Although there are various forms of prosocial behaviors, many such behaviors can be organized into actions that are exhibited in specific contexts or that are motivated by selfless or selfish desires (Eisenberg et al., 2015). Carlo and Randall (2001) presented several commonly studied forms of prosocial behaviors but two forms are particularly relevant to the present study. Altruistic prosocial behaviors are defined as helping others with little or no expectation for self-reward. Public prosocial behaviors, on the other hand, refer to helping in front of an audience. Conceptually, altruistic and public actions are expected to show a differentiated pattern of relations to other theoretically-relevant constructs and distinct developmental trajectories.

Altruistic prosocial behaviors have long been the focus of scholarly debate given the important implications regarding the relative benefit for others at the cost or expense to one's self and the nature of humans (Wilson, 1975; Batson, 1998). Altruistic prosocial behaviors are deemed to be primarily selflessly-motivated and predicted by strong moral principles and/or sympathy and often incur a (psychological or physical) risk to one's self (Staub, 2013; Eisenberg et al., 2015). Indeed, extreme examples of altruistic behaviors often draw much attention because they include acts of heroism and courage at the possible risk of one's life or health. Much of this work refers to moral or care exemplars—individuals who demonstrate committed acts of charity at great personal risk to themselves (Oliner and Oliner, 1988; Colby and Damon, 1992; Hart and Fegley, 1995). However, it is important to note that altruistic behaviors can also include somewhat less dramatic yet costly actions such as committed acts of charity, volunteerism, donations, and generosity (Staub, 2013; Carlo, 2014).

In contrast, scholars assert that public prosocial behaviors are primarily selfishly-motivated with an instrumental purpose (Carlo and Randall, 2001). These actions might be exhibited as a means to gain the approval of others or to elevate one's social status. Moreover, sometimes these actions can be a means to manipulate or harm others by gaining trust from others. Researchers have suggested that public prosocial behaviors, in contrast to most other forms of prosocial behaviors, can lead to subsequent aggressive and maladaptive behavioral outcomes (Carlo, 2014). On the other hand, it is possible that such actions might serve an immediate purpose to improve one's mood or to gain approval of others (e.g., parents, teachers) in appropriate social contexts.

Researchers have yielded substantive evidence that there are distinct correlates and developmental trajectories of altruistic and public prosocial behaviors (Carlo, 2014; see McGinley et al., 2014 and Xiao et al., 2019, for reviews). For example, as expected, high levels of prosocial moral reasoning and sympathy have been linked to altruistic prosocial behaviors but negatively linked to public prosocial behaviors among U.S. Latinx youth. Furthermore, social desirability is modestly, but significantly, positively related to public prosocial behaviors but not altruistic prosocial behaviors (Carlo and Randall, 2002). A recent study showed that public prosocial behaviors were associated with hoarding and stockpiling supplies during the COVID-19 pandemic (Dinić and Bodroža, 2020).

Altruistic prosocial behaviors (i.e., actions primarily intended to benefit others with little to no regard for self-benefit) are of special interest to researchers interested in growth in character strengths following exposure to adversity and trauma. This interest is situated in a small, but growing, body of literature focused on the altruism-born-of-suffering hypothesis that trauma events can lead to reorganization and self-reflection on life's purpose and meaning, which can result in new schemas to reduce suffering, and improve the lives of others (Staub, 2013). Additionally, exposure to trauma could sensitize some individuals to the plight and suffering of others, which can move individuals to engage in actions intended to improve and assist others (Davis et al., 2016). There is suggestive evidence primarily based on cross-sectional or retrospective report study designs (see Staub and Vollhart, 2008, for a review), that some children and youth exposed to trauma (e.g., war, violence) exhibit high levels of altruistic behaviors (Eisenberg et al., 2015). Moreover, there is evidence of some adults' altruistic behaviors, even at risk to one's self, under extreme trauma circumstances (e.g., Holocaust, war) or subsequent to traumatic events in childhood (Oliner and Oliner, 1988; Harel et al., 1993; Bonanno, 2004; Grossman et al., 2006). However, direct longitudinal evidence on this notion of change in altruistic behaviors as a result of earlier trauma exposure is lacking, especially in U.S. minoritized youth.

