Over the past 50 years, families in the U.S. have faced diverging economic fates. Wealth inequality has exploded, with the gap between the richest and poorest Americans more than doubling since the late 1980s (Schaeffer, Reference Schaeffer2020). Income inequality too is on the rise, with upper-income households bringing in increasingly larger shares of the country’s total income (Schaeffer, Reference Schaeffer2020). Moreover, the percentage of American’s experiencing income volatility has grown (Dynan et al., Reference Dynan, Elmendorf and Sichel2012; Menasce Horowitz et al., Reference Menasce Horowitz, Igielnik and Kochhar2020; Moffitt & Zhang, Reference Moffitt and Zhang2018; Weller, Reference Weller2018). Low- and middle-income families face substantial swings in monthly income; one report found that these families, on average, experienced five months of income volatility (i.e., monthly income change of ±25%) within a single year (Hannagan & Morduch, Reference Hannagan and Morduch2015). These trends have led to a rise in the prevalence of families experiencing economic hardship (Morduch & Schneider, Reference Morduch and Schneider2017; Schaeffer, Reference Schaeffer2020). Economic hardship, as we define it, encompasses both material deprivation, which encapsulates the extent to which families lack necessary or important materials or experiences, and subjective feelings of financial stress, which represents the psychological distress individuals experience as a result of not having enough money to meet needs (Schenck-Fontaine & Panico, Reference Schenck-Fontaine and Panico2019). In 2018, nearly 40% of all families reported having trouble paying for housing, utilities, food, and/or medical care, and among low-income families this number was 60% (Karpman et al., Reference Karpman, Gonzalez and Kenney2020). Moreover, the COVID-19 pandemic intensified the breadth and depth of economic hardship experienced by American families (Menasce Horowitz et al., Reference Menasce Horowitz, Igielnik and Kochhar2020). For example, food insecurity rose by 25% and financial instability grew by 20% during the latter part of 2020 (Cooney & Shaefer, Reference Cooney and Shaefer2021; Hardy & Logan, Reference Hardy and Logan2020), and experiences of economic hardship crept into the middle class (Menasce Horowitz et al., Reference Menasce Horowitz, Igielnik and Kochhar2020). With upper-income families pulling further away financially from middle- and lower-income families, financial pressure is increasingly common in all but the wealthiest of households (Menasce Horowitz et al., Reference Menasce Horowitz, Igielnik and Kochhar2020; Schaeffer, Reference Schaeffer2020).
At the same time, there has been an unprecedented rise in mental health problems among American youth (CDC, 2021; Murthy, Reference Murthy2022). Given documented links between income inequality, economic hardship, and children’s behavioral functioning (Edmunds & Alcaraz, Reference Edmunds and Alcaraz2021; Gershoff et al., Reference Gershoff, Aber, Raver and Lennon2007; Odgers, Reference Odgers2015; Piera Pi-Sunyer et al., Reference Piera Pi-Sunyer, Andrews, Orben, Speyer and Blakemore2023; Schenck-Fontaine & Ryan, Reference Schenck-Fontaine and Ryan2022), the exacerbation of economic inequality and hardship may be a contributing factor to the increase in these problems. Recent evidence indeed suggests that low-income children and children living in families experiencing economic precarity and food insecurity were one of several subgroups disproportionately impacted by the recent increase in adolescent mental health issues (CDC, 2021; Murthy, Reference Murthy2022; Osgood et al., Reference Osgood, Sheldon-Dean and Kimball2021). Additional research probing the associations between hardship and behavior problems during adolescence is crucial for two reasons. First, major social cognitive and sociocultural changes occur in adolescence, including improvements in perspective-taking and metacognition, elevated emotion reactivity, and increased understanding of societal structures around socioeconomic status (Crone & Dahl, Reference Crone and Dahl2012; Kraus et al., Reference Kraus, Piff, Mendoza-Denton, Rheinschmidt and Keltner2012; Stephens et al., Reference Stephens, Markus and Phillips2014; Weimer et al., Reference Weimer, Warnell, Ettekal, Cartwright, Guajardo and Liew2021). Second, adolescence is a time of high vulnerability for mood and emotional issues, substance abuse, and psychiatric disorders. Indeed, half of all lifetime mental health issues manifest by 14 years of age (Costello et al., Reference Costello, Copeland and Angold2011; Kessler et al., Reference Kessler, Berglund, Demler, Jin, Merikangas and Walters2005), and the prevalence of mental health issues in adolescents is currently on the rise (Murthy, Reference Murthy2022). As such, adolescents’ understanding of their economic circumstances may be keenly relevant to their experiences of hardship and their mental health (McLoyd et al., Reference McLoyd, Kaplan, Purtell, Bagley, Hardaway and Smalls2009; McLoyd, Reference McLoyd1990, Reference McLoyd, Carlo, Crockett and Carranza2011). Yet, much of the research only considers caregiver reports of hardship despite that youth’s own perceptions of hardship may directly relate to their behavior above and beyond their caregivers’ perceptions (Delgado et al., Reference Delgado, Killoren and Updegraff2013; Mistry et al., Reference Mistry, Benner, Tan and Kim2009; Wadsworth & Berger, Reference Wadsworth and Berger2006). This may be an especially important oversight in the literature because by adolescence, youth have gained better understanding of family finances and financial stress, as well as socioeconomic status.
