Sibling-focused family prevention with Latinx siblings in middle childhood: A randomized clinical trial spanning the COVID-19 pandemic

Abstract

This preregistered randomized clinical trial examined the efficacy of a prevention program on parenting of siblings and sibling relationships among Latinxs. Participants were 272 sibling dyads (82.9% Mexican) with 5^th graders (M_age = 10.63; SD = .38; 51.8% female), their younger siblings in 1^st to 4^th grades (M_age = 8.18; SD = 1.06; 54.8% female), and their caregivers. Families were randomized to Siblings Are Special (SIBS; n = 161) or an alternative academic skills program (n = 111). Data were collected pre- and post-program (2018-2022). Significant effects were detected for sibling-focused parenting in the expected direction for authoritarian control and non-intervention in sibling conflicts, but there were no effects for sibling relationship quality. Implications of COVID-19 and future directions are discussed.

Siblings and their relationships have a profound and enduring role in children’s social, emotional, and cognitive development and in their long-term psychosocial and physical health (Feinberg et al., 2019; McHale et al., 2012). Yet, sibling influences continue to be overlooked in developmental (Holmes et al., 2024a; 2024b) and prevention science (Kramer et al., 2019). Latinx siblings, in particular, are central in one another’s daily lives: they have more siblings than Black or White youth and spend more time together, both in comparison to other youth and to time spent with parents and other household members (Ackert & Wikle, 2022; Updegraff et al., 2011). Latinx cultural values and practices that prioritize family members as sources of support and companionship (Cahill et al., 2021; Knight et al., 2010) also underscore siblings’ significant role in children’s socio-emotional functioning and their potential as a prevention target (Updegraff et al., 2016; Updegraff & Perez-Brena, 2023).

Existing prevention and intervention programs with Latinx families, which typically focus on single parent-youth subsystems or on parenting practices, have demonstrated positive impacts on youth (for a review, see Pineros-Leano et al., 2023). The success of these programs may be in part because they capitalize on the strong family-oriented values and practices that characterize Latinx families (Gonzales et al., 2012; Stein et al., 2014). Recognizing these familial strengths and the need to expand beyond an exclusive focus on parent-child relationships/parenting practices, we tested the efficacy of a socioemotional-based family prevention (Siblings are Special; SIBS). SIBS engages multiple siblings and caregivers and has two primary targets: (a) parenting of siblings and (b) siblings’ dyadic relationships (Feinberg et al., 2013; Updegraff et al., 2016). Building on a small-scale pilot feasibility study with Latinxs (Updegraff et al., 2016) and a prior trial with predominantly White families (Feinberg et al., 2013), this pre-registered randomized clinical trial (RCT) tested the efficacy of SIBS versus Academic Skills (an attention control condition focused on math and reading skills). This study takes a step toward addressing the underrepresentation of ethnoracially minoritized participants in rigorous experimental tests of programs (i.e., RCTs), as the largest discrepancy is for Latinxs who make up only 6% of clinical trial enrollees but are 16.3% of the U.S. population (Turner et al., 2022).

Our focus on middle childhood was guided by evidence that siblings are central and influential figures in children’s socio-emotional development in this period (Feinberg et al., 2019; McHale et al., 2012) and that children in middle childhood are highly amenable to social skills-focused programming (Greenberg et al., 1995). Siblings are also children’s most frequent companions outside of school (Dunifon et al., 2017), and sibling conflict increases across middle childhood (Kim et al., 2006). Given that sibling conflict is a salient stressor for parents and contributes to both parents’ and children’s diminished mental health (Buist et al., 2013; Feinberg et al., 2019), middle childhood is an optimal time for sibling-focused prevention.

Parenting of Siblings and Sibling Relationships as Primary Targets of Prevention

The first intervention target of SIBS is the parenting of siblings. Theoretical and empirical work has underscored parents’ role in fostering children’s relational and socioemotional development in the context of their peer relationships (Parke & Buriel, 2006). Theoretically, a key pathway is through parents’ direct involvement in their children’s relationships such as by guiding, coaching, and advising them in social interactions (Parke & Buriel, 2006). This theorized pathway has been extended to parenting of children’s sibling relationships and includes parents’ efforts to support and guide sibling relationships and help their children navigate conflicts and disagreements (Cahill et al., 2024; McHale et al., 2000; Ross & Lazinski, 2014; Siddiqui & Ross, 2004). Building on this literature, our focus on the parenting of siblings centered on three constructs: positive guidance, authoritarian control, and non-intervention.

Positive guidance reflects parents’ proactive efforts to foster supportive sibling interactions (Cahill et al., 2024; Chen, 2019; McHale et al., 2000; Ross & Lazinski, 2014) and includes strategies such as promoting listening and perspective taking, emphasizing the importance and value of sibling relationships, and praising children’s efforts to get along (Cahill et al., 2024; McHale et al., 2000). Research suggests that parents engage in positive guidance across multiple ethnoracial contexts and that these strategies are associated with more positive sibling relationships (Cahill et al., 2024; McHale et al., 2000; Updegraff et al., 2016). Further, experimental work focused on parent skills training specific to children’s sibling relationships revealed positive effects: When parents were trained to guide children through disputes, children were more likely to listen to one another, share their perspectives, and explain their feelings when compared to children of parents in a no-contact control condition (Siddiqui & Ross, 2004; Smith & Ross, 2007). Whereas positive guidance reflects parents’ proactive involvement, authoritarian control and non-intervention capture parents’ responses to sibling conflicts. Authoritarian control involves stepping in and resolving children’s conflict, typically by punishing or threatening punishment (Cahill et al., 2024; Chen, 2019; McHale et al., 2000; Updegraff et al., 2016). Such responses can reduce children’s opportunities to learn how to express their points of view, negotiate, and manage their own conflicts (e.g., Updegraff et al., 2016; Chen, 2019). Non-intervention is conceptualized as parents’ efforts to stay out of sibling conflicts and let siblings resolve them on their own (Cahill et al., 2024; McHale et al., 2000). Prior experimental work shows SIBS’ effects on reductions in parents’ authoritarian control (Updegraff et al., 2016) and increases in non-intervention (Feinberg et al., 2013) compared to parents in a no-contact control condition. Accordingly in this study, we hypothesized that, compared to Academic Skills parents, SIBS’ parents would report more positive guidance (Hypothesis 1; H1, preregistered), less authoritarian control (H2, preregistered), and more non-intervention (H3, not preregistered – see online supplement) at post-test. Potential differences in program effects for mothers versus fathers also were tested. Here we expected that mothers’ primary role in caregiving in Latinx families (Azmitia & Brown, 2002; Updegraff & Umaña-Taylor, 2015) and greater involvement in family-based prevention (Pineros-Leano et al., 2023) would lead to stronger effects for mothers.

The other primary target of SIBS is sibling relationship quality. Developmental theory underscores the significance of children’s relationships for their socio-emotional maturation (Hartup, 1989), and points to siblings as key socialization agents (Dunn, 1983; Feinberg et al., 2019; McHale et al., 2012). Within the domain of sibling relationship quality, SIBS was designed to promote more positive and less conflictual sibling relationships by providing opportunities for children to learn and practice-- in the context of their sibling relationships-- key socioemotional and relational competencies (e.g., emotion expression and regulation, perspective taking, constructive time-use). These skills are commonly taught in social skills programs that focus on peers (Bierman, 2004; Greenberg et al., 1995). In middle childhood, children spend most of their free time with their siblings (Dunifon et al., 2017), and these relationships are commonly characterized by emotional intensity and conflict (McHale et al., 2012; Feinberg et al., 2019). Thus, sibling relationships offer a highly motivating but challenging ecology for the development of children’s socio-emotional and relational skills.

