Parenting programs are implemented globally as part of national and international strategies to enhance the well-being and development of children by reducing harmful and enhancing supportive parenting practices (Backhaus et al., 2023; UNICEF, 2014). Decades of rigorous intervention evaluation research show that parenting programs based on attachment and learning theory principles can indeed successfully change parenting practices and improve children’s outcomes, including reducing physical and emotional violence against children and reducing children’s behavior problems (Leijten et al., 2018; Weisz & Kazdin, 2015). However, there remain gaps in our knowledge about parenting program effects. The present study evaluates changes in parental cognitions and parenting behavior in the context of an online parenting program implemented in social services in Croatia.

Gaps in the Global Parenting Program Literature

Many parenting programs strive to benefit the well-being of parents in addition to the well-being of children (Barlow et al., 2012). Parental well-being may benefit directly (e.g., through the social support provided by other parents in the group) as well as indirectly (e.g., through reduced behavior problems). However, the extent to which parenting programs improve parental well-being varies between programs (Lindsay et al., 2011). The present study includes a wide range of indicators of parental well-being (e.g., self-esteem, parental morale, and self-efficacy) to shed light on the aspects of parental well-being that are less or more affected by parenting support.

In addition, evidence predominantly comes from North America, Western Europe, and Australia (Leijten et al., 2016). Because parenting programs are implemented globally (Backhaus et al., 2023), it is important to create an evidence base in which families across countries are well-represented. The importance of studying parenting programs in diverse settings also aligns with what some call the need for a ‘heterogeneity revolution’ (Bryan et al., 2021): systematically sampling from different populations to better understand the conditions under which our interventions are effective. The present study was conducted in Croatia, a country at the intersection of central and southeast Europe, and underrepresented in the international literature on parenting program effects.

Programs under evaluation are often implemented for research purposes, under highly controlled settings (e.g., with program facilitators supported by academic staff and resources). However, intervention effects from highly-controlled studies may not always translate to effects in real-world settings (Baker-Ericzén et al., 2010). Understanding how well parenting programs work in routine practices requires evaluating them as part of routine practice. In the present study, we observed changes in family dynamics over the course of a parenting program implemented in routine practice.

Last, the Covid-19 pandemic accelerated the use of online family support. Examples include providing group Triple P through Zoom in Portugal (Canário et al., 2021) and the Online Parenting Pro-Tips (OPPT) program, combining web-based modules and live coaching via videoconferencing, for veteran parents in the United States (Riegler et al., 2020). Evidence accumulates that online settings can in some cases successfully replace traditional in-person settings (Spencer et al., 2020), but a better understanding is needed of the conditions under which online programs are less or more successful (Canário et al., 2022; Leijten, 2023). The present study examines changes in parental cognitions and parenting behavior in the context of an online parenting program.

Parenting Support in Croatia

Croatia, the youngest member of the European Union, is home to 246,477 young and preschool children (children ≤ six years old). Parenting practices in contemporary Croatian families are slowly changing, stimulated by the UN Convention on the Rights of the Child (1989), ratified in 1991. However, physical punishment and lack of joint play and stimulation are still common. Despite the legal ban of corporal punishment in 1999, both parents and children in a representative sample of 1074 seventh graders (thirteen year-olds) report that three in four children have been hit by their mothers and two in three by their fathers (Pećnik & Tokić, 2011). In addition, a significant proportion of parents reported not regularly engaging in positive interaction such as joint play and stimulation of their young child’s learning. (Pećnik, 2013). Violence and a lack of positive parental interaction were reported especially often by parents with lower level of education and in rural areas. The WHO advises the implementation of parenting programs to reduce parental violence against children (WHO, 2016, 2023). However, fear of stigma remains a barrier to help-seeking for parents in Croatia—one in three believe that professional support serves “problematic families, while the average parent should solve parenting problems on his/her own” (Pećnik, 2013).

