Introduction

Despite disparate lines of social science research with various conceptualizations of future orientation, findings regarding future orientation as promotive are strikingly consistent across groups, circumstances, and outcomes (Seginer 2009). The idea that humans (and specifically adolescents) are inherently able to orient themselves toward the future has been long established in psychological research (Seginer 2009) and is a cornerstone of health behavior change research and practice (Prochaska and Velicer 1997). Adolescent future orientation is generally regarded as a positive predictor of adolescent academic and health outcomes (Lindstrom Johnson et al. 2014). Future orientation has specifically been linked to higher grade point average among African American high school students through its effect on their perceptions of the usefulness of education and valuing academics (Brown and Jones 2004). Research also consistently has identified positive parenting to be beneficial in the development of adolescents’ future orientation from theoretical (Seginer et al. 2004), empirical (e.g., Kerpelman et al. 2008) and intervention standpoints (Brody et al. 2004).

The present study extends previous research by examining future orientation from a multidimensional perspective with a sample of primarily African American early adolescents who presented for assault injury in the emergency department in two urban, high violence contexts. The primary sources of adolescent exposure to violence in the U.S. include maltreatment and family or community violence (Finkelhor et al. 2015), often in situations which adolescents cannot avoid. Thus, imagining “possible selves,” or mental representations of who they will be and who they hope not to be in the future, represents both a method of cognitively removing oneself from current circumstances and a method of improving self-regulation (Oyserman et al. 2004). Additional research regarding these constructs and the context is necessary, as future orientation might be an important intervention target for youth with low material resources or who might otherwise be viewed as “at risk” (e.g., exposed to violence). The present study employs a person-centered approach to examine several future orientation variables and their associations with aspects of parenting, academic outcomes, and aggressive behavior. Such an approach enables exploration of not only how aspects of adolescent future orientation work together but also how these common patterns are associated with risk and protective factors. Using a similar approach (cluster analysis), Dixson et al. (2017) found differential associations with dimensions of hope and academic outcomes, including school belonging, academic investment, and academic self-concept, suggesting that a person-centered approach might be useful in differentiating latent groups of adolescents both on a wider array of future orientation components and academic outcomes.

Conceptualizations of Future Orientation

Future orientation has been conceptualized in multiple theoretical and disciplinary perspectives including neuropsychology, personality, development, motivation, and cognition (Seginer 2009). Two such conceptualizations relevant to the present study are hope theory and the neural bases of future cognition. Hope theory incorporates several aspects of future orientation (Snyder 2002). From this theoretical perspective, hope can serve to promote positive goals as well as mitigate the risk for negative outcomes and it does so through both cognitive and behavioral components. Specifically, hope theory proposes that individuals utilize mental representations of the pathways available to them (based on past, present, and future circumstances) and cognitions about their capacity to reach those goals to facilitate goal-directed action. Thus, future orientation can represent not only future oriented thinking about positive and negative outcomes, but also action taken in the present to influence future goals.

In addition to this traditional social science conceptualization, recent work has demonstrated that aspects of future cognition (i.e., episodic future thinking) share neural bases with the recall of past events (Schacter et al. 2017). Similar brain regions have been implicated in response to trauma (Hart and Rubia 2012). For example, Maheu and colleagues (2010) found in a preliminary study that adolescents who had experienced neglect in foster care or orphanages expressed heightened medial temporal lobe activation (a region purported to be shared by recall and prospection) when processing a threatening facial cue. More work is needed, but the possibility that brain regions involved in future thinking are also affected by traumatic experiences suggests that research at this intersection will be helpful in understanding pathways by which violence exposure influences outcomes related to cognition and self-regulation.

Future Orientation and Youth Exposure to Violence

A range of social science research has consistently demonstrated the connections of adolescent future orientation with exposure to and involvement in violence. The evidence suggesting strong future orientation (from various conceptualizations) is an important protective factor for youth exposed to violence is well established. Future life expectations, including fatalism, have been related to avoidance of aggressive and delinquent behavior generally (Chen et al. 2016), and an intervention focused on future goals, barriers, and planning reduced fighting behavior and marijuana use (Lindstrom Johnson et al. 2015). Further, both future educational expectations and fatalism (inversely in separate analyses) mediated the relationship between experiencing adverse childhood events and violent behavior among a large, U.S.-representative sample (Brumley et al. 2017). This is especially important given the dose-response relationship between adverse childhood experiences (many of which are related to violence exposure) and negative health outcomes (Anda et al. 2006). Additionally, work exploring adolescents’ future orientation in urban, low-income contexts, found that career-focused future thinking is associated with lower engagement in violent behavior in youth considered academically “at risk” (Stoddard et al. 2011). Similar work suggests that hopefulness mediates the relationship between adolescents’ perceptions of their family’s closeness and their involvement in violent behavior (Stoddard et al. 2011).

