Introduction

Military-connected (MC) elementary school students experience unique stressors, including frequent moves in and out of civilian schools (De Pedro et al., 2011). Frequent school transitions have been linked with poor parent–teacher relationships, child internalizing symptoms, and decreased school connectedness (Mehana & Reynolds, 2004; Scanlon & Devine, 2001). Scholars have called for targeted support for MC students (Esqueda et al., 2012; Fenning, 2021), but research on how schools can support this population is lacking (De Pedro et al., 2011). Contributing to the paucity of research in this area is a lack of adequate tools for assessing school supports specific to MC students. This study describes development of an inventory designed to assess the extent to which MC youth experience various types of school support and the extent to which those supports matter to MC students. We hypothesized that scores from this inventory would be positively linked to youth reports of school connectedness and academic efficacy, and to their parents’ reports of their involvement in school and parent–teacher relationship quality. We also expected youth who reported having a supportive adult in their life, either at school or outside of school, would report higher school supports scores.

School Mobility of Military-Connected Students and School Supports

Over 90 percent of MC students attend civilian schools (Esqueda et al., 2012), and it is estimated that MC students change schools three times more frequently than civilian students (Esqueda et al., 2012; Kitmitto et al., 2011; Masten, 2013). Military-related moves, or “permanent change of station” (PCS), often include a change in school districts for MC children. Frequent moves can be disruptive for MC families (Jelleyman & Spencer, 2008) and are linked with declines in students’ academic and socioemotional functioning (Rumberger, 2015). High mobility also predicts lower educational attainment in early adulthood (Hagan et al., 1996; Herbers et al., 2013). Some work also suggests that high mobility during elementary school is linked to future academic performance issues and maladjustment (Mehana & Reynolds, 2004).

There is some work suggesting that school supports might help to mitigate the negative effects of frequent school transitions. For example, students who feel supported at school and engage in supportive school practices tend to have positive academic, behavioral, and socioemotional outcomes (Bradshaw et al., 2008). Indeed, school supports designed to aid transitioning students, provide tailored academic support, and foster positive peer relationships have been linked with student flourishing (Caspe et al., 2011). Additionally, past research suggests that when schools provide support that reflects understanding of families’ values and student needs, the quality of the parent–teacher relationship improves (Torres-Rendón & Zinsser, 2022).

Assessing School Supports for Military-Connected Students

Military service is more than a profession; it represents a distinct culture (Cozza & Lerner, 2013) with a distinct language and hierarchy, and a unique sense of rules, self-expectations, and self-sacrifice (Cole, 2014). Military family life also has distinct features, including risk of physical and psychological harm to deployed service members, the challenge of service members reintegrating into civilian life, and “the relatively frequent churning of schools and peer groups experienced by children of active-duty service members” (Mancini et al., 2020, p. 647). Within this culture, a premium is placed on making sacrifices, being resilient, and weathering stressors with equanimity (Meyer, 2015). Because many MC students tend to behave in ways that reflect the disciplined, structured nature of their families, they have been described as “one of the most invisible minorities in public education” (Hanna, 2020, p. 184). However, because this is often the assumption of MC students, the idea these students are resilient can contribute to some falling through the cracks. This idea also could explain why school supports for MC families are relatively less developed compared to some other highly mobile youth populations (i.e., those who are unhoused, children of migrant workers, and youth in the foster care system; Griffin et al., 2019; Hamilton, 2013; Tyre, 2012).

The tendency for schools to overlook MC students is in accord with research indicating that schools struggle to support these students and their parents (Esqueda et al., 2012). School staff typically lack understanding of military culture, which can limit the level and quality of support offered to MC students. MC parents often feel unseen and underappreciated by educators (Cole, 2014) and a majority believe civilian schools fail to provide adequate support for their students (Berkowitz et al., 2014). As a result, the needs of MC students tend to go unnoticed until significant problems arise (Cole, 2014).

There is evidence suggesting MC families are highly resilient (Cramm et al., 2018) and that MC parents are hesitant to seek support from schools and other entities that could negatively affect their military standing (De Pedro et al., 2011). Studies also indicate that many MC students cope well with military-related moves and school changes (Masten, 2013), although these findings run counter to research indicating highly mobile MC students are at greater risk for poor functioning than civilian students (Masten, 2013). Cramm et al. (2018) suggested that apparent contradictions in the field’s understanding of MC students’ and families’ needs likely contribute to uncertainty on the part of schools about how best to support these families.