Growing Up Together Online

Growing up Together (in Croatian: Rastimo zajedno) is developed in Croatia and in the early stages of being established as an empirically supported program. Growing up Together was first developed as a universal program (Pećnik & Starc, 2010), stimulated by the Council of Europe’s Recommendation (2006) to increase the availability of community-level parenting support and strengthen the professional competencies of parenting support providers. Grounded in children’s rights as enshrined in the UNCRC, including a right to support in fulfilling parental responsibilities, and the relational view of socialization and social relational theory (Kuczynski & Parkin, 2006), the program evolved around the key concept of ‘parenting in the best interest of the child’ (Pećnik, 2007). The principles of children’s provision, protection, and participation rights are referred to in the program as the ‘four pillars of parenting’: nurturing behavior (i.e., responsiveness and acceptance), providing structure and guidance (i.e., routines and boundaries), acknowledging the child as a person (i.e., the child’s own perspective), and enabling child’s empowerment (i.e., facilitating the child’s agency and initiative). This conceptualization is consistent with ‘needs-supportive parenting’ (Grolnick et al., 1997; Joussemet et al., 2008) and the ‘empathic dialogues’ of positive parent – child interaction (Hundeide & Armstrong, 2011). The program builds on attachment theory-based and social learning theory-based universal programs (e.g., International Child Development Program, Hundeide & Armstrong, 2011; Positive Discipline in Everyday Parenting, Durrant, 2016) and integrates their core common elements (e.g., parent-child relationship enhancement approaches and parental use of differential reinforcement to promote the child’s self-regulation) with elements unique to embedding the children’s rights-based and self-determination theory-based ‘four pillars of parenting’ to the local context (e.g., recognizing child’s feelings and needs when interacting with the child).

Both the program developer (Pećnik, 2019; Pećnik & Starc, 2010) and independent evaluators (Keresteš et al., 2017) found that at the end of the program, participants (N = 2114) reported less agreement with beliefs unresponsive to the needs and rights of a young child (e.g., supporting corporal punishment, ignoring child’s crying, not providing explanation of requests and limits) and higher parenting self-efficacy. Participants also reported more frequent positive parenting practices (e.g., reading picture books) and less frequent harsh practices (e.g., yelling, hitting a child). Data collected on a subsample (n = 224) showed that improvements in parenting self-efficacy and parental verbal and physical violence were maintained six months after program completion; improvements in positive parent-child interactions returned to pre-program levels.

A few years later, because socioeconomically disadvantaged parents were underrepresented among the users of the universal Growing up Together program (Keresteš et al., 2017), a more comprehensive program (Growing up Together – Count us In!; Pećnik et al., 2020) was developed to address both the universal and specific needs for support of socioeconomically vulnerable families engaged with social welfare or child protection services, and the needs of their preschool children. In addition to the content of the universal program, this program included content related to more general parental resources, such as self-regulation, co-parenting, and social network enhancement as well as an additional emphasis on strengths-based approach to facilitate parent’s trust and engagement. This program has been implemented through family centers in Croatia since 2018. Pre- and post- intervention self-report data suggest improvements in parenting stress and harsh parenting, as well as an increase in parenting morale, attempted understanding of the child’s perspective, and positive practices (e.g., joint play; Pećnik et al., 2021).

Growing up Together Online was developed in 2020 in response to the Covid-19 pandemic. The program was co-created with parents of preschool children involved with social services, who experienced difficulties in coping with parental responsibilities even before the additional stress produced by the pandemic. After physical distancing restrictions were relaxed, the program continues to be offered to parents who were unable to participate in in-person programs, because of a lack of transport in rural areas or lack of childcare during program hours. The content is similar to that of Growing up Together – Count us In!. It consists of ten sessions, structured into four consecutive modules (Table 1). It thus combines established common elements of parenting support programs, for which sound empirical support is available (e.g., Kaehler et al., 2016), with elements to address the specific needs of this population at risk for child maltreatment, increasing an emphasis on parents’ own well-being (Fukkink & Vink, 2014) and self-care (Melendez-Torres et al., 2019). In addition, similar to the approach used by other programs (e.g., Incredible Years), the key program content (i.e., the general principles of nurturing behavior, providing structure and guidance, acknowledging personal experience, and enabling empowerment) is reflected in how facilitators support parents.

Table 1 Growing up Together Online Program Content

Growing up Together Online may change both parental cognitions and behavior. It is well-known that parenting programs can have these effects. For example, universal programs for first-time parents can enhance parental feelings of self-efficacy (Amin et al., 2018) and different programs increase parental use of positive reinforcement techniques and reduce parental use of corporal punishment (e.g., Leijten et al, 2018; Sanders et al., 2014). Meta-analyses by, among others, Leijten et al. (2023) and Spencer et al. (2020) show that similar effects can be obtained in online parenting programs. Parental cognitions and behavior are both directly targeted in the program, but parental cognitions and behavior are also intertwined such that if one of the two changes, the other is likely to also change. Interventions designed to specifically strengthen parental feelings of self-efficacy, for example, are known to have spillover effects on parenting behavior, and even child behavior (Mouton & Roskam, 2015). This may be because when parents feel more confident, they are better able to remain calm in challenging parenting situations and refrain from harshness and hostility (Jones & Prinz, 2005). Similarly, specifically targeting parental behavior can have spillover effects on parental cognitions (Bodenmann et al., 2008). This may be because when parents see the positive effects of changes in their behavior on their children (i.e., mastery experiences), this strengthens their feelings of self-efficacy (Bandura, 1977). In the present study we therefore assess changes in both parental cognitions and behavior.