A small body of work suggests that sibling relationships can be improved by fostering children’s socio-emotional and relational skills within this relational context. With sibling dyads from predominantly White families, Kramer and colleagues demonstrated the positive effects of a program that taught social and emotional skills to 4- to 8-year-old sibling pairs: Compared to a no-contact control group, sibling dyads randomized to the program demonstrated improvements in sibling warmth and children’s emotion regulation with siblings (Kramer et al., 2019). Sibling-focused interventions also have targeted children in out-of-home (foster) care, a group for whom preserving these family ties may be especially important (Kothari et al., 2017). Among sibling pairs (7 to 15 years of age) from both White and ethnoracial minoritized backgrounds, a 12-session sibling relationship intervention led to improved sibling relationship quality compared to a ‘services as usual’ control condition (Kothari et al., 2017). Further, prior trials comparing SIBS to no-attention control conditions found that parents (though not children) reported increases in sibling positivity in both White and Latinx families (Feinberg et al., 2013; Updegraff et al., 2016), and that sibling conflict declined in Latinx families (Updegraff et al., 2016). Accordingly, we hypothesized that parents and children in SIBS would report more positive (H4, pre-registered) and less negative (H5, pre-registered) sibling relationships at post-test compared to those in Academic Skills.

Advancing Sibling-focused Prevention

Sibling relationships have rarely been a focus of family-based prevention. The few such programs typically have targeted parents via skills training and education (Ross & Lazinski, 2014; Siddiqui & Ross, 2004) or worked directly with siblings to improve their relationships (e.g., Kennedy & Kramer, 2008; Kothari et al., 2017). Except for a program for siblings at elevated risk for conduct disorders in middle childhood (Bank & Snyder, 2004; Feinberg et al., 2019), SIBS is the only program to have a combined focus on parents and siblings simultaneously. In the afterschool component of SIBS, based in part on social learning principles (Bandura, 1977), activities were designed for siblings to work together, learn and practice skills with each other, and observe and model both their own siblings and other dyads within the group. Thus, the active ingredient in the afterschool portion of SIBS was the combination of socioemotional content and structure that facilitated learning and application of these skills within siblings’ dyadic relationships, with support from group facilitators.

SIBS was also intentional in its primary time commitment being the sibling sessions (i.e., 12 weekly afterschool sessions), with less time demands on parents (i.e., three family nights) relative to typical parenting programs (Pineros-Leano et al., 2023). The focus of the three family nights was to help parents navigate their role in children’s sibling relationships (online supplement, Table S1). Topics of the first family night were fostering close relationships, constructive time use, and emotion expression and regulation. The second family night focused on listening, negotiation, and conflict resolution, with the opportunity for parents to practice coaching strategies with sibling pairs. The final family night focused on team building, conflict resolution, and specifically helping parents decide when to step in versus allow children to resolve their own conflicts.

Comparing SIBS to an attention-control condition represents (to our knowledge) the first randomized test of a sibling-focused program that involves a contact- and time-equivalent control condition. Existing experimental designs have involved control conditions such as: wait list, services as usual, or providing parents in the control group with reading materials (e.g., a book about siblings). The control condition in the current study, Academic Skills, was designed to be equally appealing to parents and schools (given the context of program delivery), and feasible with the same age group targeted by SIBS. The focus on reading and math skills, however, required that children work in grade-level groups rather than with their siblings, given siblings’ age differences. The family nights, in contrast, were more similar in structure across conditions; nevertheless, they differed in content, as the Academic Skills family nights focused on reading, math, and homework strategies (online supplement, Table S2), whereas SIBS focused on relationship dynamics as described above. Despite these differences, the current design provides a more rigorous experimental test of SIBS versus a control condition compared to previous studies, given the equivalent dosage across conditions.

COVID-19 Impact on Program Efficacy

This clinical trial spanned the onset of the COVID-19 pandemic, an unprecedented period that substantially and suddenly altered children’s sibling relationships. Due to school closures, social distancing, and stay-at-home orders, siblings became many children’s primary peer-like companions in the early pandemic period (Browne et al., 2021; Whiteman et al., 2023). Only one study of pandemic effects, though, has focused specifically on Latinx sibling relationships (Sun et al., 2021). These findings suggested that Latinx siblings were important sources of support with relationships becoming more positive over time during the initial period of social isolation following school closures (Sun et al., 2021). More generally, the pandemic heightened stressors for some Latinx parents, who experienced disproportionate rates of COVID-19 infections, hospitalizations, and mortality as compared to their population percentage and to Whites (Basset et al., 2020; Majano et al., 2023). Furthermore, spikes in unemployment and increased financial difficulties were also pervasive among Latinxs (Noe-Bustamante et al., 2021). Given the timing of the pandemic and potential implications of social isolation and related stressors for family and sibling relationships and program delivery, COVID-19 onset was explored as a moderator of program effects.

The Current Study

Building on a pilot-feasibility study with Latinxs (Updegraff et al., 2016), this pre-registered RCT tested the effectiveness of SIBS versus Academic Skills in a sample of Latinx siblings in middle childhood and their parents. Compared to Academic Skills and controlling for T1 levels of dependent variables, we hypothesized that SIBS participants would report more positive guidance (H1, preregistered) and lower authoritarian control (H2, preregistered) at post-test. Although non-intervention in sibling conflicts was not pre-registered, it was hypothesized that SIBS parents would report more non-intervention (H3) based on prior literature (Feinberg et al., 2013). SIBS parents and siblings also were expected to report more sibling positivity (H4, pre-registered) and less negativity (H5, pre-registered) at post-test versus Academic Skills, controlling for prior levels. We also explored moderation by COVID-19 timing. Given our clustered design with data from mothers, fathers, and two siblings in each family, we also tested for parent (mother vs father) and sibling (older vs younger) differences in program effects as indicated in our pre-registration. Sibling differences were not anticipated, but because program effects could be stronger for mothers than fathers given gender-differentiated roles in Latinx families (Updegraff & Umaña-Taylor, 2015) we tested this possibility. Pre-registered covariates included parent education, parent immigrant status, and sibling characteristics (age spacing, sibling dyad sex constellation, number of siblings), consistent with prior work (Updegraff et al., 2016). Given sample variation in family configurations, we also accounted for whether caregivers were biological parents of children and whether siblings were biologically related.

Method

Participants

Participants were f272 sibling dyads and their caregivers from 270 Latinx families as two families included two eligible dyads. Families were recruited from 11 elementary schools in the southwest U.S. that served at least 60% Latinx students (range 60% to 89%). All schools held schoolwide Title I designations (i.e., > 40% of students qualified for federal free/reduced lunch), with the range for participating schools from 78% to 97%. Eligibility criteria were as follows: Latinx 5^th grader and a sibling (1^st to 4^th grade) attending the same school and living together for at least three years; neither child having significant disabilities that would prevent participation; a parent/guardian able to participate in English or Spanish; availability after school one day/week, and able to provide transportation home or allow staff to transport via rideshare service. This study was pre-registered at clinicaltrials.gov (NCT03706014); any deviations from pre-registration are noted in the online supplemental materials.

For all three cohorts (C1 in 2018-19, C2 in 2019-20, and C3 in 2021-22), program materials in Spanish and English were sent home with students and mailed to parents of Latinx 5^th graders with a younger sibling in 1^st to 4^th grade (N = 563 families). Bilingual staff successfully contacted 91% of families by phone to assess eligibility and interest. The majority were eligible (72%), and 272 sibling dyads in 270 families enrolled and were randomized (online supplement Figure S1). There were no cohort differences in the percentage of eligible families who agreed versus refused to participate, χ² (2) = 1.19, p = .55.