Because civilian schools struggle to meet the needs of MC students and families, scholars have called for research on the processes by which school staff can support highly mobile MC students (Fenning, 2021). An impediment to this work is a lack of adequate measurement tools. Despite research documenting the need of such tools (De Pedro et al., 2011; Esqueda et al., 2012), we could find no published studies that used psychometrically sound measures to assess the extent to which MC students viewed school supports as important or available at their school. Absent sound measurement, the question of whether MC students access and benefit from available school supports will remain unanswered.

The Transactional Nature of School Support for MC Students

Theoretically, our work is informed by an organizational approach to child development, one that frames children as active agents who exercise individual choice and who engage in recurring self-organization (Hayden & Mash, 2014). We assume these processes can influence how children come to access and experience supportive activities and relationships and can produce bidirectional transactions between what is available and what children actually experience as support from their environment (Pianta et al., 2003; Zee et al., 2013). Therefore, availability of school supports must be understood in the context of individual differences in children’s capacity for and propensity to access various sources of support (Ryan & Deci, 2002; Zee et al., 2013). However, to some extent, though children do have the capacity to choose which available school supports with which to engage, it is also important to note that socioeconomic and social contextual factors can facilitate or limit access to supports for some children (Barker et al., 2005). Specific to schools, we draw from Wentzel’s (2022) theoretical review of the relations among the constructs of school support, teacher-student interactions, and peer relationships. Wentzel (2022) posited that students in supportive schools will experience greater school belonging as a function of both individual-level (i.e., with friends or teachers) and group-level (i.e., broad social acceptance or inclusion) relationships. Wentzel also noted that students in supportive school environments will have a sense of inclusion that motivates them to engage in academic tasks, which is in line with recent work on the relation between school connectedness and academic efficacy and performance (Zeinalipour, 2022).

Wentzel also posited links between school support and the quality of parents’ relationships with teachers and their level of school involvement: Parents are more likely to invest in a relationship with their child’s teacher and to get involved in their child’s schooling if their child sees school as a haven of support and not as an uncaring or unprofessional environment. For example, Niehaus and Adelson (2014) found a positive association between perception of school support and parents’ involvement in their child’s school.

Current Study

In this study, we developed and evaluated an inventory designed to assess the extent to which MC children experienced various forms of school support as well as the degree to which those supports were thought to matter to MC students. We used an iterative process to develop the inventory and then examined its psychometric properties, including its internal structure, reliability, convergent validity, and criterion validity. We hypothesized that scores from this inventory would be positively linked to children’s reports of school connectedness, academic efficacy, parental involvement in school, and to parent ratings of parent–teacher relationship quality. Based on our supposition that children’s access to school supports reflects a transactional, bidirectional process, we also examined the degree to which scores from this inventory were associated with having a supportive non-parental adult in their life (either at school or outside of school). Assuming MC children differ in their tendency to access available supports (e.g., Rhodes, 2005; Ryan & Deci, 2002), we hypothesized that children who reported higher levels of school support would also report greater support from their family and from non-parental adults in their lives.

Method

Participants

Participants were 444 children (52% boys, 48% girls) with at least one active-duty military parent (5.3% were dual military families). All children were in grade three (45%) or five (55%), and all were enrolled in one of the 13 public elementary schools within a single school district near a large US military installation in the Pacific Northwest. We recruited families with students in 3rd and 5th grade because past research suggests that children begin to reliably complete most self-report measures at this age (Woolley et al., 2004). In addition, due to the longitudinal nature of the larger study from which these data are drawn and to maximize our potential sample, we recruited families with students in non-consecutive elementary school grades in two consecutive years. In most families, the service member was in the Army (86.4%), followed by Air Force (6%), Marines (2.1%), Navy (1.6%), National Guard (1.6%), Coast Guard (0.2%), or multiple branches (1.9%). As reported by parents, 57.3% of children were White, 10.7% Black or African American, 6.2% Native American, 5% Asian, 3.3% Pacific Islander, and 17.5% as another racial identity (i.e., bi/multiracial); 18.6% of parents identified their child as Hispanic, Latinx or Spanish origin. Nearly half of all parents (48.3%) indicated their child had changed schools at least two times since kindergarten, and 15% of children had a parent who was currently deployed (i.e., away for more than two weeks for military-related work/training), and the median length of current deployment was reported to be 12 or more months. Parent reports were provided by one parent associated with each child in the study. Most (93%) parents completing a survey indicated that only one parent in the home was active duty, 5% indicated that both parents were active duty, and about 2% indicated their family no longer had an active-duty service member at data collection. Among parents who responded to demographic questions, most were married or living with a partner (90%), and the remainder were divorced (7%), separated (1.9%), or single or never married (1.2%). Most parents identified as White (62.5%), 12% Black or African American, 6.5% Asian, 5.5% Pacific Islander, 4% Native American and 9.5% as another racial identity (e.g., bi/multiracial). In addition, 19% identified as Latinx. At the time of the survey, 15% of parents reported their child had experienced at least one parent deployment since they entered kindergarten, with approximately 55% of those deployments lasting 12 months or longer.