For parental cognitions, we distinguish between parenting-related cognitions and general cognitions. Because, unlike most other programs, Growing up Together Online also explicitly targets parents’ general self-esteem, we assessed whether parents hold more adaptive cognitions after the program not only about themselves as a parent, but also about themselves as a person. There is some evidence from previous studies that this can be done successfully (e.g., Fuscaldo et al., 1998). Although the ultimate goal of most parenting programs is to enhance children’s well-being and development, a more detailed understanding of the benefits of parenting programs for parents themselves provides a more comprehensive picture of the societal value of parenting programs. Some of the cost-effectiveness of parenting programs seems driven by reduced loss of productivity in parents (Posthumus, 2009). Such effects could perhaps be explained by parenting programs potentially increasing parents’ general self-esteem.

For parental behavior, we distinguish between increases in parenting behaviors known to contribute to healthy child development, such as parental attempts to see things from their child’s perspective and appropriate reinforcement of positive child behavior (Dishion et al., 2008; Eisenberg et al., 2005), and reductions in parenting behaviors known to compromise healthy child development, such as hostility and physical punishment (Gershoff, 2002). Importantly, these two aspects of parenting are not just two sides of the same coin. Some parents show both high levels of harshness and positive involvement (Weeland et al., 2022), or low levels of harshness and positive involvement (e.g., lax or neglecting parents; Karavasilis et al., 2003). Previous findings are inconsistent about whether parenting programs affect both aspects of parenting equally. A recent individual participant data meta-analysis suggests that parenting program effects on harsh parenting are at least as substantial as effects on, for example, positive reinforcement (Leijten et al., 2018), but meta-analyses of studies specifically focusing on families at risk for child maltreatment suggest harshness can be difficult to reduce (Euser et al., 2015).

The Present Study

The aim of the present study was to expand our understanding of the benefits of parenting programs for parental cognitions and behavior, by examining changes in parents who participate in an online parenting program implemented in social services in Croatia. We did so in a so called ‘hybrid design’ where the implementation and evaluation are combined in one study (Curran et al., 2022). Such hybrid studies are increasingly encouraged to narrow the gap between the development and use of innovations (Shepardson & Polaha, 2023). Because the program was implemented as routine practice, no randomization took place, but parents were asked to report on their thoughts and feelings, and on their interactions with their child, immediately prior and after the intervention. We hypothesized changes in parental thoughts, feelings, and behavior and examined both the magnitude of these changes as well as the percentage of families showing reliable change, following Jacobson and Truax’s (1991) guidelines.

Methods

Participants

Initially, seventy-four parents with at least one preschool child enrolled in the program. This was 67% of the 110 parents invited to enroll. The main reasons that parents provided for not enrolling were working hours (shifts; farming) that were incompatible with the program schedule; child care commitments (e.g., doctor appointments; lack of child care); lack of interest in a group-based or online intervention. Of the 74 enrolled parents, 62 (84%) completed the program, defined as attending at least six sessions (M = 7.55) and staying enrolled until posttest.

More detailed attendance rates were available for around 50% of the groups (e.g., 22% attended all 10 sessions). When parents missed a session, facilitators offered parents an individual catch-up session. Twenty-six percent of parents used one to four catch-up sessions (M = 1.55). These catch-up sessions are included in the abovementioned attendance rates. In total, 16% of parents dropped out. Reasons for drop-out included work obligations (e.g., unexpected shift changes, finding a new job), study obligations (e.g., one mother was balancing participation with finishing her college education), weak internet connection, and addiction problems.

Parents participated in 18 groups in 11 cities throughout Croatia. They were predominantly female (89%) and had a mean age of 33 years (SD = 5.34, range 21–44 years). The majority of parents did not attend higher education; specifically, nearly three-quarters reported primary or secondary school as their highest academic level on a 4-point scale (1 = primary education, 2 = secondary education, 3 = vocational education, 4 = university degree). All parents had at least one preschool-aged child and ten parents were raising more than one preschooler. Ages of preschool children ranged from 6 months to 7 years (M = 4.04 years; SD = 1.81).