Older and younger siblings averaged 10.63 years (SD = .38) and 8.18 years (SD = 1.06), respectively, and were identified as girls (53.3%) or boys (46.7%) by parents. Siblings averaged 2.44 years apart in age (SD = 1.03) and slightly more were same-sex (51.5%) than mixed-sex dyads (48.5%). Older and younger siblings were primarily born in the U.S. (92.4% and 93.8%, respectively) and completed surveys in English (95.5% and 94.2%, respectively). Children lived in households with an average of 2.58 siblings (SD = 1.31) and 6.23 people (SD = 1.88); families reported (74% reporting) annual pre-tax household income that averaged $33,434 (SD = 22,204); and 70.4% of families included a male and female caregiver who were married or living together. Characteristics of participating female (n = 269) and male caregivers (n = 120), referred to hereafter as mothers and fathers, are in Table 1.

Table 1.

Sociodemographic Characteristics of Mothers and Fathers

	Mothers		Fathers
	M	SD	M	SD
Age in Years	36.72	6.73	38.60	10.72
	n = 267	%	n = 119	%
Primary Caregiver Relationship to Siblings
Biological Parent	255	95.5%	95	80%
Biological Parent/Stepparent1	4	1.5%	5	4.2%
Adoptive Parents	2	1%	2	1.6%
Stepparents	0	0%	13	11%
Grandparent	6	2%	2	1.6%
Older Sibling	0	0%	1	0.8%
Parents’ Romantic Partner	0	0%	1	0.8%
Education Level 2
Less than high school	148	56%	67	56%
High school degree or GED	78	29%	38	32%
Post-secondary training or degree	40	15%	12	10%
Nativity Status
Born outside the U.S.	186	70%	91	76%
Born in the U.S.	81	30%	28	24%
Interview Language Preference
Spanish	179	66%	83	70%
English	91	34%	36	30%
Ethnoracial Origin 3
Mexican	222	83.1%	97	82%
Guatemalan	12	4.5%	8	6.7%
Salvadoran	7	2.62%	1	0.8%
Puerto Rican	2	0.75%	0	0
Honduran	1	0.37%	2	1.7%
Latinx subgroup not specified	5	1.87%	2	1.7%
White (not Latinx)	9	3.37%	7	5.9%
American Indian/Alaskan Native (not Latinx)	2	0.75%	0	0
Asian (not Latinx)	2	0.75%	0	0
Hawaiian or Pacific Islander (not Latinx)	1	0.37%	0	0
Multi-ethnoracial4	4	1.5%	1	0.8%

¹Parents are biologically related to one sibling and the stepparent of the other sibling.

²One mother and two fathers were missing education.

³One father was missing ethnicity.

⁴Multi-ethnoracial caregivers were Mexican-Venezuelan, Mexican-Guatemalan, White-Black, and Latinx-Native American.

Procedures

A home visit was conducted with each family by two staff members. When parents preferred Spanish, at least one home visitor was fluent in Spanish. The project was briefly described, consent and assent forms were read aloud, and parents’ questions were answered prior to obtaining participants’ signatures. Interviewers then completed a demographic questionnaire by reading each item and entering parents’ responses on paper surveys. For both pre- and post-test, each parent completed a phone interview lasting 45-60 minutes in English or Spanish. Interviewers read items aloud and entered participants’ responses to account for variability in reading levels. Surveys were conducted with children in afterschool sessions lasting 45 minutes for older siblings and 35 minutes for younger siblings. Surveys were read aloud to all students, with interviewers recording responses for younger children (1^st to 3^rd graders) and children who needed assistance. Older children entered their own responses on paper surveys. An exception was the C2 post-test: 34.6% of children completed post-test during afterschool sessions and 65.4% completed post-test over the phone with interviewers reading survey questions and entering child responses into Qualtrics surveys due to COVID-19 school closures. Pre-test data were collected prior to randomization. Post-test data were collected one week after program completion for children and within 2-4 weeks for most parents by project staff (see online supplement for protocol for assigning interviewers for post-test). Honoraria were $20 for the home visit, $35 for pre-test parent interviews, $45 for post-test parent interviews, and goody bags for children at pre- and post-test ($3-$5/bag). All procedures were approved by the Human Subjects Review Board at Arizona State University and the school district’s Research Board.

Program Assignment and Structure.

Within each school, families were randomly assigned to SIBS or Academic Skills conditions prior to the first program session, with the goal of randomizing a slightly higher percentage to SIBS (60%) versus Academic Skills (40%) to increase the statistical power to test moderators specific to SIBS (not a focus of this paper). Each SIBS group (n = 44) had up to four sibling dyads (M = 3.62 dyads; SD = .48) with at least two group leaders (one with Spanish fluency) who had a minimum of two years of post-secondary education. Group leaders participated in 15-18 hours of training on the curriculum. Academic Skills groups (n = 25) maintained the same ratio of 1 adult per 4 children and ranged from four to nine dyads (M = 4.32 dyads; SD = 1.50) with 2-5 group leaders in each. Group leader training involved six hours on the Academic Skills curriculum, which included games that engaged math and reading skills (with three levels of difficulty for 1^st-2 graders, 3^rd-4^th graders, and 5^th graders, respectively). Group leaders in each condition met with project coordinators weekly (separately for the two conditions). Program manuals and all program materials were provided (see supplement Tables S1 and S2 for SIBS and Academic Skills curriculum outlines, respectively). Program conditions were equivalent in the number and length of afterschool sessions (12 sessions, 90 minutes each) and family nights (3 sessions, 2 hours each), but differed in structure. In SIBS, the focus was on the sibling dyad, with the goal of fostering positive sibling relationships by encouraging positive and constructive interactions and learning and applying socioemotional strategies (e.g., listening, perspective-taking) with siblings. The Academic Skills condition, in contrast, focused on age/grade appropriate reading and math skills. Although siblings attended the program together, activities and games were conducted with children grouped by grade level. Family nights were similar in structure but differed in content. Both conditions included dinner (30 min), a breakout session with parents and children separately (30-45 min), a presentation of social/academic skills to parents by children (15 min), and a family activity and wrap up (30 min). At each family night, a raffle was held for a $25 store gift card. Families received money for gasoline ($5 in C1 and C2; $10 in C3 due to increased fuel costs) or rideshare services, free childcare, food, and/or materials. All sessions took place in the participating elementary schools.

Program Implementation and Fidelity.

Program attendance was high, indicating success in engaging participants. Older and younger siblings averaged 9.60 (SD = 3.10) and 9.68 (SD = 3.01) sessions, respectively, and 77.21% of older and 78.31% of younger siblings attended at least 8 of 12 sessions. Average family night attendance was 2.41 (SD = .77) for mothers and 1.84 for fathers (SD = .74) across the three family nights. Attendance did not differ across conditions for family night, t (263) = .82, p = .411, or older and younger siblings’ session attendance, t (270) = 1.35, p = .177, and t (270) = 1.42, p = .157, respectively. There also were no significant differences in afterschool session attendance, F (2,269) = .23, p = .79, or family night attendance F (2, 262) = .09, p = .91, across cohorts. To evaluate fidelity, 30% of SIBS sessions were videotaped (see online supplement for details).

Cultural Adaptation.

SIBS was originally developed to be relevant to White, Black, and Latinx families. Surface structure adaptations for Latinxs (i.e., adaptations to the method of delivery/materials, but not core content) were tested in a pilot study and included delivering the program with a predominantly Latinx and bilingual staff, translating materials to Spanish, and ensuring cultural relevance of examples/graphics (Updegraff et al., 2016).