Procedures

For the larger study, two cohorts of families were recruited, beginning in January 2019 for Cohort 1 and January 2020 for Cohort 2. For both cohorts, an initial email was sent by a district-level administrator to all 3rd- and 5th-grade parents announcing the study, alerting them that more study information would be sent home with their student, and encouraging them to complete the consent form online or return it to their child’s teacher. Teachers were provided a brief one-minute video that introduced students to the study noting that participation required obtaining parent consent and returning the form to their teacher. Students were instructed to return the form regardless of their parents’ military status or decision to participate; teachers in classrooms that returned 75% of the forms were given a $25 Amazon gift card to be used for class activities. Reminder emails were sent to all 3rd- and 5th-grade parents. In addition, parents who had identified with the school district as MC were mailed a copy of the consent form and asked to sign and return it to their child’s teacher. A total of 4672 consent forms were distributed across both Cohorts. Of these, 3324 (71%) were returned with 673 (20.2%) parents indicating that their family was MC. Of these MC families, 531 (79%) parent–child dyads provided signed consent indicating their desire to participate in the study. Parents were sent a survey to the email on file with the school district for them to also complete for the larger study and we did not hear back from any parents expressing concern that they had not provided consent for their child to be in the study. Youth were given the opportunity to provide assent to complete survey measures at the beginning of their electronic questionnaire. Of the 531 parent–child dyads who consented, 444 children and 433 parents completed the baseline survey.

Cohort 1 youth completed electronic surveys via Qualtrics at school in Spring 2019. Cohort 2 youth’s Qualtrics surveys were administered remotely over Zoom by research team members in Spring 2020, after schools closed due to the COVID-19 pandemic. Youth received a $10 gift card for each survey they completed. Parents in both cohorts completed surveys online on the same schedule as their child. Individual survey links were sent to parents via email with reminders by phone, text, and email. Parents received a $20 gift card for completing the survey. All study procedures were approved by Boston University’s Institutional Review Board.

Measures

Inventory of School Supports-Youth Report (ISS-YR)

Development of the ISS-YR began with a review of published research and open-source platforms (e.g., government, nonprofit, and education websites, handbooks) that were relevant to MC student transitions. Based on this review, we identified a range of 34 possible school supports that could be offered to MC students as well as to other highly mobile student populations (e.g., homeless, migrant, foster care involved). We initially grouped these supports into the following broad categories: (a) proactive supports (e.g., school data systems and new student surveys to help schools track student mobility, guidebooks to foster school connectedness); (b) technology-based supports (e.g., mobile apps, virtual school tours, social media efforts); (c) school-specific programs (e.g., counseling and mental health services, academic credit recovery programs, newcomer clubs, summer bridge programs); and (d) district-wide practices (e.g., administrative policies, curricula development).

We next sought feedback on these content categories from various stakeholder groups, including MC parents and students as well as school staff with experience working with MC students and families (e.g., military family liaison, supervisor of Positive Behavioral Intervention Supports, elementary school teachers and principals). Specifically, we interviewed 6 principals of elementary schools with a high MC student population, and several MC students. Additionally, we conducted four phone interviews with MC parents outside of the sample school district. We also conducted small focus groups with two MC parents and one student as well as piloted a preliminary version of the inventory with these families. Feedback from these groups offered insight into military-specific and universal student supports, as well as unmet support needs, relevant to MC students. Based on feedback obtained, we reduced the inventory to 17 items to include only supports that elementary students could recognize and removed supports on which students might have difficulty reporting (i.e., IEP or 504 plans, etc.). We then reorganized items as follows: (a) formal academic supports for students, (b) formal social–emotional supports for students, (c) informal supports for students, (d) welcoming and transitional supports, and (e) supports designed specifically for MC students. In an early version of the measure, follow-up questions for each category of support explored whether the child’s school featured this support; how they had learned about the support; and which members of the family had accessed the support. These follow-up questions were later removed based on stakeholder feedback. At this point, we also removed four supports items deemed by stakeholders as redundant, resulting in a 12-item inventory.