Procedure

Of the 62 parents that completed the program, 46 were referred to the program by child protection services departments. Of these 46 parents, 17 (37%) were mandated to attend by legal measures, while others were recommended to enroll voluntarily. Of the remaining 16 participants, ten parents were referred to enroll voluntarily by family centers (preventive part of social services providing family counseling or mediation), four were recommended to enroll by kindergarten professionals, and two parents had heard about the program from former participants and enrolled self-referred. The six parents who were not referred by social services were eligible because they were coping with multiple stressors (e.g., single parents with financial problems who experienced problems in their parental role). Parents were excluded if their level of mental health and addiction problems prevented them from participation in a group program. There were no other exclusion criteria.

In an intake interview parents received information about the program and decided if they would enroll. Data were collected at the beginning of the first session (T1) and at the end of the final session (T2). Parents were informed of the purpose of data collection and invited to provide it voluntarily. To reduce social desirability, parents filled in the online survey anonymously on tablets, lent to them during program participation, with a self-generated code that allowed for linking individual participant data at T1 and T2. There were three waves of program implementation. Wave 1 (autumn 2020) consisted of four groups (n = 15) in Pula, Split, Varaždin, and Vukovar; wave 2 (spring 2021) of three groups (n = 16) in Split, Varaždin, and Vukovar; and wave 3 (autumn 2021) of eleven groups (n = 31) in Čakovec, Dubrovnik, Osijek, Požega, Senj, and Slavonski Brod in addition to the original four cities.

Intervention

Growing up Together Online (in Croatian: Rastimo Zajedno Online; Ustić et al., 2021) uses a group format to create a collaborative and supportive environment where parents share experiences, learn about family needs and ways to satisfy these under various life circumstances, increase understanding of themselves as parents, especially their strengths, receive support in finding their own solutions to the difficult situations they cope with, and develop their social and parenting skills. The program consists of ten thematically linked sessions (Table 1). Sessions focus not exclusively on parent-child interactions, but also on parents as individuals, addressing parents’ personal strengths and their relationships with other adults. Facilitators model the four pillars corresponding to the four themes that are central in the program: empowering parents to feel nurtured and acknowledged (‘seen’) by facilitators, providing structure and guidance, and encouraging parents to appreciate and further develop their personal and parental strengths.

The program starts with a 2-hour in-person session at the family center where parents bring their children, complete pre-intervention assessments, and receive the intervention materials. Intervention materials included borrowed tablets with internet connection, a package with reading materials, hand-outs with parent-child activities to use during the program, and some creative materials for the children. Parents received instructions on how to access the secured online environment of their group on the Growing up Together website where parents exchange output (e.g., pictures of creative parent-child activities) and interact with each other on discussion boards.

The eight main sessions (each lasting around 1.5-hour) are delivered remotely via video call. Each session consists of an introductory block (recap of previous session and home assignments, check-in activity), a main block (short interactive presentations, prompted discussions), and a closing block (personal growth and take-home journal, suggestions for home assignments, parental feedback on the session). Throughout the program parents are encouraged to share experiences and take active steps in self-care and positive parenting practices through keeping a shared online journal of self-care activities and enjoyable moments spent with their children and families. Through a mobile app facilitators reminded parents about upcoming sessions and answered questions about assignments and participation in the program.

The program ends with a 2-hour in-person session at the family center for celebrating program completion, post-intervention assessments, and handing back the tablets. Due to pandemic-related restrictions, however, 11 of 18 groups had conducted their final session through videocall.

Parenting groups consisted of 5 to 6 parents and two co-facilitators. Facilitators were MA level psychologists, social workers, and social pedagogues (specialists in behavioral disorders) who already worked at the family centers and received additional training to deliver the program. In 7 of the 18 groups at least one of co-facilitators was also the program developer. The remaining 11 groups were led by facilitators who had been trained by the program developers and were supported by them through group supervision. Facilitators were self-selected from a national network of facilitators licensed for providing in-person parenting support program and with experience in working with vulnerable families. Fidelity of program implementation was monitored by detailed, structured facilitators’ written reports of every session and group supervision.