Thus, we sought to test the altruism born of suffering hypothesis by examining the patterns of developmental change in altruism following exposure to adversity. As a sensitivity test of the hypothesis, and because the nature of public prosocial behaviors is decidedly distinct from altruistic prosocial behaviors in terms of conceptual definition, developmental trends and antecedents (Carlo, 2014; see McGinley et al., 2014 and Xiao et al., 2019), we also examined associations between trauma exposure and developmental change in public prosocial behaviors.

Despite the existing work on the altruism-born-of-suffering hypothesis and the speculation on how moral growth could occur, there are no direct studies of possible growth triggers mechanisms that could spur positive adaptations. Most models of prosocial and moral development have emphasized sociocognitive and socioemotive mechanisms (e.g., perspective taking, moral reasoning, empathy/sympathy) that predict prosocial behaviors (Eisenberg et al., 2015). Other scholars note the important influence of psychological (e.g., temperament, self-regulation) and socialization (e.g., parental practices and styles, peer affiliation, media exposure) mechanisms (Hoffman, 2000; Grusec, 2011; Eisenberg et al., 2015). In recent years, however, there has been increasing attention to integrative approaches that simultaneously consider the interplay of culture-group-specific mechanisms (e.g., cultural values, cultural stress) and non-culture-group-specific mechanisms (e.g., self-regulation or coping) to better account for individual and group differences in prosocial and moral developmental outcomes. Carlo and de Guzman (2009) posited, for example, that stressors (e.g., trauma), cultural values (e.g., familism), and coping at earlier ages could jointly account for prosocial behaviors at later ages. In their proposed ecocultural stress-based model, early life experiences (e.g., trauma exposure) are processed via moderating and mediating mechanisms (e.g., problem-focused coping behaviors, familism values) that subsequently predict individual and group differences in prosocial development among ethnic/racial minority youth.

In the context of understanding variability in the development of prosocial behaviors following trauma exposure, one might expect that trauma would result in distinct developmental trajectories based on individual differences in growth triggers such as problem-focused coping and/or the strength of their affinity to their family unit (i.e., familism). Researchers have suggested that trauma-exposed individuals who have well developed coping or self-regulation capacities might more effectively modulate their emotional and behavioral reactions to trauma, which can facilitate positive adaptation after trauma (Linley and Joseph, 2004). Indeed, problem-focused coping (i.e., attempts to reduce or eliminate stress source) have been linked to posttraumatic growth in adults (Linley and Joseph, 2004) and self-regulation and coping is also conceptually and empirically linked to higher levels of prosocial behaviors (Carlo et al., 2012; Eisenberg et al., 2015). For example, researchers have demonstrated that youth who exhibit relatively high levels of problem-focused coping also exhibit high levels of prosocial behaviors (Carlo et al., 2012). Although problem-focused coping is positively linked to prosocial behaviors, evidence on direct relations between problem-focused coping and specific forms of prosocial behaviors does not exist. However, individuals high in certain forms of problem-focused coping – in particular positive cognitive restructuring and seeking understanding – might be particularly prone to demonstrate growth following exposure to acute traumatic events (Linley and Joseph, 2004). Therefore, based primarily on theory (Carlo and de Guzman, 2009), we expected that trauma-exposed youth who are high (vs. low) on problem-focused coping might exhibit increases in altruistic prosocial behaviors and decreases (or no change) in public prosocial behaviors following trauma exposure.

An additional candidate variable that could empower positive adaptation to adversity is linked to relatively recent research on prosocial development in U.S. Latino/a youth (Carlo and de Guzman, 2009). Cultural researchers have identified familism (i.e., identity with, support, and obligation to kin) as a relatively common value strongly endorsed by many Latino/a families (Sabogal et al., 1987; Knight et al., 2010). Conceptually, familism embodies learning to be respectful and considerate of family members, to be socially responsible and responsive to the needs of family members, and to provide support to family members. These expectations and norms can provide important training grounds to apply prosocial attitudes and altruistic tendencies to others. Moreover, in general, familism beliefs might provide U.S. Mexican youth with psychological and social support needed to positively adapt to trauma exposure. Indeed, there is substantive, accumulated evidence that familism is positively related to several forms of prosocial behaviors in U.S. Latino/a youth (e.g., Armenta et al., 2011; Knight et al., 2014; Zhao et al., 2022; see Carlo, 2014).