Note that this process for constructing the ISS-YR was informed by the behavior analytic approach to scale development (Goldfried & D’Zurilla, 1969). This inductive approach to measure development, which has been widely used (see Elias et al., 1992; Farrell & Bettencourt, 2020), differs from more traditional approaches that begin with a large set of theory-derived items that are then trimmed via exploratory factor analyses. Items for the ISS-YR were written in the first person and described the student having experienced a specific type of school support (e.g., “Teachers helped me feel welcome when I first started at this school”). Some items were accompanied by a brief example to clarify their meaning (e.g., “Example: Showed me around and introduced me to my classmates”). Items were presented twice, first to ask the extent to which students experienced the support (“How true or not true is this statement for YOU?”; 1 = Not at all true for me to 4 = Totally true for me) and then to ask the extent to which students viewed the support as important to MC students (“How much do you think this kind of support matters to kids from military families?”; 1 = It doesn’t matter at all to 4 = It matters a lot). The second set of “importance” items was almost identical to the “experience” items, but were framed so that students could report on their perception of the support’s importance instead of whether they experienced it (e.g., “The school’s open house at the beginning of the year helped me feel welcome when I was new” was the framing in the first set of experience items whereas, “The school having an open house at the beginning of the year to help new students feel welcome” was the framing in the second set of importance items). The full 12-item ISS-YR is presented in Supplemental Table S1.

To capture students’ perceptions of both having experienced each support and the perceived importance of that support, we computed a cross-product score (experience X importance) for each ISS-YR item. However, before computing these scores, we first recoded as “missing” all items rated by youth as one that “doesn’t matter at all.” In this way, each student’s score on the ISS-YR reflected the extent to which they experienced only those supports reported as having some level of importance to MC students. We also dichotomized ratings of experiencing the support such that “Not at all true for me” was coded as zero and all other ratings (i.e., “A little true for me,” “Mostly true for me,” or “Totally true for me”) were coded as one. ISS-YR item scores can be interpreted as follows: 0 = not experienced; 1 = experienced but of limited importance; 2 = experienced and somewhat important; 3 = experienced and fairly important; 4 = experienced and very important. Scores for the ISS-YR and its subscales were formed by computing an average across relevant items.

We initially examined the internal structure of the ISS-YR via exploratory factor analyses (EFAs) using oblique rotation. Using a factor loading cutoff of ≥ 0.32 to retain items (Tabachnick & Fidell, 2001), two underlying factors emerged, each with six items. We labeled these factors as functional (i.e., focused on academic or military-specific supports) and relational (i.e., focused on welcoming, transition, and socioemotional supports) support, respectively (see Table 1). Items on the functional factor capture support activities and provisions that are informational, tangible, and instrumental focused; items on the relational factor capture emotional and relationship-focused supports. These factors parallel those found in previous research on the kinds of supports children experience at school. Dubow and Ullman (1989), for example, proposed a typology that included emotional, informational, and tangible support components.

Table 1 Youth school supports inventory (ISS-YR) Items descriptives
Full size table

School Connectedness

Students used the School Connectedness Scale (SCS; McNeely, 2005) to rate the extent to which they felt connected to their school. The SCS is a 6-item measure rated on a four-point scale (1 = Not at all true for me; 4 = Totally true for me). An example item is, “I feel like I am a part of this school.” The SCS has demonstrated evidence of strong psychometric properties, including convergent, discriminant, and criterion validity (McNeeley, 2005). In this study, the SCS had strong internal consistency reliability (Cronbach’s α = 0.81).

Academic Efficacy

We assessed students’ academic efficacy using the five-item Patterns of Adaptive Learning Scales–Academic Efficacy Scale (PALS-AES; Midgley et al., 2000). Items on this self-report measure are rated on a four-point scale (1 = Not at all true for me; 4 = Totally true for me). An example item is, “I can do almost all the work in school if I don’t give up.” In past research, the PALS-AES has demonstrated strong construct validity, stability, and internal consistency reliability (Midgley et al., 2000). In this study, the PALS-AES had good internal consistency reliability (Cronbach’s α = 0.80).

Parental Involvement in School

Students used the Parental Involvement and Support Scale (PISS; Anderson-Butcher et al., 2015) to rate the extent to which their parents were involved in their school. The PISS is a 4-item measure rated on a four-point scale (1 = Not at all true for me; 4 = Totally true for me). An example item is, “My parent(s) ask me about my school work.” The PISS has demonstrated evidence of strong construct validity and internal consistency reliability (Anderson-Butcher et al., 2015). In this study, the PISS had adequate internal consistency reliability (Cronbach’s α = 0.68).

Supportive Non-Parental Adult

Students were asked whether they had “an adult in their life, other than their parent, who they can really count on when they need help and who cares a lot about what happens to them.” A similar item has been used in previous studies to index whether youth have a supportive adult mentor in their life (Rhodes et al., 1992). Students have three response options and can check more than one box: (a) No, (b) Yes, an adult at school, and (c) Yes, an adult outside of school (but NOT my parent or another adult who lives with me).