Fidelity

Adherence to the program manual was monitored by the Growing up Together Center and program authors in all groups. Facilitators completed structured session report sheets on the duration of each activity in minutes and qualitative data about (1) facilitators’ experience with each activity, and (2) perceived parents’ responses to each activity. These data were used to monitor fidelity in all cohorts, as well as for refinement of the manual in the first cohort. Deviations from the manual (e.g., omitting an activity) occurred in less than 4% of all sessions, with the except of the first cohort, where one group did not thoroughly implement the empowerment journal as a homework task. Nonadherence mostly manifested in minor changes in time allocated to particular activity or in the order of activities, to adapt to parents’ needs and resources.

Facilitators had access to the report sheets from other ongoing groups, to allow for timely problem-solving when implementation difficulties arose. Facilitators had monthly group supervision meetings with the program authors. These ‘reflection spaces’ (reports and supervision) were used to enable authors and facilitators to seek the balance between fidelity to the manual and responsiveness to needs of participants in a particular group. In so doing, we tried to respected the argument that fidelity and adaptation are both necessary for quality implementation (Anyon et al., 2019).

Measures

We collected information on family sociodemographic (parent gender, educational level, and number of preschool children) at pretest, and on parental cognitions (general self-esteem and parenting self-efficacy, experience, and morale) and on parenting behavior (attempted understanding, positive involvement, angry outbursts, and physical and emotional harshness) at pretest and posttest. All measures are based on validated questionnaires. It is possible, however, that some of their psychometrics changed as a result of translation to Croatian and because in three of them we were able to only include a subset of the items.

Parental Cognitions

Parents general self-esteem

Parents’ general self-esteem was assessed with six items from the Rosenberg Self-esteem Scale (RSE; Rosenberg, 1965; Bezinović, 1988). Using a five-point Likert scale (1 = disagree strongly to 5 = agree strongly), parents were asked to complete the stem question “How true are these statements for you?” for items such as “I have a number of good qualities” and “On the whole, I am satisfied with myself.” Internal consistency was good at both time points (α > 0.83).

Parenting self-efficacy

Parenting self-efficacy was assessed with nine items from the Positive Engagement Subscale of the Fathering Self-efficacy Scale – (FSES; Sevigny et al., 2016). Using a five-point Likert scale (1 = disagree strongly to 5 = agree strongly), parents were asked to complete the stem question “How true are these statements for you?” for items such as “I am able to explain rules to my child in a way she or she can understand” and “I can sense when my child is starting to feel frustrated or upset.” Internal consistency was good at both time points (α = 0.85).

Parenting experience

Parents’ parenting experience was assessed with six items from the Parenting Experience Scale (PES; Benzies et al., 2013). Using a five-point Likert scale (1 = disagree strongly to 5 = agree strongly), parents were asked to complete the stem question “How true are these statements for you?” with items such as “I have someone to talk to when I need support” and “I am able to manage stress.” Internal consistency was good at both time points (α > 0.76).

Parenting morale

Parents’ parenting morale was assessed with ten items of the Parenting Morale Index (PMI; Trute et al., 2009). Using a five-point Likert scale (1 = never to 5 = all the time), parents answered the stem question “When you think about your daily life as a parent, how often do you feel … ?” with items such as “optimistic” and “worried” (reversely scored). Internal consistency was good at both time points (α > 0.71).

Parenting Behavior

Attempted understanding

Parents’ attempted understanding of their child’s misbehavior was measured with the stem question “How do you react when your child has done something you really don’t like?” The Attempted Understanding Scale (Stattin et al., 2011) included four items such as “I try to understand how my child thought and felt” and parents assessed the frequency of each reaction on a five-point Likert scale (1 = never to 5 = all the time). The reliability of the scale was good for both time points (α > 0.70).

Positive involvement

Parents’ positive involvement and reinforcement frequency was assessed with a stem question “How often in the last 7 days did you …?”, followed by four items adapted from Pećnik (2019), such as “praise the child” and “told story, read or look at picture book.” Parents could answer the items on a five-point Likert scale (1 = never to 5 = all the time). The reliability of the scale was good for both time points (α = 0.79).

Angry outbursts

The frequency of angry outbursts was measured with four items of the Angry Outburst Scale (Stattin et al., 2011) on a five-point Likert scale (1 = never to 5 = all the time). The stem question was “How do you react when your child has done something you really don’t like?”, followed by items such as “I get angry and ‘explode’” and “My first reaction is anger and yell at my child.” The reliability of the scale was good for both time points (α > 0.76).