The present study focused on late childhood through early adulthood based on known age-related changes in prosocial development (e.g., Van der Graaff et al., 2018; McGinley et al., 2021). We longitudinally examined whether earlier reported trauma exposure predicted changes in two distinct forms of prosocial behaviors. Specifically, we examined the association between youth exposure to traumatic events during late childhood to middle adolescence (grades 5, 7, and 10) and later development of altruistic and public prosocial behaviors from middle adolescence to early adulthood (grades 10, 12, and 5 years post high school) among U.S. Mexican youth. Problem-focused coping and familism values were examined as two growth-promoting mechanisms that might facilitate resilience, recovery, or growth following exposure to adverse traumatic events. In addition, given gender socialization theory and prior reported gender differences in prosocial behaviors (Xiao et al., 2019), we then examined whether gender moderated these relations.

Given prior conceptual notions (Carlo and de Guzman, 2009; Staub, 2013) and the prior evidence on the relations between familism and prosocial behaviors, we hypothesized that trauma exposed U.S. Mexican youth who also endorsed higher levels of familism or problem-focused coping (relative to lower levels) might yield increases in altruistic prosocial behaviors and decreases (or no changes) in public prosocial behaviors across time. We did not have assessments of prosocial behaviors prior to trauma exposure. Our design, instead, compares developmental changes in prosocial behaviors among adolescent who were and were not exposed to trauma. This design incorporates the strengths from resilience and posttraumatic growth approaches (Infurna and Jayawickreme, 2019).

Mplus 8.0 (Muthén and Muthén, 2017) was used to evaluate the study hypotheses. First, we used a series of growth curve models (GCMs) to assess gender equivalence for growth factors (intercept and linear slope) for each subscale of prosocial behaviors at Waves 3, 4, and 6 (coded as 0, 2, 6 in GCMs to capture the time that elapsed from 10th grade to young adulthood). The criterion for determining gender equivalence in growth patterns was a significant chi-square difference test (α level of 0.01) between the model constraining all paths to be equal for boys and girls and the model allowing the means of the growth factors to differ. If the chi-square difference test was not significant, the subsequent analyses were completed for the entire sample; otherwise, the subsequent steps were completed separately by gender. Next, two GCMs were conducted including childhood trauma (W1-W3) as a predictor of the trajectories of altruistic and public prosocial behaviors. This was accomplished by adding to the models in step 1 the binary trauma variable as a predictor of the growth factors (i.e., intercept and slope). Finally, we examined the potentially moderating roles of coping and familism on the relations between trauma exposure and the intercept and slope of the prosocial behavior subscales. Each moderator was tested separately by adding familism or coping and the interaction term to the models in step 2 as the predictors of the growth factors. Interaction terms were created between the trauma variable and each moderator (i.e., familism and coping), after the variable was centered. For significant interaction terms, simple slopes were examined for trauma on the growth factor at different levels of the moderator (i.e., higher and lower levels at +1 and −1 standard deviation from the mean, respectively). A probability level of < 0.05 was used to determine significant main and interaction effects.

Maximum Likelihood (ML) estimation was employed. Skew and kurtosis of the items were within the cutoffs provided by West et al. (1995) for use of ML estimation: skew ranged from −1.18 to 0.4 and kurtosis ranged from −0.68 to 1.32. To facilitate missing data strategy, auxiliary variables that were theoretically related to prosocial behaviors (e.g., consideration of others, social competence) were included in the models. To evaluate the goodness of fit of each model, root mean square error of approximation (RMSEA), comparative fit index (CFI), standardized root mean square residual (SRMR), and chi-square tests were used as fit indices. Adequate fit was based on the following cut-off scores: RMSEA < 0.08 and CFI > 0.95 (Hu and Bentler, 1999; Yu, 2002).

Missing data were handled in all models with full information maximum likelihood estimation (FIML). Due to a large attrition rate at Wave 6, attrition analyses were completed for the following Wave 1 variables: demographics (age, gender, income, nativity, household structure), coping, familism, and trauma. Compared to those who completed Wave 6 assessment, children who did not participate in the assessment were more likely to be boys [52.7% vs. 41.8%; χ² (1) = 8.99, p < 0.01], older [M_attrited = 10.91 vs. M_retained = 10.82; t (1) = 7.88, p < 0.01], and have lower family income (M_attrited = 5.96 vs. M_retained = 7.42; t = 20.51, p < 0.01) at Wave 1, as well as more likely to report lower familism (M_attrited = 4.50 vs. M_retained = 4.57; t = 4.07, p < 0.01) and lower coping (M_attrited = 3.75 vs. M_retained = 3.90; t = 7.94, p < 0.01) at Wave 1. No differences were found for trauma, child nativity, or household structure at Wave 1. We were unable to explore differences in prosocial behaviors at Wave 1 or 2 because this construct was not administered then due to developmental considerations.