COVID-19 Impact.

C1 took place prior to the onset of COVID-19. The onset of the pandemic occurred during the implementation year for C2 (2019-2020). As a result, we did not complete Session 12 and Family Night #3 at one school due to state-wide school closures on March 16, 2020. In addition, about two-thirds of student post-program surveys were collected over the phone rather than in afterschool sessions due to statewide shelter-in-place orders, which began on March 31, 2020 (Executive Order 2020-18) and extended through the end of the school year/data collection period. No programming was conducted during the 2020-2021 school year due to COVID-19 restrictions. For C3 (2021-2022) pre- and post-test and program delivery, project staff wore face coverings due to COVID-19 human subjects’ requirements; whether face coverings were optional or required for participants depended on each participating school’s infection rate and varied across sessions. Social distancing, enhanced cleaning procedures, and meals at family nights were altered to reduce virus spread. In Arizona, this study’s location, social isolation due to state-level executive orders were most restrictive in March-July, 2020.

Measures

Measures were piloted in prior work with English and Spanish speaking Latinx samples and were translated via forward and back translation with decentering to ensure equivalence across languages (Knight et al., 2009). All scale scores were created by averaging items such that higher scores indicated higher levels of a construct.

Parenting of Siblings.

This measure included three subscales originally adapted from McHale et al. (2000) and the measurement properties were established in prior work with this sample via Bayesian exploratory and confirmatory factor analyses (Cahill et al., 2024). Positive guidance includes ten items (e.g., “How often do you make suggestions to your children about how they can get along better?”); authoritarian control has five items (e.g., “How often do you punish your children when they are fighting or bothering one another?”), and non-intervention has four items (e.g., “When your children are having a disagreement, how often do you ignore them on purpose so that they can solve it on their own?”). Items are rated on a 5-point scale from 1 (Never) to 5 (Almost always). Alpha coefficients ranged from .83 to .84 for mothers’ and .88 to .91 for fathers’ positive guidance, .75 to .77 for mothers’ and .67 to .76 for fathers’ authoritarian control, and .67 to .71 for mothers’ and .63 to .68 for fathers’ non-intervention across timepoints.

Sibling Relationship.

Mothers and fathers reported on the frequency each sibling directed prosocial behaviors toward the other sibling (counterbalanced) with a measure developed by Stormshak et al. (1996) and adapted for sibling relationships (Updegraff et al., 2016); this construct is referred to hereafter as positivity. An example item of this 6-item scale is “[older sibling’s name] helps, shares, and cooperates with [younger sibling’s name].” Parents also reported on each sibling’s negativity with a 7-item scale (Stocker & McHale, 1992); a sample item is “How often does [older sibling’s name] start fights or cause trouble for [younger sibling’s name]?” Items were rated on a Likert scale ranging from 1 (Almost never) to 6 (Almost always). Alpha coefficients for mothers’ and fathers’ reports of older and younger sibling positivity and negativity across timepoints ranged from .81 to .91.

Older and younger siblings reported on their intimacy (referred to as positivity hereafter) using an 8-item measure (Blyth et al., 1982; Updegraff et al., 2002). Negativity was assessed via the 7-item subscale from the Sibling Relationship Inventory (Stocker & McHale, 1992). Items such as “How much do you go to your sibling for advice and support?” (positivity) and “How often do you feel mad or angry at your sibling?” (negativity) were rated on a 5-point scale ranging from 1 (Not at all) to 5 (Very much). Alpha coefficients ranged from .79 to .81 for older and younger siblings’ positivity and from .85 to .89 for both siblings’ negativity across time.

Covariates and Moderators.

Parent education was coded as 0 = high school degree/GED or 1 = no high school degree/GED, and parent nativity was coded as 0 = U.S.-born or 1 = born outside the U.S. Each parent was coded as biologically related to one or both children (0) versus not biologically related (1). Parents reported the number of siblings in the home. Sibling pairs were coded as full biological siblings (0) or not (1). Sibling age spacing was calculated as older minus younger siblings’ age in years. Sibling dyad sex constellation was coded as 0 = same-sex and 1 = mixed-sex. Siblings were coded as older (0) and younger (1). Parents were coded as mother (0) or father (1). Cohort was coded as 1, 2, or 3 based on year of enrollment. COVID-19 onset was coded for each family member based on whether post-test was completed prior to COVID-19 onset/school closure on March 16, 2020 (0) or after (1).

Missing Data

Data completion rates for each family member at pre- and post-test are presented in online supplement Figure S1. There were no missing data for family demographics, except parent education (< 1%; n = 3) and ethnicity (< 1%; n = 1). Child survey data were missing from both siblings at pre-test in 3% (n = 7) and at post-test in 7% (n = 19) of families, and parent survey data were missing from 7% of families at pre- and post-test (n = 19). There were no significant differences in program condition, χ² (1) = .15, p = .70, mother education, χ² (1) = .10, p = .75, father education, χ² (1) = 3.60, p = .058, father nativity, χ² (1) = .79, p = .38, and family income, t (199) = 1.13, p = .26, between families who participated in post-test (n = 256) versus those who did not (entire family; n = 16). A significant difference emerged for mother nativity: families were more likely to participate at post-test when mothers were born outside the U.S., χ² (1) = 5.33, p = .02. Mothers’ positive guidance was higher at pre-test for those who dropped out (M = 4.53; SD = .32) versus those who did not (M = 4.27; SD = .53), t (241) = 2.32, p = .02. No other significant differences emerged in mothers’ or fathers’ reports of sibling-focused parenting or in parents’ or siblings’ reports of sibling positivity and negativity (all t-values < 1.00).

Analytic Plan

Preliminary analyses included descriptives for all study variables by program condition. Randomization (SIBS vs Academic Skills), cohort (1, 2, 3), and COVID-19 differences (pre versus during) were examined via chi-squared analyses for categorical variables and with t-tests and ANOVAs for continuous variables at pre-test. Given the clustered nature of the data (i.e., 544 children nested in 270 families, and 270 families nested in 69 intervention groups and 11 schools), intraclass correlations were computed and then design effects were examined (< 2.00 considered negligible; Muthén & Satorra, 1995).

Using an intent-to-treat design, hypotheses were tested using multilevel modeling (MLM) in SAS 9.4 with parents and siblings clustered within families and the maximum likelihood estimator (ML) for missing data. For the models focused on parenting of siblings (i.e., positive guidance, authoritarian control, non-intervention), covariates included parent education and nativity, biological relatedness of parents to children, sibling dyad sex constellation, COVID-19 timing, parent (mother versus father), and T1 levels of dependent variables; condition (SIBS = 0; Academic Skills = 1) was the independent variable. We also tested for moderation by parent (mother versus father) and COVID-19 timing. For models focused on sibling relationships (reported by parents and siblings), covariates and moderators were the same, except we controlled for biological relatedness of the sibling dyad (rather than the parent-child dyad), sibling age spacing, and number of siblings in the home; we also tested for sibling (older versus younger) differences in program effects. Effect sizes (Cohen’s d) were calculated for t-tests (independent groups and paired) and ANOVAS (Cohen, 1988), and Hedges g was calculated for MLMs (Hedges, 2007); consistent with Cohen (1988), we classified effect sizes as small (.2), medium (.5), and large (.8).

Post-hoc sensitivity analyses were not preregistered. For any dependent variables with no significant program effects, we tested for overall changes across time using MLMs with the same covariates as the primary analyses and an effect for time (0 = pre-test, 1 = post-test). Next, we explored moderation of program condition by sibling dyad characteristics (sibling dyad sex constellation; sibling age spacing). Although prior tests of SIBS did not reveal moderation by dyad characteristics, we explored moderation here because sample sizes were smaller in prior trials reducing power to detect subgroup differences (Updegraff et al., 2016). Next, models were also estimated separately for mothers and fathers to check the robustness of missing data estimation given that our design included fewer fathers than mothers.