Physical and emotional harshness

Parents’ physical and emotional harshness was assessed by a stem question “How often in the last 7 days did you …?”. Followed by three items such as “threaten a child to hit them (but you did not)” and “hit the child on hand or bottom, or pulled their hair” measured on a five-point Likert scale (1 = never to 5 = all the time) adapted from a Pećnik (2019). The reliability of the scale was good for both time points (α > 0.76).

Program Satisfaction

At the end of the final session, parents completed twelve items on a six-point Likert scale on their satisfaction with the program: eight items on perceived benefits (e.g., “Exchanges with other parents was useful” and “The sessions have positively influenced me and my relationship with my child”) and four items on perceived burden (e.g., “Participating in the sessions was exhausting” and “The program was too long”). Scores on perceived burden were reverse coded such that higher sum scores indicate higher levels of program satisfaction. Because of an administration omission, program satisfaction items were completed only by Wave 1 and Wave 3 parents (n = 46; 79%). In addition to the quantitative survey, parents in all waves were asked qualitatively in the final session about program satisfaction and challenges.

Bias Reduction Strategies

To reduce social desirability bias, parents complete pre- and post-intervention assessments on tablets with self-generated code that allowed for linking individual participant data. To reduce researcher allegiance, the main researcher and developer of the original program (NP) was not involved in the development or implementation of the online variant of the program. In addition, analyses were conducted by an independent researcher (MN).

Analytic Strategy

Primary analyses (i.e., change in parental cognitions and behavior) were conducted in IBM SPSS, Version 27 (IBM Corporation, 2021) and secondary analyses (i.e., reliable change indices) in R (R Core Team, 2021). All alpha levels were set at .05 as criteria of significance. A power analysis (two-tailed, alpha .05, statistic power 0.80) based on a one-group pretest-posttest design, indicated that the sample size of 62 that we had an adequate statistical power of 80% to detect any effect size larger than d = 0.36. Prior to the analyses, the data were screened to verify the assumptions that should be met using linear regression. For 16 parents, data on program satisfaction were missing due to logistic reasons. One parent could not fill in questions on parent-child interactions (Positive involvement & reinforcement and Physical and emotional harshness) because the child was placed out of home. Parents with missing values did not differ from parents with complete data regarding their age, gender, number of children, and educational level. Moreover, Little’s Missing Completely at Random Test was not significant (Chi-square = 23593, df = 306, p < 0.999), indicating that these data were missing at random (MCAR; Little, 1988). We used multiple imputations by predictive mean matching (PMM; Rubin, 1986; Van Buuren et al., 2006) in IBM SPSS to address the problem of missing data. With predictive mean matching, missing values are imputed from an observed value of a (randomly) selected matching respondent. This was repeated ten times, the final dataset was formed by calculating the mean of all ten imputations for each missing value.

We examined changes in parental cognitions and behavior using paired t-tests. Especially in the case of within-person design, it is important to examine whether any observed change reflects reliable change. We therefore verified whether parents’ change was sufficiently robust to overcome the uncertainty associated with measurement unreliability. We calculated the Reliable Change Index (RCI; Jacobson & Truax, 1991) by dividing the difference between pretreatment and posttreatment scores by the standard error of the difference between the two scores. RCI scores above 1.96 or below –1.96 indicate that parents significantly improved or deteriorated, respectively.

Results

Descriptive Results

Scores on all outcome measures approach normality (i.e., skewness and kurtosis <1.50). Pre- and post-intervention scores are presented in Table 2.

Table 2 Pre- and Post-Intervention Levels of Parental Cognitions and Behavior for Parents Completing the Program

Changes in Parental Cognitions

Parents reported improvements in all four types of parental cognitions (Table 2). Parents reported more positive parenting experiences, a better parenting morale, and feeling more competent as a parent. And, outside the parenting context, parents reported an increase in their self-esteem. In terms of reliable change (Table 3), one in five or six parents reported changes in cognitions related to parenting that were so large that they met the stringent criteria for reliable improvement. This number was smaller for changes outside the parenting context (one in eleven). Reliable deterioration was very rare for all outcomes.

Table 3 Reliable Change Indices of Parental Cognitions and Behavior for Parents Completing the Program

Changes in Parenting Behavior

Parents reported improvements in two out of four indicators of parenting behavior (Table 2). Parents reported fewer angry outbursts and less physical and emotional harshness. Parents did not report change in attempted understanding of their child’s misbehavior or in positive involvement and reinforcement. Compared to parental cognitions, fewer parents reported changes in behavior that met the stringent criteria for reliable change (Table 3). The outcome with the most parents reporting reliable change was reduced angry outbursts (one in seven).