Structural equation models tested whether trauma exposure was related to the intercept (middle adolescence) and slope (change from middle adolescence to early adulthood) factors for each prosocial behavior model. Fit indices supported adequate model fit such that RMSEAs were between 0.04 and 0.07, CFIs were between 0.96 and 0.98, and SRMRs were between 0.9 and 0.12. Unstandardized path coefficients for models with trauma predicting growth factors are presented in Table 4 (i.e., Model 1).

Earlier trauma exposure predicted, on average, lower altruistic intercepts for girls but had no association with girls' altruistic slopes. These findings were in the direction of childhood trauma being costly, on average, to U.S. Mexican girls' altruistic prosocial behaviors. Conversely, childhood trauma predicted higher public intercepts for girls, but was unrelated to their public prosocial behavior slopes, suggesting childhood trauma being beneficial, on average, to U.S. Mexican girls' levels of public prosocial behaviors in middle adolescence. For boys, childhood trauma did not predict altruistic or public prosocial behavior intercepts or slopes.

Moving beyond the previous models, which examined the associations between childhood trauma and prosocial behavior growth factors, we conducted model tests that examined whether the associations between trauma exposure and prosocial behaviors varied for those youth higher and/or lower on familism or coping. Model fit was adequate for all models such that RMSEAs were between 0.01 and 0.06 and CFIs were between 0.96 and 0.99. Unstandardized path coefficients are presented in Table 4 (i.e., Models 2a and 2b). Intercept and slope means as well as relations between trauma and the growth factors did not change in direction or in significance levels in the models that added the interactions. Therefore, for efficiency, we highlight significant interaction effects and present simple slope tests when needed in text, and present all additional model parameters in Table 4.

Of particular importance were the novel findings that suggest increases in altruistic prosocial behaviors for U.S. Mexican girls from early adolescence to young adulthood. Although the evidence demonstrates that such growth occurred, the findings were gender-specific such that, among those exposed to trauma, U.S. Mexican girls who reported relatively high levels of familism displayed this growth. For boys, there was a similar but nonsignificant pattern of findings, such that those who reported relatively high levels of problem-focused coping showed such positive adaptations.

For girls who reported relatively high levels of trauma experiences, high levels of familism (compared to lower levels of familism) initially were associated with lower scores on altruistic prosocial behaviors. The initial negative relation between relatively high levels of familism and altruistic behavior scores might be due to the initial challenges of coping with trauma, such that the relative immediacy of trauma exposure undermines one's ability to selflessly consider the needs of others. This is consistent with recovery patterns (Infurna and Luthar, 2018; Infurna and Jayawickreme, 2019), such that there were relatively low levels of altruistic prosocial behaviors for highly familistic, trauma-exposed girls in middle adolescence (compared to low-familistic, trauma-exposed girls). Given the emphasis on familism in Latinas, as compared to Latinos, these findings might reflect the comparative greater benefits on a strong connection to the family for U.S. Mexican girls even when exposed to trauma. This notion is consistent with higher levels of altruistic tendencies for girls relative to boys in the present study and with prior evidence that U.S. Latinas more strongly endorse familism and exhibit closer parent-child relationships than U.S. Latinos (Updegraff et al., 2005). Thus, the closer connection and greater importance placed on the family in girls might provide an important resource to help trauma exposed girls, as compared to trauma exposed, low familistic girls, to demonstrate increasingly levels of altruistic prosocial tendencies.