Results

Preliminary Analyses

Bivariate correlations between study variables and means and standard deviations by program condition are in Tables 2 and 3 (supplemental Tables S3 and S4 report correlations by individual reporter for sibling relationships). Randomization checks revealed no significant pre-test differences in covariates or study variables. For cohort differences, there were two effects: mothers reported more authoritarian control in C1 (M = 3.43; SD = .85) versus C3 (M = 2.98; SD = .84), F (2, 240) = 4.88, p = .01, d = .53, and siblings reported more negativity in C3 (M = 2.06; SD = .79) versus C2 (M = 1.79; SD = .62), F (2, 262) = 3.51, p = .03, d = .38. Cohort was initially included as a covariate but removed as it was not significant in any model. The only significant pre-test difference by COVID-19 was fathers’ nativity, χ² (1) = 5.42, p = .02, with more immigrant fathers participating pre-covid. Fathers’ nativity was included as a covariate.

Table 2.

Bivariate Correlations and Means (Standard Deviations) for Parenting of Sibling Variables and Covariates

	1	2	3	4	5	6	7	8	9	10	11
1. Pos Guid T1	--	−.16	.02	.74**	−.06	.01	−.24**	.28**	−.12	.18	.07
2. Auth Ctr T1	−.08	--	.08	−.23*	.61**	.06	.09	−.23*	.15	−.04	.07
3. Non-inv T1	.18	.19	--	−.08	.01	.49**	.00	−.05	.01	.15	.01
4. Pos Guid T2	.58**	−.05*	.10	--	−.02	−.03	−.32**	.32**	−.14	.07	.10
5. Auth Ctr T2	.09	.64**	.08	.12	--	.16	.03	−.10	.07	−.10	−.10
6. Non-inv T2	.15	.12	.48**	.17	.12	--	.08	−.08	.01	.08	−.10
7. Parent Educ.	−.07**	.05	.03	−.10**	−.01	.13	--	−.37**	.17**	−.08	.05
8. Parent Nativ.	.04**	−.17*	.00	.03**	−.12	−.08	−.20**	--	−.30**	−.08	−.06
9. Bio Parent	−.08	.15	−.08	−.05	.12	.02	.04**	−.08**	--	.00	.00
10. Sib Dyad	.01	.05	.09	−.01	.03	−.03	−.09	−.01	.01	--	.00
11. Prog Cond	.03	.01	.00	.12	.12	−.15	−.04	.03	−.01	.00	--
Mother-report M (SD) by Program Condition
SIBS	4.27 (.52)	3.14 (.82)	2.96 (.76)	4.12 (.52)	2.88 (.73)	3.22 (.62)	1.46 (.50)	.69 (.47)	.03 (.18)	.48 (.50)	--
Academic Skills	4.30 (.54)	3.16 (.82)	2.96 (.70)	4.25 (.52)	3.06 (.79)	3.01 (.72)	1.42 (.50)	.71 (.46)	.03 (.16)	.49 (.50)	--
Father-report M (SD) by Program Condition
SIBS	4.05 (.60)	2.71 (.71)	2.85 (.77)	4.00 (.57)	2.75 (.75)	2.94 (.60)	1.36 (.48)	.82 (.39)	.16 (.37)	.48 (.50)	--
Academic Skills	4.13 (.66)	2.81 (.78)	2.86 (.81)	4.13 (.72)	2.59 (.84)	2.80 (.81)	1.42 (.50)	.76 (.43)	.16 (.37)	.49 (.50)	--

Note. Mothers reported study variables above the diagonal (n ranges from 245–272 across variables), fathers reported study variables below the diagonal (n ranges from 98–272 across variables). Pos Guid = Positive guidance; Auth Ctr = Authoritarian control; Non-inv = Non-intervention; T1 = pre-test and T2 = post-test; Parent Educ. = Parent education coded 0 = less than high school degree and 1 = high school degree or higher; Parent Nativ. = Parent nativity coded 0 = U.S. born and 1 = born outside the U.S; Bio Parent = Biological parent of one or both children (0) versus not biological parent (1); Sib Dyad = Sibling dyad sex constellation (0 = same-sex; 1 = mixed-sex); Prog Cond = Program condition (0 = SIBS Intervention; 1 = Academic Skills control).

^*p < .05;

^**p < .01.

Table 3.

Bivariate Correlations and Means (Standard Deviations) for Mothers’, Fathers’, and Siblings’ Ratings of Sibling Relationships and Covariates

	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16
1. Prosoc T1 (M)	--
2. Prosoc T1 (F)	.45**	--
3. Int T1 (S)	.29**	.10	--
4. Neg T1 (M)	−.54**	−.34**	−.18**	--
5. Neg T1 (F)	−.41**	−.57**	−.08	.56**	--
6. Neg T1 (S)	−.30**	−.14	−.38**	.31**	.28**	--
7. Prosoc T2 (M)	.63**	.22*	.31**	−.45**	−.41**	−.32**	--
8. Prosoc T2 (F)	.38**	.68**	.29**	−.45**	−.68**	−.33**	.38**	--
9. Int T2 (S)	.27**	.20*	.57**	−.11	−.18	−.34**	.35**	.29**	--
10. Neg T2 (M)	−.36**	−.19*	−.16**	.71**	.46**	.35**	−.47**	−.32**	−.17**	--
11. Neg T2 (F)	−.38**	−.43**	−.19*	.57**	.79**	.33**	−.46**	−.58**	−.23**	.52**	--
12. Neg T2 (S)	−.22**	−.07	−.29**	.32**	.33**	.66**	−.36**	−.33**	−.46**	.34**	.40**	--
13. Bio Sibs	−.04	−.08	−.05	.09	.03	−.02	−.08	−.21*	−.03	.08	.03	.04	--
14. Sib Dyad	.07	.24**	−.13*	−.10	−.24**	−.04	−.06	.14	−.08	.04	−.16	−.00	.05	--
15. Age Spacing	.05	.02	.09	.03	.00	.10	.04	.05	.12	.06	−.01	.10	.04	−.02	--
16. # of Siblings	.06	−.12	.01	−.06	.05	−.01	−.04	−.15	−.05	−.06	−.04	.01	.02	−.06	−.08	--
17. Prog Cond	.08	.06	.08	−.10	.07	.01	−.05	.10	.08	−.05	.06	−.08	−.11	.00	−.02	−.05
M (SD) by Program Condition
SIBS	4.13 (.93)	4.01 (.89)	3.23 (.64)	2.30 (.81)	1.95 (.57)	1.96 (.71)	4.20 (.91)	4.02 (.73)	3.16 (.65)	2.14 (.75)	1.85 (.50)	1.97 (.67)	.12 (.32)	.48 (.50)	2.46 (1.02)	--
Academic Skills	4.28 (.93)	4.12 (.99)	3.35 (.73)	2.14 (.75)	2.05 (.85)	1.98 (.79)	4.11 (.86)	4.20 (1.03)	3.28 (.77)	2.06 (.69)	1.93 (.76)	1.85 (.81)	.05 (.23)	.49 (.50)	2.42 (1.06)	--

Note. (M) = Mother report; (F) = Father report; (S) = Sibling report; T1 = pre-test and T2 = post-test; Prosoc = Sibling prosocial behavior; Int = Sibling intimacy; Neg = Sibling negativity; Bio Sibs = Biologically related siblings coded as 0 versus other pairs coded as 1; Sib Dyad = Sibling dyad sex constellation (0 = same-sex; 1 = mixed-sex); Age Spacing = Older sibling age minus younger sibling age in years; # of Siblings = Number of siblings in home; Prog Cond = Program condition (0 = SIBS Intervention; 1 = Academic Skills control); All sibling relationship means are averaged across mothers’ and fathers’ ratings of older and younger siblings and across older and younger siblings’ reports. Sample sizes ranged from 102–272 across reporters.