More recent studies reported mixed findings about the role of future orientation as a protective factor against involvement in delinquent and aggressive behavior for urban youth exposed to community violence. Using a future orientation composite score including perceived control, positive attitudes, and hopelessness, So et al. (2018) found that future orientation buffers the effect of violence exposure on delinquent (e.g., stealing, skipping school, fighting) but not strictly aggressive behavior. Another recent study originating from an afterschool violence prevention program found that future time perspective (FTP) was not significantly predictive of violent behavior when also considering present time orientation and subjective norms about violence, both of which positively predicted violence involvement (Kruger et al. 2018). These studies represent important substantive contributions about the nuanced, but generally promotive, role future orientation may play as a protective factor specifically for urban youth exposed to violence.

However, research has only recently emerged regarding the implications of exposure to violence on the development of future orientation. Addressing this gap, Monahan and colleagues studied juvenile offenders, a group likely to have increased exposure to violence (Monahan et al. 2015). Their findings indicated that exposure to violence in adolescence through young adulthood was related to less growth in “future outlook,” a composite of time perspective, consideration of future consequences, and optimism, but not with developmental changes in impulse control. This supports earlier work suggesting that these regulatory and cognitive components of behavior might operate through different mechanisms (Steinberg et al. 2009).

Parenting as a Contextual Promotive Factor

Despite exposure to violence placing adolescents at risk, the parental relationship can serve as a protective factor generally as well as specifically promote adolescents’ future orientation. Seginer (2008) theorizes that positive parenting practices influence the development of several aspects of future orientation indirectly through self-esteem. Further, from the bioecological perspective (Bronfenbrenner and Morris 1998), parents represent proximal drivers of psychosocial development in adolescents’ environments. Focusing primarily on the affective component of parenting (i.e., positive and negative parenting), empirical studies have consistently found positive aspects of parenting to be promotive of future orientation for adolescents irrespective of conceptualization (Seginer 2009).

McCabe and Barnett (2000) found that familial social support and mother involvement significantly predicted African American adolescents’ perceptions of their control over their futures and salience of future planning, respectively. This supports pioneering work suggesting that adolescents who perceived more supportive parenting were more optimistic and hopeful about their futures than those who perceived less parental support (Trommsdorff 1983). More specifically, adolescents’ perceived maternal (but not paternal) support was found to be predictive of African American adolescents’ education-related future oriented thoughts and planning, controlling for the significant influence of both self-efficacy and ethnic identity (Kerpelman et al. 2008).

Research Questions and Hypotheses

The present study employs a person-centered approach to empirically identify underlying patterns of adolescents’ future orientation. Such an approach enables exploration of not only how aspects of adolescent future orientation work together but also how these common patterns are associated with risk and protective factors. This study addresses the following research questions: Do assault-injured urban adolescents fall into substantively distinguishable profiles of responses to future orientation variables? Does parenting influence profile membership? Can these profiles be differentiated based on participants’ academic, aggressive, and delinquent behavior? Latent profile analyses (LPA) was employed to examine underlying patterns of responses to adolescents’ reports of their hopes, goals, expectancies, commitment to learning, and fatalism. Parental warmth and strictness were modeled as predictors of profile membership and differences by profile in academic achievement, school skipping, school suspensions, fighting, and aggressive behavior were explored.