Sensitivity Analyses

We conducted sensitivity analyses to examine if outcomes differed between (1) mothers and fathers and (2) between parents who had one or multiple children. When we removed fathers from the sample, all findings remained the same, suggesting that parent gender is not associated with program outcomes. When we removed parents with more than one child from the sample (n = 10), all findings remained the same, except for angry outbursts—the decrease in angry outbursts was no longer significant. This suggests that mainly parents with multiple children reported reductions in angry outbursts. Results tables for all sensitivity analyses are included as supplemental material.

Program Satisfaction, Challenges, and Adverse Events

Quantitatively, parents were generally positive to very positive about the program with a mean score of 54 (SD = 4.86) on a possible scale of 12 (lowest possible score on all items) to 72 (highest possible score on all items). More than 50% of the parents scored on average at least a 4.5 on the 1 to 6-point Likert scale. Scores were noticeably higher and more consistent for perceived benefits (Ms 5.04 – 5.33; SDs 0.47 – 0.85) than for perceived burden (reverse coded for comparability, with higher scores indicating less burden; Ms 4.22 – 4.54; SDs 1.16 – 1.41), indicating that parents varied more in perceived burden than in perceived benefits.

Qualitatively, some parents said they sometimes found it difficult to concentrate on the session while having their children around at home. They felt that this may have occasionally led to temporarily increased tension or irritability between them and their children. Facilitators tried to support parents by suggesting activities children could do by themselves and by building in short breaks during the session to attend to the children. In the final session, parents were asked about additional challenges and adverse events. Some parents reported difficulties obtaining a stable internet connection. No adverse events were reported.

Discussion

The present study aimed to advance our understanding of the potential impact of parenting programs on family well-being by evaluating changes in parental cognitions and behavior in an online parenting program implemented as routine care for families referred to social services in Croatia. Parents reported higher levels of their general self-esteem, parental sense of competence and morale. Parents also reported lower levels of anger, harshness, and violent parenting. Parents reported no changes in parental attempts to understand children’s motives for misbehavior or increased positive interactions (e.g., praise and play).

The finding that parents reported increases in their experiences as parents (e.g., more self-efficacy, better morale, and less stress) confirms earlier findings that one of the key results of parenting programs is that parents feel more confident and less stressed in their parenting role (Barlow et al., 2012). This is a meaningful finding, because parents who feel more equipped are more likely to use effective parenting strategies, and to persist and stay calm in difficult parenting situations (Jones & Prinz, 2005; Mouton & Roskam, 2015).

But parents not only reported changes in cognitions regarding their parenting role specifically; they also reported changes in their general self-esteem. On the one hand, this may not be surprising, especially because the first session of Growing up Together Online explicitly asks parents to reflect on their strengths outside the parenting context. On the other hand, this finding may be quite significant. If such effects would spill over to parental functioning more generally (e.g., employment, marital quality, and social relationships outside the family; all factors associated with adult well-being; Patrício et al., 2014), the potential societal impact of parenting programs may be greater than previous research was able to identify. We of course cannot conclude this based just on the present unblinded and uncontrolled design, and encourage future research to attempt to replicate this finding in more controlled study designs.

We found that parents reported mainly reductions in harsh and hostile interactions, rather than increases in attempted understanding and positive involvement and reinforcement. One possible explanation for this is that many parents scored relatively high already pre-intervention on these outcomes (M = 18 on 4 to 20 scale for attempted understanding and M = 17 on 5–25 scale for positive involvement and reinforcement), leaving less room for further improvement. It has also been suggested that parents can become more self-critical about their positive involvement once they get expert advice on this (Posthumus, 2009). Another explanation could be that the Covid-19 pandemic and the earthquakes in Croatia in 2020 further intensified the already high levels of stress of families in social services (Buljan Flander & Galić, 2021; Jurić Vukelić et al., 2022). If the program alleviated some of this stress, this maybe have impacted especially parental harshness and expressed anger, because harsh parenting is strongly driven by parental feelings of stress (Conger & Donnellan, 2007; Chung et al., 2022).