Parent–Teacher Relationship Quality

Parents rated the quality of their relationship with their child’s teacher using the Parent Teacher Involvement Questionnaire-Parent Teacher Relationship scale (PTIQ-PTR; Conduct Problems Prevention Research Group, 1999). The PTIQ-PTR is an 8-item measure (e.g., “You feel comfortable talking with your child’s teacher about your child”) rated on a five-point scale (1 = Not at All; 5 = Very Much) that has demonstrated strong construct validity and internal consistency reliability (Conduct Problems Prevention Research Group, (1999). The PTIQ-PTR had good internal consistency reliability in this sample (Cronbach’s α = 0.90).

Data Analytic Plan

Missing data were handled using full information maximum likelihood (FIML; Enders & Bandalos, 2001). Confirmatory Factor Analyses (CFAs) using R software (R Core Team, 2016) with the lavaan package (Rosseel, 2012) were conducted to examine the ISS-YR’s factor structure. We compared the fit of two models based on the results of earlier EFA testing: One involved a single factor with all items loading on that factor, and the second contained two six-item factors representing functional and relational supports, respectively (see Table 1). Model fit was evaluated using Satorra–Bentler chi-square, the comparative fit index (CFI), the root-mean-square error of approximation (RMSEA), and standardized root mean residual (SRMR). The chi-square test was included as a measure of exact model fit, with non-significant p values indicating no significant differences between the models and observed data (Barrett, 2007). However, because chi-square tests are sensitive to large samples (Vandenberg, 2006), we also used the following fit indices: CFI values above 0.90, RMSEA values below 0.08, and SRMR values below 0.10 were used as criteria for adequate fit, whereas values above 0.95, below 0.05, and below 0.08, respectively, were criteria for an excellent fit (Vandenberg & Lance, 2000). To account for nested data (i.e., students within schools), CFAs were fit using robust maximum likelihood and corrected standard errors (see Dedrick & Greenbaum, 2011). We used a factor loading cutoff of 0.32 to evaluate whether items adequately loaded onto factors (Tabachnick & Fidell, 2001).

Internal consistency reliability was assessed using Cronbach’s alpha, and ordinary least squares (OLS) regression analyses were used to examine the criterion validity of ISS-YR scores when predicting academic efficacy, parent school involvement, school connectedness, and parent–teacher relationship quality. Regression analyses controlled for youth sex (i.e., male, female) and grade (i.e., 3rd, 5th). Separate analyses were used to test the predictive utility of a 12-item ISS-YR score and the two six-item subscale scores. We used a one-way ANOVA to examine associations between ISS-YR scores and youth reports of having a supportive adult in their life.

Results

Confirmatory Factor Analyses

We examined the underlying structure of the ISS-YR by using CFAs to test both a single- (Model 1) and a two-factor model (Model 2). For Model 1, the CFI, RMSEA, and SRMR values indicated an excellent model fit (SB X2 (51) = 75.926, p < 0.05; CFI = 0.953, RMSEA = 0.034, SRMR = 0.043). However, there was no evidence of an exact model fit based on X2. Model 2 tested a two-factor structure (i.e., functional supports and relational supports). This model’s X2 indicated a non-exact fit, but the CFI score indicated an adequate fit (SB X2 (53) = 91.048, p < 0.001; CFI = 0.924). Model 2’s RMSEA and SRMR scores indicated an excellent fit to the data (RMSEA = 0.041, SRMR = 0.045). All factor loadings (see Table 2) for both models were adequate (i.e., ≥ 0.32, p < 0.001). Neither Model 1 nor Model 2 included correlated residuals.

Table 2 Confirmatory factor analysis factor loadings for models 1 and 2 (one-factor model and two-factor model, respectively)
Full size table

Reliability of ISS-YR Scales

Internal consistency reliability for the ISS-YR composite and for the two support subscales was examined using Cronbach’s alpha (Table 3). Internal consistency for all three scales was adequate (Taber, 2018), with reliability estimates of 0.72, 0.63, and 0.62, respectively.

Table 3 Means, standard deviations, and correlations
Full size table

Descriptive Statistics

Means and standard deviations for individual ISS-YR items are presented in Table 1. Students ISS-YR scores and scores on outcome variables (e.g., academic efficacy, school connectedness, etc.) did not differ across cohorts. For the relational supports subscale, mean item scores (M = 2.26, SD = 0.63) indicated that when they were available, youth generally viewed these supports as “somewhat important” or “fairly important.” In contrast, mean scores for the functional supports subscale (M = 1.35, SD = 0.75) indicated that when they were available, youth viewed these supports as “a little important” or “somewhat important.” A dependent samples t test indicated that scores on the relational supports subscale were significantly higher than scores on the functional supports subscale (t(432) = 25.672, p < 0.001).