Given the recent findings of the differential impact of future orientation on coping with violence (So et al. 2018) and violence involvement (Kruger et al. 2018), it was anticipated that different measures might be tapping into different aspects of future orientation and, therefore, that individuals might group together as high in some measures of future orientation and low in others. It was further predicted that these profiles would be conceptually distinct such that parental warmth would be positively-related and strictness negatively related to participants’ classification within the conceptually promotive profile (e.g., McCabe and Barnett 2000). It was generally expected that a pattern that clearly distinguished participants with positive academic outcomes and less aggression from those with negative academic outcomes and more aggression, as similar patterns have been observed by Dixson and colleagues (2017) in their person-centered study and others in variable-centered studies (e.g., Stoddard et al. 2011). These findings suggest that a person-centered approach might be useful in understanding patterns of future orientation within individuals and how this complexity can be used to design interventions to improve health and educational outcomes of adolescents who have been exposed to violence. This is critically important to understand in an early adolescent sample both as they have lower future orientation than older adolescents (Steinberg et al. 2009) and as their academic and behavioral outcomes have an impact on their future possibilities.

Method

Participants

Participants were recruited from two hospital emergency departments in the cities of Baltimore, Maryland and Philadelphia, Pennsylvania from June of 2014 through June of 2016. Early adolescents (ages 10 to 15 years) who presented at the emergency department for an interpersonal intentional injury, excluding child abuse, sexual abuse, sibling fights, other domestic violence-related fights involving household members and police fights. Additionally, adolescents and parents/guardians had to be able to speak and understand English and have the ability to participate in a mentoring intervention. Adolescents had to reside with parents/guardians and the caregiver could not currently be involved in an ongoing custody dispute or Child Protective Services investigation. Participants were 188 primarily (95.7%) African American early adolescents (Mage = 12.87, SDage = 1.52). Participants were 60.6% male and primarily low-income, as evidenced by high rates (89.3%) of receipt of public assistance for medical care. Most assaults did not involve weapons (84.5%) and only one adolescent was hospitalized as a result of their injury. The aggressors reported by participants were most frequently classmates (41.7%) or acquaintances (24.6%). Data analyzed for this study were collected at baseline, prior to randomization for the mentoring intervention that was the purpose of the larger study.

Procedure

Research staff used the emergency department electronic patient tracking system and medical records (i.e., age, chief complaint, diagnosis, and triage notes) to identify eligible families. Families were then contacted by mail and phone to provide information about the study and further assess eligibility. Seven hundred visits (688 adolescents) were identified through review of emergency department records. Of these 688, 105 adolescents were determined to not meet study inclusion criteria, 247 were unable to be contacted (i.e., incorrect contact information or non-response), and 148 declined to participate (i.e., too busy, too traumatic, not interested). A study visit, either at home or at the hospital/clinic, was conducted with families in order to obtain written parent/guardian consent and adolescent assent. The Institutional Review Boards of both participating hospitals approved all study procedures.

At the study visit parents and adolescents each completed a separate in-person interview and an audio-computer assisted self-interview (ACASI) questionnaire. The interview collected information about demographics and the circumstances of the injury. The ACASI questionnaire asked about potentially more sensitive information, including information about adolescent behaviors and parenting practices. This data collection strategy was used to address concerns about literacy level as well as possible response bias. The data collection process took approximately 1 h for the family to complete and families received $40 in gift cards as remuneration for their time. Additionally, a research assistant used an electronic abstraction guide to log relevant information from the emergency department medical record of the target visit.

Measures

Demographic characteristics

Demographic characteristics for each participant were collected primarily through abstraction of emergency department patient records. Age in years was centered at 10. Gender was coded 0 (male) and 1 (female). Parental education was coded 0 (high school/GED completion or less) and 1 (some college or more).

Measures of future orientation

Commitment to learning

Commitment to learning was measured using a subscale from the Developmental Assets Framework (Scales and Leffert 1999). The measure was a composite of 15 items (α = .90) and adolescents responded the degree to which items described them, with response options ranging from 1 (Not at all or Rarely) to 4 (Extremely or Almost always). Examples of items include, “Actively engaged in learning new things,” “Eager to do well in school and other activities,” and “Plan ahead and make good choices.”

Hope

Hope was assessed using items from Child Trends’ Flourishing Children Project (Lippman et al. 2014). Adolescents were prompted: “Please indicate how much these statements describe [themselves].” Items were “I expect good things to happen to me,” “I am excited about my future,” and “I trust my future will turn out well.” Response options ranged from 1 (Not at all like me) to 5 (Exactly like me). The three adolescent-reported items were averaged to create a scale (α = .64).