Analyses of reliable change indices are in line with previous findings that families differ substantially from each other in how much they benefit from a parenting program (Weeland et al., 2022). In terms of the proportions of parents that report reliable improvements in parenting, our findings are similar to various previous studies (e.g., Donovan et al. 2022; Vardanian et al., 2020). That said, the number of families showing reliable change was relatively small. One reason for this might be that criteria for reliable change are fairly conservative. In our case, families had to move approximately two standard deviations on the scale to be identified as changing reliably. If all families in the sample would change this much, the overall effect size would be Cohen’s d = 2.00, something we rarely (if ever) reach in this field. In this light, it is not surprising that only a minority of families reached this bar. Instead, it raises important new questions about the conditions that allow families to benefit optimally, and how such conditions can be created.

Various study strengths allow for our study to expand the evidence base for parenting programs in several ways. Assessing both outcomes commonly included in program evaluation studies (e.g., parental sense of competence and parental harshness) and more unique outcomes (i.e., parental self-esteem), allows for both comparing our findings to earlier studies and extending these. In addition, our analyses of reliable change indices shed light on the heterogeneity in program benefits between families. Last, the study context of an online program implemented as part of routine care in social services in Croatia gives underrepresented families a voice in the international literature.

At the same time, various study limitations require us to interpret our findings with caution. First, we used a pre-posttest design, without a control group, in a relatively small sample, and cannot rule out natural improvements over time. Second, we had limited sociodemographic data, no measures of child outcomes, and our measures of parenting outcomes relied on parents’ self-reports, which could raise questions about potential bias and social desirability in parents’ reporting. Important to note here is that parents did not exclusively report change; we found a fairly robust pattern with changes in parental cognitions and harshness, and not in positive interactions. Third, although our sample size was sufficient for the analyses conducted in this study, it was insufficient to explore predictors of individual differences in program benefits, while our reliable change analyses showed that these benefits were vastly heterogeneous. In addition, because information about mandatory versus voluntary participation could not be connected to the anonymously collected self-report data, we were unable to test if mandatory participation predicted smaller, or larger, changes.

The main implication of this study for policy and practice is that participation in Growing up Together Online, an innovative program of two in-person and eight online sessions delivered through public social services, may improve parental cognitions and behaviors relevant for their child’s well-being, including physical and emotional violence. This was found in hard-to-reach families at high risk for suboptimal child development. To illustrate, arranging transportation or child care to be able to participate in an in-person parenting program on a weekly basis was an insurmountable difficulty for many of the parents who participated. Our study shows that Growing up Together Online can be provided by regular staff in family centers, contributing to the sustainability of the program and its accessibility to parents, that it reaches a high-risk population of families facing multiple life adversities (e.g., low income, conflictual divorce, intimate partner violence, and a history of early trauma) and a lack of informal and formal support, and is able to improve parental cognitions and behavior in these families. Importantly, the program addresses not only parenting knowledge, attitudes, and skills of parent-child interaction but also parent’s more broader personal resources such as general self-esteem and coping skills.

By identifying changes in parental cognitions and behavior, this study set the stage for future research on Growing up Together. Future research is needed to establish the program further as an evidence-based approach. In addition, future research should invest in understanding the key characteristics of the program that contribute to its effects and how effects could be further improved. For example, its common elements such as the relationship between facilitators and parents and the quality of implementation may be just as important as the program’s content (Cuijpers et al., 2019). In the current study, 40% of the groups had a program developer as a co-facilitator, and other facilitators were directly trained and supervised by program developers. It will be important to see how effects hold after wider dissemination. Such evaluations ideally also include longer-term assessments and assessments of possible spill-over effects on child outcomes.

Also, a better understanding is needed of the value of offering parenting support online, and under what circumstances this may be less or more beneficial for families than offering support in-person (Leijten, 2023). Growing up Together Online was developed in response to the Covid-19 pandemic, but its continued implementation (beyond the pandemic period) enabled participation of parents who would not have been able to participate in an in-person program, due to barriers related to transportation, child care and/or low income, or due to living in an remote/rural area where it is impossible to form in-person groups for an indicated parenting support program. Another advantage of the online format was overcoming the barrier of a lack of trained professionals. Because the program was primarily offered online, it allowed facilitators from family centers in different cities to co-lead groups, something that would not have been possible if the two facilitators had to be physically present at the same family center.

Last, Growing up Together Online is the result of a successful collaboration among public social services, a specialized NGO, private donors, national and international academics, as well as practitioners and parents involved in program co-creation and piloting. This collaboration facilitated social innovation, from overcoming barriers such as lack of knowledge or funding needed for introducing an online program, to barriers such as expectations about low-income parents’ capabilities to engage successfully with an online program. We hope this project will inspire similar collaborations in other countries and guide future work in this area.