Validity Analyses

Bivariate correlations were used to examine associations between ISS-YR scores and hypothesized correlates (Table 3). As expected, significant correlations were found among composite ISS-YR scores and ISS-YR support subscales. In addition, composite ISS-YR scores were significantly and positively linked to academic efficacy, school connectedness, parent school involvement, and parent–teacher relationship quality. Similar associations were found between these variables and the two ISS-YR subscales, except that scores from the relational supports subscale were not related to parent–teacher relationship quality.

We used regression analyses to examine associations between ISS-YR scores and hypothesized correlates, after controlling for student sex and grade. ISS-YR composite scores positively predicted academic efficacy and school connectedness (Table 4). Grade negatively predicted school connectedness: 3rd grade students reported higher school connectedness scores than 5th-grade students. ISS-YR relational supports subscale scores positively predicted both academic efficacy and school connectedness, but functional supports subscale scores did not predict either criterion variable (Table 5). Similar to the previous model including ISS-YR composite scores as a predictor, grade negatively predicted school connectedness.

Table 4 ISS-YR composite scores predicting youth-report academic efficacy, school connectedness, parental involvement in school, and parent-report parent–teacher relationship quality
Full size table
Table 5 ISS-YR subscale scores predicting youth-report academic efficacy, school connectedness, parental involvement in school, and parent-report parent–teacher relationship quality
Full size table

Results for regression models predicting parent-oriented variables (i.e., youth-report parental school involvement and parent-report parent–teacher relationship quality) are presented in Tables 4 and 5. ISS-YR composite scores positively predicted both parental school involvement and parent–teacher relationship quality (Table 4). Sex did not predict either criterion variable. Grade negatively predicted parental school involvement: 3rd grade students reported more parental involvement than did 5th-grade students. However, grade positively predicted parent–teacher relationship quality: Parents of 5th-grade students reported better relationships with their child’s teacher than did parents of 3rd grade students. Models including the ISS-YR subscale scores as predictors revealed similar results (Table 5). Grade negatively predicted parental school involvement and positively predicted parent–teacher relationship quality. Both the functional and relational subscale scores positively predicted parental school involvement, but only functional supports scores predicted parent–teacher relationship quality, in the positive direction. Sex was unrelated to either of the criterion variables.

One-way ANOVAs were used to examine association between ISS-YR inventory scores and youth reports of having a supportive non-parental adult in their life. Four groups were compared: (a) youth who did not have a supportive non-parental adult in their life either at school or outside of school, (b) youth who had a supportive non-parental adult at school only, (c) youth who had a supportive non-parental adult outside of school only, and (d) youth who had a supportive non-parental adult both at school and outside of school. Significant group differences emerged for composite ISS-YR scores, F(3, 420) = 3.014, p = .030, and for the ISS-YR relational subscale scores, F(3, 419) = 3.980, p = .008, but not for the ISS-YR functional supports scores. Post hoc least significant difference (LSD) tests indicated youth who had a supportive non-parental adult both at school and outside of school had higher ISS-YR composite scores (M = 1.911, SD = 0.516) than youth who had a supportive adult at school only (M = 1.686, SD = 0.569; p = .004). Youth who had a supportive non-parental adult both at school and outside of school also had higher relational supports scores (M = 2.389, SD = 0.539) than (a) youth who did not have a supportive non-parental adult (M = 2.229, SD = 0.675, p = .038) and (b) youth who had a supportive adult at school only (M = 2.112, SD = 0.626, p < .001). Finally, youth with a supportive adult outside of school had higher relational supports scores (M = 2.303, SD = 0.611) than youth with a supportive adult at school only (M = 2.112, SD = 0.626, p = .043).

Supplemental Analyses

To provide support for the use of a “simple” version of the ISS-YR that uses solely the “importance rating” items, we conducted supplemental analyses mirroring those testing the full version of the ISS-YR (see Supplemental Materials). Results testing this version of the inventory indicated similar results: (1) a two-factor (functional and relational supports) solution emerged, (2) the “simple” ISS-YR scores were positively associated with youth-report academic efficacy, parental school involvement, and school connectedness but not to parent-report parent–teacher relationship quality, and (3) the composite “simple” ISS-YR and its two subscales had slightly higher internal consistency reliability than their cross-product counterparts.