Goal orientation

Goal orientation was assessed using the mean of seven adolescent-reported items (α = .66) from Child Trends’ Flourishing Children Project (Lippman et al. 2014). Examples include, “I have goals in my life,” “I develop step-by-step plans to reach my goals,” and “How often do you have trouble figuring out how to make your goal happen?” The former two items exemplify those five items in which adolescents were prompted to indicate how much the statement describes them and for which the response options ranged from 1 (not at all like me) to 5 (exactly like me). The latter item exemplifies the two items in which adolescents were asked how often the situation occurs for them and for which response options ranged from 1 (none of the time) to 5 (all of the time). The last example item was reverse coded such that a higher mean score on the scale indicated higher goal orientation.

Fatalism

Fatalism was assessed with the item, “Do you think you will live to 35?” adapted from AddHealth (Harris et al. 2009). Response options were “yes,” “no,” and “maybe.” Because respondents utilized only the “maybe” and “yes” options, they were recoded as dichotomous with a higher score indicating a less fatalistic view.

Expectancies

Expectancies were measured using Eccles and Barber’s (1993) 11-item life expectancies scale. Items included expectancies about future education, career, family, and finances. Example items include: “You will go to college,” “You will get the job you want,” “You will be unemployed,” and “You will have a child out of marriage”. Response options ranged from 1 (Very likely) to 7 (Not very likely), and positive expectancies were reverse coded such that higher average scores across all items reflected higher perceived likelihood of positive events and lower likelihood of negative events (α = .73).

Parenting correlates of profile membership

Aspects of parenting were measured using the adolescent-reported parental attitude scale (Lamborn et al. 1991). The parental warmth and involvement subscale was created by averaging nine items (α = .75). Examples include: “I can count on my parents to help me out if I have some kind of problem,” “When my parents want me to do something, they explain why,” and “My parents spend time just talking with me.” Responses for the warmth/involvement subscale were measured on a 0 (strongly disagree) to 3 (strongly agree) scale, such that a higher scores indicated adolescents perceiving more warmth/involvement from their parents. The parental strictness and supervision subscale was created by averaging six items (α = .89), including 3 items regarding the extent to which children perceive their parents attempt to know about how the child spends their free and unstructured time and three items regarding children’s perceptions of how much their parents actually know about how they spend that time. Response options for the parental strictness/supervision subscale were: 0 (don’t try or don’t know), 1 (try a little or know a little), and 2 (try a lot or know a lot).

Concurrent outcomes

Academic outcomes

Three measures of academic behaviors (school skipping, suspensions, and academic achievement) were assessed based on similar measures used in the National Longitudinal Study of Adolescent to Adult Health (Harris et al. 2009). School skipping was measured using the adolescent self-reported item, “During the past 12 months, how many times have you skipped or cut school?” Suspensions were measured using the parent-reported item, “How many times has your child been suspended from school in the last year?” Participants provided counts of skipping events and suspensions, which were respectively coded as 0 (no skipped days or no suspensions), 1 (1–2 skipped days or 1–2 suspensions), and 2 (3 or more skipped days or 3 or more suspensions) for the purpose of analyses, as statistical models to address the zero-inflated nature of these variables are not yet well documented in the context of person-centered latent variable approaches. Parents reported their adolescent’s academic achievement (i.e., grades on the most recent report card) on a scale of 1 (D’s or lower) to 7 (A’s).

Violence outcomes

Fighting was measured with the following item adapted from the Youth Risk Behavior Surveillance questionnaire (CDC 1991–2019), “Not including your ER visit, how many times were you in a physical fight with persons other than your brother(s) or sister(s) in the last 12 months?” Adolescents responded with the discrete count of physical fights. Aggression was measured using the modified aggression scale (Bosworth and Espelage 1995), which asked adolescents how often in the past 30 days they had exhibited specific aggressive behavior. Specific aggressive behaviors were generally directed at other kids and included, for example, encouraging fighting, hitting back or hitting in anticipation of being hit, teasing, threatening, spreading rumors, and name-calling. Response options ranged from 0 (never) to 5 (5 or more times). The scale exhibited good internal consistency (α = .92).