Furthermore, there are two methodological limitations in extant studies that further limit our understanding of links between hardship and youth behavioral functioning. First, many of the studies have failed to get multiple measures hardship and behavioral outcomes, which ignores instability in these constructs. Moreover, longitudinal studies have generally used hardship measures that predated measures of youth behavioral outcomes by several years, which is problematic if these constructs are not stable (Gard et al., Reference Gard, McLoyd, Mitchell and Hyde2020; Simons et al., Reference Simons, Lei, Beach, Philibert, Cutrona, Gibbons and Barr2016). Recent research shows that month-to-month variability in economic resources, including economic hardship, is common in U.S. households (Hannagan & Morduch, Reference Hannagan and Morduch2015; Liu et al., Reference Liu, Zalewski, Lengua, Gunnar, Giuliani and Fisher2022) – variation that prior studies have largely ignored. While extant studies strived to establish clear temporal precedence, if the underlying processes are unfolding on shorter durations than the assessment interval, prioritizing temporal precedence may be problematic. In particular, if the lag structure does not reflect the underlying dynamic processes that are at play in the lives of families and youth, this approach may obfuscate associations. Indeed, when constructs vary substantially over time, it is important to measure and model the pace of change over the proper time scale instead of prioritizing lagged associations (Granger, Reference Granger1969).
Using an intensive longitudinal, repeated measures design, this study samples 104 adolescent–caregiver dyads and collects monthly measures of caregiver and youth reports of hardship, including parent reports of material deprivation (e.g., utilities cutoff for nonpayment, eviction, housing inadequacies) and youth reports of material deprivation relevant to their lives (e.g., unable to participate in school activities because of lack of money) and financial stress (e.g., worry about caregiver because money is limited), and youth behavior problems (externalizing and internalizing) over a period of nine months to advance the current literature by addressing two aims. First, we examine the month-to-month variability of reports of both caregiver and youth perceptions of hardship and youth behavior problems. Second, we estimate associations between variation in monthly measures of caregiver- and youth-reported hardship and youth internalizing and externalizing behavior problems. The results provide novel information about a critical process driving changes in problem behaviors during adolescence.
Economic hardship predicts youth behavioral functioning
Economic hardship includes material deprivation, which denotes the households lack of vital goods and experiences, like food, adequate housing, and medical care (Gershoff et al., Reference Gershoff, Aber, Raver and Lennon2007; Schenck-Fontaine & Ryan, Reference Schenck-Fontaine and Ryan2022; Thomas & Waldfogel, Reference Thomas and Waldfogel2022), as well as subjective feelings of financial stress, which encompass the psychological distress caregivers and adolescents may feel upon evaluating their economic circumstances (Schenck-Fontaine & Panico, Reference Schenck-Fontaine and Panico2019). Both material deprivation and feelings of financial stress pose risks to adolescents’ wellbeing. While studies have documented the indirect effects that financial stress may have on children through parents and the family system (Conger et al., Reference Conger, Ge, Elder, Lorenz and Simons1994), both material deprivation and financial stress can have direct effects, such as when lack of medical care or worry about caregivers’ lack of money compromises children’s mental health (Thomas et al., Reference Thomas, Miller and Morrissey2019). Indeed, research shows that material deprivation and financial stress are highly stressful for adolescents (Ennis et al., Reference Ennis, Hobfoll and Schröder2000; Wadsworth et al., Reference Wadsworth, Raviv, Compas and Connor-Smith2005). When it comes to their behavioral functioning, this stress is a source of anxiety, frustration, and demoralization and can also manifest in increased conflict between youth and parents, other family members, and/or peers, all of which can lead to internalizing and externalizing problems (Wadsworth et al., Reference Wadsworth, Raviv, Compas and Connor-Smith2005). Moreover, stress triggers a physiological response, including the release of catecholamines and glutocorticoids (Gunnar & Quevedo, Reference Gunnar and Quevedo2007; Steeger et al., Reference Steeger, Cook and Connell2017). Frequent activations of the stress response could result in dysregulated stress responses, that is, hyper- or hypo-reactivity or arousal, which in turn place adolescents at increased risk for behavioral problems (Gunnar & Quevedo, Reference Gunnar and Quevedo2007; Steeger et al., Reference Steeger, Cook and Connell2017). Indeed, a study by Steeger et al. (Reference Steeger, Cook and Connell2017), which explored adolescent physiological stress responses, illustrated that hardships are associated with increased adolescent problems behaviors via physiological responses to stress.