^*p < .05;

^**p < .01.

Across outcomes, ICCs ranged from .09 to .50 for family, .02 to .08 for group, and .01 to .02 for school levels that converged. Nonconvergence may be due to the small number of schools and low clustering. As the highest group and school level ICC was .08 and all design effects were less than 2.00, we only accounted for family level clustering in the analyses.

Program Effects

The final models are shown in Table 4 (Models 1- 7), including covariates, main effects, and significant interactions. Non-significant interactions were trimmed as their inclusion increases bias in standard errors (Cohen et al., 2013). In the text, we report effect sizes (only calculated for program effects) and results of follow-up tests for significant interactions.

Table 4.

Results from Multilevel Models with Intervention Condition Predicting Parenting of Siblings and Sibling Relationship Dynamics

	Parenting of Siblings T2			Sibling Relationship Dynamics T2
Fixed effects	Model 1: PR Positive Guidance γ (SE)	Model 2: PR Authoritarian Control γ (SE)	Model 3: PR Non-intervention γ (SE)	Model 4: PR Sibling Prosocial γ (SE)	Model 5: YR Sibling Intimacy γ (SE)	Model 6: PR Sibling Negativity γ (SE)	Model 7: YR Sibling Negativity γ (SE)
Intercept	1.73 (0.15)***	1.41 (0.14)***	2.05 (0.13)***	2.51 (0.21)***	1.56 (0.21)***	1.09 (0.14)***	0.91 (0.18)***
Covariates and Moderators
T1 report of outcome	0.57 (0.03)***	0.49 (0.03)***	0.39 (0.03)***	0.43 (0.03)***	0.49 (0.04)***	0.56 (0.03)***	0.51 (0.04)***
Parent nativity	0.03 (0.05)	−0.07 (0.07)	−0.02 (0.07)	0.14 (0.09)	−0.04 (0.08)	−0.28 (0.06)***	−0.07 (0.08)
Parent education	−0.07 (0.04)	0.06 (0.05)	−0.00 (0.05)	−0.18 (0.07)*	−0.07 (0.08)	0.03 (0.05)	0.06 (0.07)
COVID-19 timing	−0.05 (0.05)	−0.12 (0.07)	−0.04 (0.06)	0.13 (0.08)	−0.05 (0.08)	−0.01 (0.05)	−0.06 (0.07)
Biological relatedness	−0.03 (0.08)	−0.01 (0.10)	0.02 (0.10)	−0.15 (0.14)	−0.02 (0.13)	−0.01 (0.09)	0.06 (0.12)
Parent (Mother vs Father)	−0.04 (0.03)	0.09 (0.05)	−0.16 (0.04)***	−0.21 (0.08)**	---	−0.06 (0.05)	---
Sibling dyad constellation	−0.01 (0.05)	−0.02 (0.07)	−0.01 (0.07)	−0.08 (0.08)	−0.04 (0.07)	0.08 (0.05)	0.03 (0.07)
Sibling age spacing	---	---	---	0.01 (0.04)	0.06 (0.04)	0.00 (0.03)	0.03 (0.03)
Number of siblings	---	---	---	−0.06 (0.03)	−0.04 (0.03)	−0.03 (0.02)	0.01 (0.03)
Sibling (Older vs Younger)	---	---	---	0.05 (0.04)	0.13 (0.06)*	−0.03 (0.04)	−0.03 (0.06)
Predictor
Program condition	0.12 (0.05)*	0.23 (0.07)**	−0.14 (0.07)*	−0.17 (0.09)	0.06 (0.08)	−0.00 (0.05)	−0.14 (0.07)
Interactions
Program X Parent	---	−0.39 (0.07)***	---	0.28 (0.11)*	---	---	---

Note. Unstandardized coefficients (γ) with standard error (SE) are shown for main effects and significant interactions only. PR = Parent-reported outcome; YR = Youth-reported outcome; T1 = pre-test and T2 = post-test; Parent nativity coded 0 = U.S. born and 1 = born in the U.S.; Parent education coded 0 = less than high school degree and 1 = high school degree or higher; COVID-19 timing coded 0 = data collected before COVID-19 and 1 = data collected during COVID-19; Biological relatedness for parent outcomes coded 0 = biological parent and 1 = non-biological parent; Biological relatedness for sibling outcomes coded 0 = biological siblings and 1 = non biological siblings; Parent coded 0 = mother and 1 = father; Sibling dyad sex constellation coded 0 = same-sex and 1 = mixed-sex; Sibling age spacing = Older sibling age minus younger sibling age in years; Number of siblings = Number of siblings in home; Sibling birth order coded 0 = older sibling and 1 = younger sibling; Program condition coded 0 = SIBS Intervention and 1 = Academic Skills control. Dashes indicate not included in model. For YR outcomes, mother report nativity and education were included as covariates due to more data available from mothers than fathers.

^*p < .05;

^**p < .01;

^***p < .001.

Parenting of Siblings.

The program effect was significant for positive guidance (Table 4, Model 1) such that SIBS parents reported lower positive guidance than Academic Skills parents at post-test (g = .26), contrary to preregistered H1. For authoritarian control, there was a significant program x parent interaction (Model 2), with a medium effect size (g = .65). Follow-up analyses probing the interaction were consistent with H2 (preregistered) for mothers: SIBS mothers reported less authoritarian control at post-test relative to Academic Skills mothers (γ = 0.23, SE = 0.08, p = .002); fathers did not differ by program condition (γ = −0.16, SE = 0.08, p = .079). For non-intervention, a significant program effect (Model 3) showed that parents in SIBS reported more non-intervention at T2 than parents in Academic Skills (g = .24), consistent with H3 (not preregistered). A significant parent effect also emerged for non-intervention such that mothers reported higher levels than fathers.

Sibling Relationship Quality.

Testing H4 (preregistered) we examined parent- and child-reported positivity. For parent reports (Model 4), a parent effect was qualified by a significant parent X program condition interaction (g = .36). Follow-ups to this interaction, however, revealed no differences in parent-rated positivity across SIBS and Academic Skills for mothers (γ = −0.16, SE = 0.09, p = .07) and fathers (γ = 0.11, SE = 0.12, p = .33). In addition, parents without a high school degree reported less positivity than those with a high school degree. For youth-reported sibling positivity (Model 5; also H4), younger siblings reported significantly more positivity than older siblings, but there was no significant program effect. For H5 (preregistered), we examined both parent and sibling reports of negativity and found no significant program effects. For covariates, U.S.-born parents reported significantly more sibling negativity than immigrant parents (Model 6).

Post-Hoc Sensitivity Analyses

Changes in Sibling Positivity and Negativity.

As there was no evidence of program effects on sibling relationship quality, we estimated two additional MLMs, testing whether sibling positivity and negativity changed across time (regardless of program condition). All covariates were the same as the primary analyses. For sibling positivity, there was no significant time effect for parent or child reports. For sibling negativity, a time effect revealed a significant decline from pre- to post-test in parents’ reports of sibling negativity, γ = −0.15, SE = 0.03, p < .001, g = .19, that did not differ by parent, and was not found in child reports. In addition, parents reported more positivity among siblings recruited during COVID-19 versus prior to COVID-19, γ = −0.28, SE = 0.08, p < .001. In both negativity models, COVID-19 timing also was significant such that sibling negativity was lower during COVID-19 versus pre-COVID-19 for parents’, γ = −0.23, SE = 0.07, p < .001, and siblings’, γ = −0.18, SE = 0.08, p = .035, reports.