Discussion

In this study, we addressed the need for a way to index school supports for military-connected students. The high mobility rates of these students coupled with increasing recognition of the importance of support and connections for students’ well-being and achievement (Wentzel, 2022) call for greater attention to the nature and impact of school supports. We developed and evaluated the Inventory of School Supports-Youth Report (ISS-YR), a tool by which youth rate the degree to which they received school supports that matter to MC students. The iterative process of developing the ISS-YR helped to ensure its content validity, and confirmatory factor analyses supported both a one-factor model with all 12 items and a two-factor model represented by functional and relational supports. The 12-item composite ISS-YR scale demonstrated acceptable internal consistency reliability (Taber, 2018), whereas reliability estimates for the 6-item relational and functional supports subscales were minimally acceptable. It is also possible that limited subscale reliability could be explained by the fact that the ISS-YR measures do not capture an underlying construct such as depression or anxiety, but a collection of similar phenomena such as available school supports (Cleary, 1981; Evans, 1991). In essence, the ISS-YR operates similarly to measures that allow participants to report on similar experiences, such as the Adverse Childhood Experiences measure (ACE; Felitti et al., 1998). However, research indicates that internal consistency reliability estimates are in part a function of the number of items in a scale (Taber, 2018). Lower subscale reliability could also be attributed to them having a low number of items.

Bivariate correlations and regression analyses supported hypotheses about the criterion-related validity of the ISS-YR. The composite ISS-YR score was positively associated with youth academic efficacy, school connectedness, parental involvement in school, and parent-reported quality of the parent–teacher relationship. Similar patterns of associations emerged for the two subscale scores, but there were also key differences. Namely, parent-reported parent–teacher relationship quality was positively linked to the ISS-YR composite and functional supports scores but not to relational support scores. This is perhaps not surprising given that items on the relational supports subscale focus on the social climate and well-being of youth and not parents, whereas items in the functional supports subscale focus on academic and military-specific supports, which likely influence how parents view their child’s teacher.

Regression analyses indicated composite ISS-YR scores positively predicted academic efficacy, school connectedness, parental involvement in school, and parent–teacher relationship quality when controlling for youth grade and sex. We also found that scores from the ISS-YR relational supports subscale significantly predicted academic efficacy, whereas this was not the case for functional supports subscale scores. Because some items on the functional supports subscale relate specifically to academic supports, this finding is somewhat surprising. However, there is research highlighting important linkages between children’s social–emotional functioning and their academic performance (Gallardo et al., 2016). Notably, MC youth reported that relational supports were more important, on average, than functional supports. This supports the idea that fostering socioemotional learning can improve both peer relationships and academic performance (Bierman et al., 2010). Both relational and functional supports subscales significantly predicted parental school involvement; however, only functional supports scores predicted parent reports of the quality of the parent–teacher relationship. Perhaps parents are more aware of children’s receipt of functional supports (i.e., academic and military-specific supports) than relational supports, which could explain why parents of children with higher functional supports scores reported a better relationship with their child’s teacher.

Regression analyses also indicated that youth grade was a negative predictor of parental school involvement but a positive predictor of parent–teacher relationship quality. Grade-related differences in parental school involvement are consistent with work suggesting parents are often more involved during their child’s early education experiences than later in elementary school (Avvisati et al., 2010). Previous research has shown that parent–teacher relationship quality also tends to increase as children progress from lower to upper grades within a school district (Yazdani et al., 2020), suggesting that parents might become more comfortable with the school and its teachers the longer the child remains in that school district.

We also found significant differences in ISS-YR composite and relational supports scores (but not functional supports scores) based on whether youth had a supportive non-parental adult in their life. Youth who had a supportive non-parental adult both at school and outside of school reported the highest ISS-YR scores compared to youth who lacked having a non-parental adult in their life or had this kind of support but only at school or outside of school. Thus, youth who reported experiencing support outside of school were also more likely to report experiencing support in school. These findings could mean that youth who have supportive relationships in multiple contexts are somehow privileged in their receipt of school supports, but our supposition is that youth who have a capacity to access and benefit from relationships in other contexts are better able to access school supports as measured by the ISS-YR (Hagler et al., 2019; Rueger et al., 2016). To the degree this interpretation is valid, it would mean that some MC youth are less likely to experience school supports even if made available. However, as mentioned previously, though individual differences in youth’s ability to access supports likely exist, it is important to note that broader structural factors (i.e., socioeconomic status, race, ethnicity, etc.) also play a role in increasing or limiting access to supports (Barker et al., 2005).

Strengths, Limitations, Implications, and Future Directions

This study has several strengths, including adding to an emerging body of the literature on how best to support MC parents, students, and families (Astor & Benbenishty, 2014; Fenning, 2021; Spencer et al., 2020). To our knowledge, this is the first study to develop a measure indexing both the extent to which MC youth perceive school supports to be important and the extent to which they receive these supports. We used an in-depth, iterative process to develop the ISS-YR including obtaining feedback from key stakeholders. The study also benefited from a large, diverse sample of MC elementary school students.