Analytical Strategy

Latent profile analysis was implemented to determine patterns of responses to the future orientation variables commitment to learning, goals, hope, expectancies, and fatalism, as well as modeling auxiliary variables as predictors of profile membership and differences in outcomes by profile. Analyses were conducted using Mplus version 7 (Muthén and Muthén 1998–2019). Five latent variable models were analyzed: models with between one and five latent profiles with 1000 random starts, 100 final stage optimizations, and 1000 iterations. The five-profile solution included profiles with less than 10% of the sample in a single profile, so solutions with more profiles were not estimated. Several fit indices and tests were used to select a solution. These included: Akaike Information Criterion (AIC; Akaike 1987), Bayesian Information Criterion (BIC; Schwarz 1978), Adjusted BIC (Sclove 1987), Lo-Mendell-Rubin Likelihood Ratio Test (LMR-LRT; Lo et al. 2001), and Adjusted LMR-LRT. Improvements in model fit were judged by lower AIC, BIC, and Adjusted BIC. The likelihood ratio tests assess improvement in model fit for nested models by comparing a solution with a solution with one less profile. A significant p-value for the LMR-LRT indicates that the addition of a profile improves model fit (Nylund et al. 2007). Further, each solution was reviewed for substantive interpretability.

Two approaches were used to analyze the relations of the future orientation profiles to auxiliary variables (i.e., predictors, covariates, outcomes), as the substantive research questions posed in this study were related to both the prediction of individuals’ profile membership as well as the prediction of distal outcomes from profile membership. Multinomial logistic regression was conducted Mplus using the R3STEP procedure to model parental warmth and strictness, as predictors of profile membership. Age, gender, and parental education were also included as covariates, as certain aspects of future orientation have been found to differ by age and gender (Steinberg et al. 2009), and increased parental education has been found to influence parental warmth (Davis-Kean 2005). Exposure to community violence was included as a covariate due to its relationship with future orientation and the outcomes, but given its lack of significance the more parsimonious model is presented in this paper. R3STEP regresses the categorical latent profiles on variables of interest in an automatic 3-step process in which (a) an LPA solution is estimated, (b) posterior probabilities of profile membership are saved, and (c) multinomial logistic regression is conducted (Asparouhov and Muthén 2013). The BCH method (described by Vermunt 2010) was implemented to assess mean differences between profiles in academic achievement, fighting, aggression, and school skipping and suspensions. Mean differences in school skipping and suspensions were also tested using raw count data. Several methods exist for modeling auxiliary variables in the LPA framework, and suggestions for the selection of such methods are still being developed as this is an ongoing area of statistical research. Previous versions of these analyses attempted model the academic and violence outcomes by utilizing most likely profiles as predictors in zero-inflated negative binomial models but were not successful due to convergence issues. Both the R3STEP and BCH methods take into account error in the classification of individuals into latent profiles, and thus represent an improvement over categorizing participants into their most likely classes to conduct additional analyses.

Results

Descriptive Analyses

The four continuous future orientation variables (commitment to learning, goals, hope, expectancies) were each significantly (at α = .01) and positively correlated (r = .21–.59) with other future orientation variables, with the lowest being expectancies with commitment to learning and hope, respectively. Regarding fatalism, 68.6% of participants said they think they will live to age 35, and the remainder responded “maybe.” The parental warmth/involvement and strictness/supervision scales were also significantly and positively correlated (r = .26–.35) with the future orientation variables except expectancies, which was not significantly correlated with the parenting variables. Table 1 provides means, standard deviations, and correlations among continuous study variables.

Table 1 Correlations between continuous future orientation and parenting variables
Full size table

Latent Profile Model Selection

Fit for each model was assessed to determine the number of profiles to retain; these indices are presented in Table 2. At the traditional α = .05 cutoff, LMR-LRT for four- and five-profile solutions were clearly not significant. LMR-LRT and adjusted LMR-LRT suggested selecting the two-profile solution; however, p-values for a three-profile solution were fairly close to .05. Additionally, the log-likelihood, Akaike information criterion, and Bayesian information criterion indices continued to improve between the three and four profile solution. The substantive meaning of the profiles was also examined, and the three- and four-profile solutions were found to be appropriate. Because it seemed that the addition of a fourth profile simply split some participants from the low and high classes to create a medium profile not substantively distinct from either of the existing profiles, the three-profile solution was selected. The three-profile solution had entropy of .84, indicating good profile separation. Probabilities of profile membership were, on average, .87, .93, and .94, indicating high probabilities of participants being assigned into their most likely classes.