Variation in Program Effects by Sibling Dyad Characteristics.

We tested whether program effects varied by sibling dyad sex constellation for all outcomes and by sibling age spacing for sibling relationship outcomes by adding program condition by dyad constellation/age spacing interactions to the MLMs. There were no significant interactions.

Discussion

Siblings have been neglected in developmental science (Holmes et al., 2024a; 2024b) despite their critical role in children’s socioemotional development and adjustment (Feinberg et al., 2019; McHale et al., 2012). Studying Latinx children and parents, a group that is underrepresented in tests of evidence-based prevention (Turner et al., 2022), we examined the effectiveness of SIBS on the primary intervention targets: parenting of siblings (positive guidance, authoritarian control, and non-intervention in conflicts) and sibling relationship quality (positivity and negativity). This pre-registered RCT provided a stringent test of program effects with intent-to-treat analyses, successful randomization, low attrition, and evidence that the program was implemented as intended. Furthermore, high rates of attendance at afterschool sessions and family nights provide evidence of the value and appeal of this approach to family-based prevention with low-income Latinx families. Although significant effects were fewer than predicted, this was also the first sibling-focused program (to our knowledge) to use an RCT with an attention control condition, increasing confidence in the program effects we detected.

Primary Intervention Targets: Parenting of Siblings and Sibling Relationships

Parents play a key role in shaping sibling relationships in middle childhood, including fostering positive interactions and managing negative exchanges that have documented implications for youth development and adjustment (Feinberg et al., 2019; McHale et al., 2000). Our findings revealed that the combination of weekly sessions with siblings and family nights (which allowed parents to learn about and practice strategies for supporting siblings) resulted in small to modest program effects in the expected direction for two dimensions of sibling-focused parenting. For conflict management in particular—a major stressor for parents— program effects were in expected directions (i.e., for authoritarian control and non-intervention). In both primary and post-hoc analyses and supporting H2, SIBS mothers reported significantly lower levels of authoritarian control at post-test versus Academic Skills mothers, controlling for pre-test levels and background characteristics. This medium effect size was similar in magnitude to a feasibility study with Latinxs (Updegraff et al., 2016). Lower authoritarian control has the potential to benefit siblings by promoting autonomy in youth’s handling of sibling conflicts as well as more opportunities to practice resolving conflicts independently of parents (Updegraff et al., 2016). Longer-term follow-up will be needed to determine whether there are downstream impacts of parents’ less authoritarian approach to sibling conflicts, such as greater autonomy in managing interpersonal conflicts.

Turning to (deliberate) non-intervention in sibling conflicts, compared to Academic Skills parents, SIBS parents reported higher levels at post-test, consistent with H3. The lack of a parent by program interaction effect for non-intervention in the clustered models combined with the pattern from the post-hoc analyses, suggested consistency of this effect across mothers and fathers. Although this non-intervention effect was not pre-registered, findings were consistent with SIBS program effects in a sample of predominantly White mothers and fathers (Feinberg et al., 2013). Together, these findings suggest that SIBS parents were more likely, post intervention, to support their children’s efforts to manage their conflicts, stepping in less often to resolve those conflicts or to punish children for fighting. As children in SIBS learn to express their points of view, negotiate, and come up with their own solutions to problems, parents may feel more comfortable stepping back and allowing siblings to put their skills into action (Feinberg et al., 2019). It also may be that conflicts between SIBS children were less intense and demanding of parental intervention. Another possibility, also not directly measured here, is that SIBS children less often sought out their parents for assistance with conflicts. These results are consistent with experimental evidence of the benefits of teaching parents about sibling relationship management (Ross & Lazinski, 2014; Siddiqui & Ross, 2004) and children about socioemotional skills to apply in the sibling relationship context (Kothari et al., 2017; Kramer et al., 2019).

Turning to positive guidance (H1), a small-sized program effect revealed that Academic Skills parents reported more positive guidance at post-test than SIBS parents. Our post-hoc analyses suggested however, that, particularly for mothers, this unexpected finding was only evident before the COVID-19 pandemic. In contrast, the post COVID-19 findings align with prior SIBS trials with White families in showing no significant differences for SIBS as compared to no-contact control parents (Feinberg et al., 2013). It is notable that levels of positive guidance were high for parents in both groups at pre- and post-test (i.e., above 4.0 on a 5.0 scale), suggesting the possibility of ceiling effects in Latinx families (Updegraff et al., 2016). Within this sociocultural context, parents’ efforts to foster positive sibling relationships is a strength that is consistent with their family-oriented cultural values (Cahill et al., 2021). Importantly this pattern of results should be viewed with caution, given that, across the intervention targets, the only attrition effect was for mothers’ positive guidance, such that mothers with higher pre-test scores were less likely to participate at post-test.

For sibling relationship outcomes, starting with H4, post-test differences in siblings’ positivity were not significant for mothers’, fathers’ or children’s reports in the primary analyses. These findings contrast program effects showing more positive sibling relationships in SIBS versus no-attention control conditions based on mothers’ reports in a prior pilot feasibility study with Latinxs (Updegraff et al., 2016) and on mother and observer reports in a trial with predominantly White families (Feinberg et al., 2013). The null effects on children’s reports of positivity in this study are consistent with prior SIBS trials. Our measure of child-reported sibling positivity focused on children’s feelings of support/closeness rather than behavioral manifestations: The program’s impact on such sentiments may not be as easily detected, at least in the short-term and/or from children’s perspectives, as would more behaviorally focused reports of positivity. Turning to findings for fathers, they may have had fewer opportunities to observe or were less attuned to siblings’ positivity; our smaller sample of fathers also may have limited power to detect effects. Replication with a larger sample of Latinx fathers will be important in future work.

Contrary to H5 and prior work revealing a medium effect size with Latinxs based on mother-reported sibling negativity (Updegraff et al., 2016), we found no program effects on sibling negativity as reported by parents or siblings. Our post-hoc analyses, however, showed a small decline for the whole sample (i.e., both SIBS and Academic Skills groups) from pre-test to post-test based on parents’ reports. One possibility for this lack of effects is that program experiences (regardless of specific content) resulted in lower levels of sibling negativity at post-test. That is, although siblings in the Academic Skills program did not work together during the weekly sessions, in both intervention and control groups, siblings attended the program together each week, and like those in SIBS, Academic Skills siblings had shared experiences such as learning games that they could play together at home. Further, the family nights in both conditions included a family dinner and an activity wherein siblings taught their families games and played them together. These experiences may have led to shared activities at home as well as fostered positive and deterred negative sibling exchanges. In short, benefits for sibling relationships in both groups were consistent with evidence that shared sibling and family activities are beneficial for sibling relationship quality (McHale et al., 2000; Updegraff et al., 2005). Longer-term follow-up will help to determine whether these effects fade out for Academic Skills children more so than SIBS’ participants given socioemotional training in the latter. A design that includes a no-contact control group in addition to the intervention and attention control conditions will be informative in future research as well.