There are also several limitations to the study. Our sample was drawn from 3rd- and 5th-grade students in a single school district. Other school districts and grade levels may support MC families in other ways. This is particularly important given differences we found in school connectedness, parental school involvement, and parent–teacher relationship quality by grade. Notably, a considerable proportion of the students in our district were military-connected (i.e., about 20%) given the district’s close proximity with a prominent military base in the Pacific Northwest. Future work will need to assess the ISS-YR’s utility in districts with a higher or lower proportions of MC students, as the number of tailored supports for these children might differ substantially. Relatedly, given that the proportion of MC students in schools can be highly variable, districts will likely also differ in the extent to which schools consider tailored supports for MC students to be worthwhile. More research is needed to assess school supports for MC students in districts with a smaller proportion of MC students. Also, because the parents of most MC students in this sample were affiliated with the Army, generalizability of our findings to MC families affiliated with other military branches is limited. Our study was also limited by our use of youth self-reports to assess their experience with school supports. Future studies should use additional means to assess those supports youth experience and the degree to which they benefit academically and socially from those experiences. Additionally, it is possible that the wording of the “importance ratings” items in the third person (i.e., “How much do you think this kind of support matters to kids from military families?”) could present issues with interpretation of the ISS-YR scores. Our intention with this wording was to allow students to comment on whether they believed certain supports were important to MC students generally. We also made this decision, in part, to maintain consistency with a parent-report inventory of school supports also developed for the larger study. However, the possibility exists that some students could complete the measure considering only their own experience. More research is needed to assess what potential impacts the wording of the ISS-YR could have on how scores related to other constructs. Finally, longitudinal studies are needed to examine the degree to which ISS-YR scores are temporally stable and predictive of student functioning over time.

We also see merit in efforts to adapt the ISS-YR for use with other highly mobile student populations, including foster youth, homeless youth, and children of migrant workers. Most items in the ISS-YR are not specific to MC children. Thus, it could be used to understand how schools can use various practices to support other highly mobile student groups. We would expect both common and unique needs across these groups.

Our study also has potential implications for how schools can support MC students. For example, the ISS-YR could be used to identify gaps in the kinds of supports made available to MC students, as well as to gauge those supports youth consider important and are currently accessing. Schools could use this information when making decisions about how to allocate resources to support MC students. For example, the current findings suggest schools might consider emphasizing relational school supports versus functional supports if the aim is to enhance MC students’ self-rated academic efficacy. However, it is important to reiterate that these findings are preliminary and should be replicated in future research.

In this study, we aimed to use the ISS-YR to capture both students’ experience of school supports and the extent to which students viewed those supports as important to MC families. To do so, we created a cross-product score that provides an index of the importance of school supports MC students actually experienced at school, rather than a simple rating of how important all supports were to MC students. Though we believe this cross-product score obtained from the ISS-YR produces perhaps the most meaningful information about school supports, we recognize that calculating this score could be cumbersome for teachers or school officials. Scoring syntax for the cross-product ISS-YR is included in the Supplemental Materials. To provide a simple index of MC students’ perceptions of supports, school officials could use either students’ reports of how many supports they experienced at school to assess support engagement, or they could use students’ reports of how important these supports are to assess potential support impact or relevancy.

Results from supplemental analyses testing a “simple” version of the ISS-YR, including only importance ratings, indicated that this version performs similarly to the cross-product version of the ISS-YR. Specifically, we found the same two-factor solution with functional and relational supports. This version of the ISS-YR was similarly related to other study constructs, and internal consistency reliability estimates for the composite scale and both subscales were adequate. The latter finding supports the idea that the multiplicative scoring process could have contributed to lower internal consistency reliability. Broadly, these results indicate that schools can use either version of the ISS-YR with confidence; the former (i.e., simple) version of the measure can be used to index students’ general perceptions of which supports are important to MC students and the latter (i.e., cross-product) version of the measure can be used to index how important are the supports that were available or experienced by students at school.

Conclusion

This study contributes to a growing body of research on the unique educational experiences of MC students and the need for a measure to assess youth perceptions of supports at their schools. Our findings provide preliminary support for the psychometric utility of the ISS-YR, including significant associations with academic efficacy, school connectedness, parental school involvement, and parent–teacher relationship quality. Our findings also suggest that differences in the support MC students experience in other contexts (e.g., with non-parental adults outside of school) could relate to students’ ability to access and benefit from school-based supports. More research is needed to determine how to increase access to school supports for MC students and other highly mobile populations.