In contrast, findings for trauma-exposed U.S. Mexican boys who were high (compared to low) on problem-focused coping, showed trend (at the p < 0.07, statistically nonsignificant level) evidence of increases in altruistic prosocial behaviors from middle adolescence to early adulthood. Given that boys' altruistic prosocial behaviors were increasing over developmental time (see Table 3 for slope), their pattern suggests that trauma-exposed boys with high levels of problem-focused coping were showing relatively steeper increases in altruistic prosocial behaviors as compared to all other groups of boys. Thus, for trauma-exposed boys with high problem-focused coping resources, the positive trend in altruistic prosocial behaviors is consistent with growth (Infurna and Luthar, 2018; Infurna and Jayawickreme, 2019). The pattern is also in accord with stress and coping scholars who assert that problem-focused coping is generally a constructive set of strategies that can help persons effectively deal with stressors (Lazarus and Folkman, 1984; Compas et al., 2001). That the beneficial trend effects of problem-focused coping were manifested in U.S. Mexican boys rather than girls might reflect the strong instrumental orientation that boys tend to exhibit relative to girls (Maccoby, 1990). Alternatively, boys who are strong in problem-focused coping might be more likely to take responsibility for others in ways that are congruent with traditional gender-typed expectations linked to masculinity (machismo). However, given the nonsignificant findings, there is a great need for future studies to determine whether these findings are reliable.

To our knowledge, these are first empirical findings that show age-related increases in altruistic prosocial behaviors in a U.S. ethnic/racial minority group exposed to childhood trauma experiences. That, for girls, familism (and non-significantly, problem-focused coping for boys) served as a trigger for patterns of positive adaptation from relatively low baseline levels of altruistic prosocial behaviors demonstrates suggestive evidence of growth and recovery stemming from culture- and non-culture-specific person traits. Such evidence is consistent with resilience and posttraumatic growth scholars who assert the need to identify protective and enhancer mechanisms for high-risk youth (Masten and Narayan, 2012; Infurna and Luthar, 2018; Carlo et al., 2022). These overall findings are also in accord with prior research that links familism, problem-focused coping, and other stressors (e.g., discrimination) to prosocial behaviors and with contemporary culture-specific approaches to understanding prosocial development in U.S. ethnic/racial minority youth (Carlo and de Guzman, 2009; Knight and Carlo, 2012). By considering the interplay of culture-group and non-culture-group specific mechanisms, researchers might be better able to account for distinct developmental trajectories of prosocial development in ethnic/racial minority youth even when such youth experience exposure to trauma.

Moral developmental scholars (Staub and Vollhart, 2008) who postulate that moral growth can manifest from exposure to adverse circumstances (altruism-born-of-suffering) have not directly examined how such growth occurs. The present findings show a nuanced pattern of moral development increases that differed across gender and growth triggers. Increases in altruistic prosocial behaviors were revealed by patterns of recovery, as well as, growth. In other words, altruism-born-of-suffering for some youth were present after initial declines in altruistic prosocial behaviors followed by growth (girls high on familism); whereas, for trauma exposed boys who reported relatively high levels of problem focused coping, altruistic prosocial behaviors showed a nonsignificant trend with little initial decline in such behaviors. Because altruistic prosocial behaviors were not assessed prior to trauma exposure, determination of resiliency growth or chronic low trajectories were not possible. Nonetheless, the overall patterns of distinct forms of altruism-born-of-suffering provocatively suggests distinct trajectories of some ethnic/racial minority youth who are capable of overcoming adverse earlier life experiences. Such findings might inform the moral development of ethnic/racial minority persons from adverse life circumstances who significantly contribute to improving the lives of needy others.

Beyond the effects in predicting altruistic prosocial behaviors, there were also several nuanced patterns of relations among trauma exposure, familism and coping in predicting public prosocial behaviors. For both boys and girls, regardless of trauma exposure, coping was associated with high levels of public prosocial behaviors. These findings might reflect the possibility that public forms of prosocial behaviors might serve similar instrumental function as that of problem-focused coping given that public prosocial behaviors can result in positive feedback and approval from others and subsequently reduce stress (Carlo and Randall, 2002). However, higher levels of coping were also linked to decreases in public prosocial behaviors over time. This latter pattern of relations suggests that, over time, the efficiency of problem-focused coping in managing stress could trigger less need to engage in such prosocial behaviors designed to gain others' approval or positive social feedback. Although further longitudinal research is needed to better understand these findings, it is possible that the effects of coping or self-regulation on other specific forms of prosocial behaviors could change over time.

