P. (2004). Child welfare involvement among children in homeless families. Child Welfare, 83(5), 423–436.

    PubMed  Google Scholar 

  33. Pelaez, M., Field, T., Pickens, J. N., & Hart, S. (2008). Disengaged and authoritarian parenting behavior of depressed mothers with their toddlers. Infant Behavior and Development, 31, 145–148.

    Article  PubMed  Google Scholar 

  34. Perlman, S., & Fantuzzo, J. (2010). Timing and influence of early experiences of child maltreatment and homelessness on children’s educational well-being. Children and Youth Services Review, 32, 874–883.

    Article  Google Scholar 

  35. Perlman, S., Sheller, S., Hudson, K. M., & Wilson, C. L. (2014). Parenting in the face of homelessness. In M. E. Haskett, S. Perlman, & B. Cowan (Eds.), Supporting families experiencing homelessness. (pp. 57–77). Springer, New York, NY.

  36. Prinz, R. J., Sanders, M. R., Shapiro, C. J., Whitaker, D. J., & Lutzker, J. R. (2009). Population-based prevention of child maltreatment: the U.S. Triple p system population trial. Prevention Science, 10, 1–12.

    Article  PubMed  PubMed Central  Google Scholar 

  37. Rhoades, K. A., & O’Leary, S. G. (2007). Factor structure and validity of the Parenting Scale. Journal of Child and Adolescent Psychology, 36, 137–146.

    Article  Google Scholar 

  38. Sanders, M. R. (1999). Triple P-Positive Parenting Program: towards an empirically validated multilevel parenting and family support strategy for the prevention of behavior and emotional problems in children. Clinical Child and Family Psychology Review, 2, 71–90.

    Article  PubMed  Google Scholar 

  39. Sanders, M. R., Kirby, J. N., Tellegen, C. L., & Day, J. J. (2014). The Triple P-Positive Parenting Program: a systematic review and meta-analysis of a multi-level system of parenting support. Clinical Psychology Review, 34, 337–357.

    Article  PubMed  Google Scholar 

  40. Schilling, S., Lanier, P., Rose, R. A., Shanahan, M., & Zolotor, A. J. (2020). A quasi-experimental effectiveness study of Triple P on child maltreatment. Journal of Family Violence, 35, 373–283.

    Article  Google Scholar 

  41. Shapiro, C. J., Kilburn, J., & Hardin, J. W. (2014). Prevention of behavior problems in a selected population: stepping Stones Triple P for parents of young children with disabilities. Research in Developmental Disabilities, 35, 2958–2975.

    Article  PubMed  Google Scholar 

  42. Smith, S. M., Holtrop, K., & Reyonlds, J. (2015). Do positive parenting practices moderate parental mental health and child behavior among homeless families? Family Relations, 64, 606–620.

    Article  Google Scholar 

  43. Steele, R. G., Nesbitt-Daly, J. S., Daniel, R. C., & Forehand, R. (2005). Factor structure of the Parenting Scale in low-income African-American sample. Journal of Child and Family Studies, 14, 535–549.

    Article  Google Scholar 

  44. Swick, K. J., & Williams, R. (2010). The voices of single parent mothers who are homeless: implications for early childhood professionals. Early Childhood Education Journal, 38, 49–55.

    Article  Google Scholar 

  45. Tully, L. A. & Hunt, C. (2017). A randomized controlled trial of a brief versus standard group parenting program for toddler aggression. Aggressive Behavior, 43, 291–303.

    Article  PubMed  Google Scholar 

  46. U.S. Department of Education, National Center for Homeless Education (2020). Number of Homeless Children/Youth Enrolled in Public School, by Year.

  47. Walker, C. A., & Davies, J. (2010). A critical review of the psychometric evidence base of the Child Abuse Potential Inventory. Journal of Family Violence, 25, 215–227.

    Article  Google Scholar 

  48. Walker, C. A., & Davies, J. (2012). A cross-cultural validation of the Brief Child Abuse Potential Inventory (BCAP). Journal of Family Violence, 27, 697–705.

    Article  Google Scholar 

  49. Wessels, I., & Ward, C. L. (2016). Battered women and parenting: acceptability of an evidence-based parenting programme to women in shelters. Journal of Child and Adolescent Mental Health, 28, 21–31.

    Article  PubMed  Google Scholar 

  50. Yu, M., North, C. S., Lavesser, P. D., Osborne, V. A., & Spitznagel, E. L. (2008). A comparison study of psychiatric and behavior disorders and cognitive ability among homeless and housed children. Community Mental Health Journal, 44, 1–10.

    Article  PubMed  Google Scholar 

  51. Zugazaga, C. B. (2008). Understanding social support of the homeless: a comparison of single men, single women, and women with children. Families in Society: The Journal of Contemporary Social Services, 89, 447–455.

    Article  Google Scholar 

Download references


Partial financial support was received from Doris Duke Charitable Foundation, North Carolina State University Suniti-Anand Gupta Scholarship to Support Graduate Student Research, and NC State University Department of Psychology John Oliver Cook Dissertation Fellowship.

Author information



Corresponding author

Correspondence to Mary E. Haskett.

Ethics declarations

Conflict of Interest

The authors declare no competing interests.

Ethical Approval

This study was performed in line with the principles of the Declaration of Helsinki. Approval was granted by the Internal Review Board of NC State University.

Informed Consent

Written informed consent was obtained from the participants.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Armstrong, J.M., Owens, C.R. & Haskett, M.E. Effects of a Brief Parenting Intervention In Shelters For Mothers And Their Children Experiencing Homelessness. J Child Fam Stud 30, 2097–2107 (2021).

Download citation


  • Parent training
  • Triple P
  • Maltreatment prevention
  • Homelessness
  • Family shelters