Table 2 Model fit statistics for latent profile analyses
Full size table

Description of Latent Profiles

The three latent profiles can be distinguished by inspecting the standardized means and frequency of the future orientation variables included as indicators. The descriptive characteristics for the indicators of each profile are provided in Fig. 1. Generally, the profiles can be described as following a pattern of low, medium, and high responses. However, the medium profile did not follow this pattern for expectancies. Thus, the profiles were termed Low Future Orientation, High Future Orientation, and Discordant Future Orientation, with the latter profile characterized by mid-range responses to all indicator variables except expectancies, for which respondents in this profile scored lowest. Participants were most likely to be classified into the High profile (61.7%), followed by the Discordant (27.1%) and Low (11.2%) profiles.

Fig. 1
figure 1

Standardized means of indicators presented by profile. Error bars represent ± standard error of the standardized mean. Fatalism indicator is presented as a percentage below the legend

Full size image

Demographic and Parenting Factors as Predictors of Latent Profile Membership

Results of multinomial logistic regression (profile membership on risk and protective factors and covariates) are provided in Table 3. Gender, age, and parental education did not significantly predict profile membership over and above other predictors but were retained as covariates due to potential differences in parenting and dimensions of future orientation on these factors. Parental strictness and supervision was also not a significant predictor of profile membership. Increases in parental warmth predicted significant increases in the log odds of participants being classified in the High or Low profiles in comparison to the Discordant profile.

Table 3 Results of multinomial logistic regression of profile membership on parental warmth and strictness
Full size table

Concurrent Outcomes

Exploratory results of profile comparisons of academic and violence-related outcomes are provided in Table 4. Significant differences were found when comparing the High profile and Low profile across all three academic outcomes. Adolescents in the High profile earned higher grades on average and had fewer suspensions and skipped days than adolescents in the Low profile. While the differences were not significant when compared to either the Low or High profiles, adolescents in the discordant group reported the highest grades on average but also the most suspensions and skipping. Analyses of mean differences in the count data for school skipping and suspensions did not reveal significant differences, but were also unable to account for the zero-inflated nature of the data. Participants in the High profile also had significantly higher grades than those in the Low profile. Finally, participants in the Low profile scored significantly higher on the aggression scale than participants in the High profile. While the mean number of fights was clearly highest for participants in the Low profile, this difference was not significant at α = .05.

Table 4 Mean differences of academic and violence-related outcomes across profiles
Full size table

Discussion

Previous research on the associations of adolescent future orientation and exposure to violence have taken a unidimensional approach to future orientation. The present study implements a person-centered approach (LPA) to understand profiles of future orientation from a multidimensional perspective among a sample of primarily African American early adolescents who have been injured by assault. The future orientation profiles were examined for their associations with parenting practices and adolescents’ academic outcomes and involvement in violence-related behavior. The findings reported in this study extend research that has found that adolescents who are more oriented toward their futures generally had more positive health and academic outcomes than their peers who are less future oriented, and positive parenting was associated with greater future orientation for adolescents (McCabe and Barnett 2000). This is in the context of recent research which has found that the relationship between future orientation and violence involvement might be more nuanced for urban youth exposed to community violence in terms of both coping with violence exposure (So et al. 2018) and predicting involvement in violence (Kruger et al. 2018). The present findings have intervention implications, as being low or discordant across several dimensions of future orientation might be a psychological marker for risk and enable targeted interventions to promote future-oriented cognition and action for adolescents as well as improved parenting. Additionally, a sole focus on adolescents who have been identified as particularly at-risk or resilient might mean missing a prime opportunity to intervene with those who are hopeful and have future goals but might need resources or material support to realize their goals.