Girls low on problem-focused coping and unexposed to trauma scored higher on public prosocial behaviors than those low on problem-focused coping who reported trauma. These findings are somewhat surprising given prior work in U.S. Latino/a adolescents that revealed positive relations between acculturative stress (including discrimination) and public prosocial behaviors (Brittian et al., 2013; Davis et al., 2016). Scholars have recognized that prosocial behaviors can sometimes serve as a coping mechanism to alleviate one's own distress or to improve one's mood (e.g., Schacter and Margolin, 2019). Other scholars note that stress experiences might orient persons to focus on one's self (rather than others), which might facilitate public prosocial actions that could evoke positive social feedback to gain social support and resources to deal with such experiences (Davis et al., 2016). However, the present findings suggest that U.S. Mexican girls exposed to trauma are less engaged in problem-focused strategies to manage such stress, which inhibits public prosocial behaviors. The fact that this pattern held for girls and not boys could be due to the relative importance of familism (Carlo, 2014), peer relationships, and social approval in girls during adolescence (e.g., Brown et al., 1999).

In addition, girls' and boys' familism was positively linked to public prosocial behaviors, regardless of trauma exposure. This finding aligns with prior research that familism consistently positively predicts public prosocial behaviors in U.S. Latino/a youth (see Carlo, 2014). Although prior research shows that familism also positively predicts other care-based forms of prosocial behaviors (such as when requested or in crisis situations), familism might orient youth toward helping family members in particular, especially in order to gain family members approval or to comply with family duties (Carlo and de Guzman, 2009). Indeed, encouragement and practice of public prosocial behaviors toward family members earlier in life might play an important role in the subsequent development of other forms of prosocial behaviors and, eventually, in such actions toward non-family members (Zhao et al., 2022). However, research on the developmental trajectories and interrelations of specific forms of prosocial behaviors toward distinct targets across childhood and adolescence will be needed to examine this possibility.

The strength-based approach in the present study also revealed age- and gender-related trends in specific forms of U.S. Mexican youth prosocial behaviors. There were increases in altruistic prosocial behaviors for both genders from adolescence to young adulthood though girls scored higher than boys on both altruistic and public prosocial behaviors. On the other hand, the findings also yielded evidence for declines in public prosocial behaviors for both genders. The overall decline in public prosocial behaviors for both genders (although boys scored higher than girls) complements the earlier reported increases in altruistic prosocial behaviors and suggests a greater consideration of others' needs and less attention to gaining the approval of others and practical concerns during adolescence. Such findings might inform prior work that suggested declines in prosocial behaviors in early to middle adolescence but apparent increases in late adolescence in European heritage youth (Van der Graaff et al., 2018; Shi et al., 2021). The present findings are novel in extending these increases into young adulthood, in specific forms of prosocial behaviors, and in a sample of U.S. Mexican youth. Additionally, the demonstrated evidence of increases of altruistic prosocial behaviors in U.S. Mexican heritage youth across adolescence and into young adulthood run counter to conceptions of pathology and deficits in ethnic/racial minority youth.

Despite the many strengths of the present study, there were several limitations that call for caution in interpretation of the findings. First, the longitudinal study design did not allow for a fully prospective investigation, which tempers our ability to draw confident inferences regarding causality and direction of effects. Other study designs (e.g., intervention studies designed to increase coping and/or familism in trauma-exposed youth) are needed to draw stronger inferences. Second, the present sample focused on U.S. Latino/a youth of Mexican heritage and generalizability of the findings to other U.S. Latino/a groups is limited. And third, the present findings rely on self-reports of trauma, familism, coping, and prosocial behaviors. Studies using multiple methods are desirable to replicate the present findings and reduce concerns with self-presentational demands and shared method variance. Moreover, a number of concerns (e.g., narrowness of assessment, possible confounds between stress and coping) have been raised about the use of checklist measures of major life events and the need to consider alternative assessments (e.g., narrative measures) to better contextualize these constructs (Infurna and Jayawickreme, 2019). Future studies could also examine the impact of additional trauma experiences (such as parental/sibling death) that could have resulted in stronger effects of trauma exposure. Finally, as noted previously, there is evidence that scores on the self-report measure of prosocial behavior are significantly related to behavioral tasks and show changes from prosocial behavior intervention programs. However, replication of the present findings using behavioral, other-reported (e.g., peers, teachers), or observational measures of prosocial behaviors is desirable in future studies.