COVID-19 Timing and Program Effects

There are several potential implications of the COVID-19 pandemic for this RCT. First, it is possible that the onset and impact of the COVID-19 pandemic made it more difficult to detect program effects, particularly in the case of sibling relationship quality. Children’s abruptly and substantially altered daily contexts at the onset of COVID-19 meant that, for many children, siblings were their only playmates and companions. Prior work with a subset of this sample revealed that in the early period of the pandemic (i.e., immediately following school closures), sibling relationships became more positive as children spent more days at home together (Sun et al., 2021). Our findings here also showed that parents described sibling relationships as more positive and parents and children described sibling relationships as less negative among dyads recruited during versus prior to the pandemic’s onset. This evolving context may have had implications for sibling relationship quality that contributed to null program effects. Learning more about the short- and long-term effects of COVID-19 on children’s sibling relationships may inform future sibling-focused programming.

The pandemic onset also occurred when post-test data collection was ongoing for our largest cohort of families, resulting in close to two-thirds of post-test child surveys being collected by phone. This may have impacted data quality, especially from younger siblings, whose attention spans were shorter. The pandemic also had implications for program delivery for our final cohort. Most notably, staff wore facial coverings during all participant contacts and children and parents also wore masks at times, depending on school infection rates. Given some of the SIBS’ content on socioemotional expression, limited opportunities for non-verbal input due to face coverings may have impacted how effectively this content could be conveyed. Further, social distancing protocols implemented in afterschool sessions and family nights may have diminished some of the benefits that children and parents gained from interactions with others during the program. It is also possible that the daily challenges of the pandemic for our vulnerable group of families may have overshadowed some potential program benefits.

Another possibility to consider is that both programs may have offered some benefits to all families in the COVID-19 pandemic. Notably, we were equally successful in recruiting families before and during the pandemic. In addition, there were no significant differences in program attendance at weekly sessions or family nights in cohorts recruited both before and during the pandemic. This pattern suggests that families were motivated to attend despite potential risks during the pandemic. It may be that both programs provided parents and children with opportunities for social interactions that helped alleviate feelings of social isolation during a time of limited social contacts. Indeed, parents reported valuing both the positive interactions with their children and with other parents (Cahill et al., 2023). More generally, what we can learn from research conducted during the COVID-19 pandemic has the potential to expand our understanding of the role of larger sociohistorical events in shaping the implications of prevention programing for children’s development and well-being.

Strengths and Limitations

Strengths of this study include its integration of family, developmental, and prevention science theory and research to deliver a culturally relevant program for Latinx families, a group that has been noticeably absent from rigorous tests of prevention programs (Turner et al., 2022). SIBS was intentionally designed to be more time intensive for siblings than for parents, given the well-documented challenges of recruiting and retaining economically disadvantaged and minoritized parents into parenting programs (Pineros-Leano et al., 2023). Our success in recruiting, engaging, and retaining families, including during a global pandemic, is suggestive of the appeal of the program’s focus and approach. With a relatively small time commitment from parents (i.e., six hours of in-person time over three months), in combination with the more intensive time with siblings, we found evidence of predicted declines in parent involvement in sibling conflicts among SIBS versus attention control parents. Managing sibling conflicts is a common childrearing concern and a source of both stress and diminished mental health in parents (Feinberg et al., 2019). Helping children navigate sibling conflicts— potentially leading to less need for parental involvement, may be particularly beneficial in Latinx families where sibships are large (Ackert & Wikle, 2022) and family-oriented values prioritize harmonious relationships (Cahill et al., 2021). Further, the reduction in parental authoritarian control in sibling conflict, a finding that was consistent with the pilot study with Latinxs (Updegraff et al., 2016), is important given more generally documented negative impacts of authoritarian parenting styles (Steinberg, 2001) and evidence that it may be ineffective in sibling conflict (Kramer et al., 2019).

Our study also has limitations. From a design perspective, it will be important for future evaluations to employ independent reporters (beyond parents and children) and an external evaluation team. As this was the first efficacy trial of this program with Latinxs, our focus was on a culturally responsive approach to ensure that we engaged and retained families in both programs. An additional design limitation is the possibility of demand characteristics because interviewers read all items to parents and entered their responses. With this format, it is possible that parents may have felt compelled to respond in a way that they think was expected by program staff. Although demand characteristics must be considered, it is also important to consider the variability in participants’ reading levels, with some parents indicating that they were unable to read. Further, not all parents may have felt comfortable sharing that they had difficulties reading. Thus, reading questions and entering responses prevented potential undue burden on parents with reading difficulties and increased the quality and consistency of data across parents. Related to the curriculum, revisions could include more targeted parent skill development and emphasis on home practice activities, additional role plays at family nights, and more reinforcement of siblings’ skills during sessions. In addition, much of what we know about sibling-focused program effects pertain to mothers, as fathers are rarely included (except Feinberg et al., 2013). Although all family members were encouraged to attend family nights, only one parent was required to attend, and mothers participated more often than fathers. There are documented challenges to engaging fathers in prevention programming (Gonzales et al., 2023) and ways to increase fathers’ engagement are an important future direction. Obstacles for low-income families were addressed here (e.g., transportation, childcare), but employment challenges like long hours and inflexible schedules that characterize unskilled jobs (Dillman Carpentier et al., 2007; Winslow et al., 2009) are difficult to resolve and may reduce fathers’ participation. The COVID-19 pandemic likely exacerbated these challenges given its economic implications (Noe-Bustamante et al., 2021). Further, our sample included primarily two-parent households. Thus, it will be important for future work to identify creative solutions to engage fathers from different family configurations given their distinct and influential roles in children’s lives (Cabrera & Bradley, 2012).

As noted, a key difference between our study and prior work testing sibling-focused programs (Kramer et al., 2019) is that Academic Skills siblings experienced an equivalent intervention in terms of program contact time. Our focus on academic skills in the control condition, however, necessitated a different structure in the afterschool sessions, one that delivered content by grade level given the span of reading and math skills in our target age range. Thus, although the number and length of sessions were the same across programs, SIBS children worked together with their siblings and Academic Skills children worked with grade-level peers. To address this limitation and better understand the active ingredients of sibling-focused programing, a future evaluation should include an attention control condition with afterschool sessions that are similarly structured to SIBS but distinct in content. Such a design would increase our understanding of essential program features to affect positive change in sibling relationships.

In the face of this structural difference across conditions in the afterschool sessions, children in both conditions attended the program with a sibling, traveled home together, and had opportunities to interact during breaks. Further, the family nights in both conditions provided for constructive sibling interactions supervised by group leaders and parents. As such, these experiences may have contributed to improved sibling relationships evidenced by declines in sibling negativity in both groups and limited our ability to detect group differences. Future studies that vary these key aspects of the program will be informative.

Another limitation was our focus on a single sibling dyad in each family. Consistent with U.S. demographics (Ackert & Wikle, 2022), many families in this study had additional siblings. Sibling-focused programs that expand beyond a single target sibling dyad may yield stronger program effects on relational and individual well-being. Further, efforts to address multiple relationships, although logistically challenging, may be essential in families where multiple sibling dyads are the norm. Finally, this efficacy trial was conducted in a large school district in the southwest U.S. with strong educational infrastructure to support programing for Latinx families. In geographic regions with more limited infrastructure and less availability of bilingual individuals to deliver programs, implementation may be more challenging.

Conclusion

Targeting sibling relationships offers an important and underutilized opportunity to build essential socioemotional and relational competencies in middle childhood—a time when most children are learning and practicing these abilities. Further, relational skills that are learned in this dyadic context may allow siblings to become valuable sources of support later as children transition to adolescence and begin to navigate new complexities in their relationships outside the family. Latinx families tend to have more siblings than other groups (Ackert & Wikle, 2022) and a body of evidence highlights their strong family orientations (Cahill et al., 2021), factors that may provide an ideal context for sibling-focused prevention. Thus, it is critical to consider how prevention approaches can capitalize on familial strengths and resources to support the healthy socioemotional development of the largest ethnoracial minoritized group in the U.S..