The first aim of the present study was to identify subgroups of adolescents based on their response to five future orientation variables. Rather than qualitatively assign adolescents to subgroups based on their responses, LPA was employed to empirically identify the patterns. Participants could be classified into three distinguishable profiles characterized by responding high or low on all future orientation variables or moderately on all variables except expectancies (i.e., the Discordant profile). Similar to other person-centered analysis (e.g., Dixson et al. 2017), the “high” group was most prominent. However, the present study found the existence of a Discordant profile in which participants are distinguishable based on having low expectancies despite having hope and goals, being committed to learning, and not endorsing fatalistic attitudes. This group may be motivated but may rightly or wrongly perceive barriers to attaining their desired future outcomes. This is consistent with the educational “aspiration-expectation gap” for urban youth, in which some adolescents who aspire to higher levels of education, they do not anticipate that they will meet these goals (Perry and Raeburn 2017).

In addition to identifying latent profiles, a primary aim of this study was to understand how membership in these profiles are predicted by parental warmth and strictness. Parental warmth expectedly predicted a higher likelihood of being in the High versus Discordant profile. Conversely, increased parental warmth also predicted increased likelihood of being in the Low versus Discordant profile. Taking into account the intercept, adolescents were effectively equally likely to be classified as Low or Discordant with increased parental warmth. While unexpected, parental warmth might act on the domains of future orientation that are individually-controlled (e.g., hope, goals, fatalism), and expectancies might be reflective of adolescents’ understandings of their access to material resources. Meanwhile, parental strictness was not a significant predictor of profile membership. The research literature regarding parental warmth versus monitoring as a support for future orientation is limited (Lowe and Dotterer 2013), and more work in this area is needed. From a developmental perspective, early adolescence might represent a developmental period during which individuals are primed to benefit from interventions to increase parental warmth, as they still spend a large portion of their time with their parents. Improvements in parental warmth could mitigate the risk of negative effects of violence exposure on the development of future orientation (Monahan et al. 2015), thus improving trajectories of future orientation (and ultimately academic and delinquent outcomes).

Finally, this study aimed to determine whether the established latent profiles could be differentiated based on academic and violence outcomes. The High and Low profiles could be differentiated across all outcomes except fighting. This is not surprising given the breadth of literature suggesting that future orientation is a positive factor in adolescents’ academic outcomes, including a person-centered study of the relationship of dimensions of hope with a host of academic outcomes (Dixson et al. 2017). It is, however, interesting that significant differences in fighting were not distinguishable across profiles. Given the context of the present sample, it might be that some baseline level of fighting is normative or protective in the community setting (Salzinger et al. 2002), as both the High and Discordant profiles reported more than one fight on average. In situations of imminent danger, immediate self-preservation takes precedence over concern about future consequences (cf. Wilson and Daly 1997). This is consistent with So and colleagues (2018) findings that adolescents who were both more future oriented and coped with violence by avoiding violent areas were less likely to engage in aggressive and delinquent behavior. Most research, however, has considered aggression more generally rather than fighting specifically, so more work is needed to elucidate the role of future orientation in preventing specific forms of delinquency and violence. However, Lindstrom Johnson and colleagues (2015) randomized controlled trial of a motivational interviewing intervention reduced fighting by 63%.

Participants in the Low and Discordant profiles skipped school more frequently than those in the High profile. This finding is consistent with previous work showing that having greater FTP is associated with higher academic engagement for Black adolescents (Brown and Jones 2004). Although the difference in skipping for High versus Discordant adolescents was significant only at p < .10, this result suggests that school officials like truancy officers or attendance clerks might interface with adolescents in the Discordant profile in a way that has meaningful implications for intervention. Specifically, given their positive attitude towards learning and the future, this might be an important group to focus dropout prevention initiatives. It also might suggest the importance of being trauma-informed, given the influence of violence exposure on neural systems (Maheu et al. 2010) and the development of future orientation (Monahan et al. 2015).

This study was limited in several ways. First, the cross-sectional nature of the data does not allow for investigation of directional relations between the hypothesized predictors and outcomes. Second, due to limitations in both the state of research regarding modeling auxiliary variables with LPA and available measures of academic outcomes, the present analysis of auxiliary academic outcomes was primarily exploratory. Third, this sample is relatively small and was selected based on having been injured by assault. This impacts the ability to distinguish groups, which may be particularly relevant for the discordant group (e.g., large standard errors). While the nature of the sample limits the generalizability of this study’s findings, this sample represents an important group of youth who might be at higher risk for attenuated educational aspirations and increased violence involvement given their own exposure to violence (i.e., assault injury) as adolescents. In a general sample, it might be that more adolescents would fall in a middle profile or that additional profiles would exist. It could be that these adolescents have moved to the more “extreme” profiles (High and Low) as a function of their injuries. For example, those who are coping with the injury by becoming more future oriented might be more likely to be in the High profile, and those who are not coping with the injury well might be more likely to be in the Low profile. Additional work in the vein of that by So and colleagues on outcomes beyond delinquency will be beneficial in understanding the role of future orientation in coping for adolescents who have been assault-injured.

Notwithstanding these limitations, the present study has important implications for practice and more generally for adolescent development. The findings of this study highlight that research and practice focused on identifying students who are particularly “promising” or “at risk” based on perceptions of their futures might miss a proportion of students who might be perceived as academically competent (e.g., having good academic performance) and oriented toward the future but might be at risk of skipping school or being suspended and having low expectancies about their futures (the Discordant group). This could be important for schools seeking to implement trauma-informed strategies, as parents, educators and school interventionists might need to pay specific attention to providing material resources and removing educational barriers for these students rather than attempting to simply instill hope or encourage students to envision their futures, as even those students who perceive having hope and goals but have low expectancies are potentially at risk for skipping, getting suspended, getting into fights, or otherwise exhibiting aggressive behavior. Practitioners might also consider the inclusion of future orientation in their efforts to identify at risk students in an effort to facilitate early intervention. The present study reinforces the importance of future orientation more generally, but also highlights that the construct can be distinguished in into multiple profiles among early adolescents, who have previously been found to have lower future than their older adolescent peers. This variability suggests that factors from earlier life influences future orientation in early adulthood and might influence differential changes in future orientation over time.

Future studies should examine whether these profiles hold with larger and different samples, as well as whether parenting remains predictive of the profiles and whether profiles can be differentiated on academic and violence involvement outcomes. This is especially the case for violence involvement, as Kruger and colleagues (2018) found that FTP was not associated with early adolescent violent behavior above the effects of present time perspective and subjective norms about violence. Thus, adolescents’ perceptions of pressure to engage or not engage in violent (or other deviant) behavior might importantly represent their understanding of their context and have implications for both future orientation directly and interact with future orientation to influence involvement in violent behavior. That is, contextual factors (including perceptions of context) and individual promotive factors (e.g., ethnic identity; Kerpelman et al. 2008) represent other important variables in understanding the relationship of future orientation with academic and violence-related outcomes. Studies might also consider additional future orientation variables, like FTP, as indicators of latent profiles. Future studies must also understand the development and influencing factors of future orientation throughout childhood, adolescence and adulthood. Future research should build upon this work to understand whether these findings are consistent across samples of adolescents who have and have not been exposed to violence (including other types of violence) and whether profile membership is related to having been exposed to violence.

Conclusion

Previous research has approached future orientation from a unidimensional perspective and, thus, has not explored how a multidimensional perspective can inform how researchers understand the promotion of future orientation through the parental relationship and how future orientation or the lack thereof can serve as a risk or protective factor relative to academics and violence-related behavior. Using latent profile analysis, the present study identified three patterns of future orientation among early adolescents exposed to violence (High, Low, and Discordant) and connected these profiles to parental relationships and adolescents’ academic and violence-related behavior. The profiles notably included a Discordant profile of future orientation in which participants were average on all future orientation variables except expectancies, for which they scored the lowest of the three profiles; participants in the two other profiles scored high or low on all future orientation variables, respectively. Both the Low and Discordant represent groups of adolescents who might be at risk for negative future outcomes and a target for intervention. However, students represented by the Discordant profile might not be targeted for intervention unless they also exhibit more traditional risk indicators, like school skipping. The ability to distinguish early adolescents into profiles of future orientation supports the developmental appropriateness of research on future orientation during this developmental period. This could be important for schools seeking to implement trauma-informed strategies, as parents, educators and school interventionists might need to pay specific attention to providing material resources and removing educational barriers for these students rather than attempting to simply instill hope or encourage students to envision their futures, as even those students who perceive having hope and goals but have low expectancies are potentially at risk for skipping, getting suspended, getting into fights, or otherwise exhibiting aggressive behavior. The present study reinforces the importance of future orientation more generally, but also highlights that the construct can be distinguished in into multiple profiles among early adolescents, who have previously been found to have lower future than their older adolescent peers.