The empirical literature has consistently linked family economic hardship to child and adolescent adjustment (Delgado et al., Reference Delgado, Killoren and Updegraff2013; Gershoff et al., Reference Gershoff, Aber, Raver and Lennon2007; Schenck-Fontaine & Panico, Reference Schenck-Fontaine and Panico2019; Simons & Steele, Reference Simons and Steele2020). In particular, youth experiencing higher than average levels of family economic hardship tend to display elevations in internalizing and externalizing problems (Delgado et al., Reference Delgado, Killoren and Updegraff2013; Schenck-Fontaine & Panico, Reference Schenck-Fontaine and Panico2019; Zilanawala & Pilkauskas, Reference Zilanawala and Pilkauskas2012). These links have been identified using measures of hardship that assess material deprivation (Schenck-Fontaine & Ryan, Reference Schenck-Fontaine and Ryan2022) and measures that tap financial stress (Delgado et al., Reference Delgado, Killoren and Updegraff2013), as well as measures of hardship that incorporate elements of both material deprivation and financial stress into a single predictor (Ponnet et al., Reference Ponnet, Wouters, Goedemé and Mortelmans2016). In their seminal work illustrating the importance of hardship when looking at links between economic circumstances and child development, Gershoff et al. (Reference Gershoff, Aber, Raver and Lennon2007) found that hardship, conceptualized as material deprivation, contemporaneously predicted worse socioemotional competence even when controlling for family income. In a more recent study focused specifically on adolescents, Delgado et al. (Reference Delgado, Killoren and Updegraff2013) found that parental reports of material deprivation predicted adolescent reports of financial stress and their behavioral functioning measured two years later. Given this growing literature, it is critical to consider family economic hardship as a key predictor of adolescent behavioral development.
The importance of youth’s own perceptions of hardship in predicting behavior
During adolescence, a multitude of complex social cognitive and affective processes mature and become refined. These processes include: (a) understanding the mental states of other individuals (theory of mind); (b) thinking about thinking (metacognition); and (c) emotion reactivity. Growth in these capacities enables adolescents to understand, explain, and predict the actions and cognitions of others (Crone & Dahl, Reference Crone and Dahl2012; Keating, Reference Keating2004; Weimer et al., Reference Weimer, Warnell, Ettekal, Cartwright, Guajardo and Liew2021). Several studies have documented adolescents’ rapid development in perspective-taking processes, resulting in greater accuracy compared to children in their ability to assess the mental states of others (Bosco et al., Reference Bosco, Colle and Tirassa2009; Choudhury et al., Reference Choudhury, Blakemore and Charman2006; Dumontheil et al., Reference Dumontheil, Apperly and Blakemore2010; Lonigro et al., Reference Lonigro, Baiocco, Baumgartner and Laghi2017). From a metacognitive standpoint, adolescents gain greater abilities to recognize their own thoughts and emotions, to relate them to relevant interpersonal events, and to understand the mental states of other people and keep them distinct from their mental states (Demetriou & Bakracevic, Reference Demetriou and Bakracevic2009; Weil et al., Reference Weil, Fleming, Dumontheil, Kilford, Weil, Rees, Dolan and Blakemore2013). Not only do these competencies suggest youth may have unique perspectives of economic hardship, these processes are also linked to adolescents’ self-regulation and executive control, which is particularly important in shaping behavioral development, including internalizing and externalizing behaviors (Fernandez-Duque et al., Reference Fernandez-Duque, Baird and Posner2000; Fleming et al., Reference Fleming, Dolan and Frith2012; Shimamura, Reference Shimamura, Stuss and Knight2002).
Additionally, studies have documented that family members can have unique perceptions of family economic circumstances despite living in the same household (Delgado et al., Reference Delgado, Killoren and Updegraff2013; McLoyd et al., Reference McLoyd, Jayaratne, Ceballo and Borquez1994; Mistry & Elenbaas, Reference Mistry and Elenbaas2021; Rivenbark et al., Reference Rivenbark, Arseneault, Caspi, Danese, Fisher, Moffitt, Rasmussen, Russell and Odgers2020). In fact, caregiver and youth reports of hardship only correlate modestly – around r = .25 (Delgado et al., Reference Delgado, Killoren and Updegraff2013; McLoyd et al., Reference McLoyd, Jayaratne, Ceballo and Borquez1994; Mistry & Elenbaas, Reference Mistry and Elenbaas2021). This could be due to caregivers’ attempts to shield youth from economic hardship, which could result in differing perceptions of hardship between caregivers and their children (Mistry & Lowe, Reference Mistry, Lowe and owe2006). Alternatively, there may be individual differences between caregivers and youth when it comes to how distressing or impactful shared hardships are based on personality or utilization of coping mechanisms (Wadsworth & Berger, Reference Wadsworth and Berger2006).
Emotionally, data suggest that adolescents may be uniquely vigilant toward and more likely than adults to react to certain cues of threat (Dreyfuss et al., Reference Dreyfuss, Caudle, Drysdale, Johnston, Cohen, Somerville, Galván, Tottenham, Hare and Casey2014). Compared to adults, adolescents are less able to accurately discern different expressions of negative emotions (Thomas et al., Reference Thomas, De Bellis, Graham and LaBar2007), have greater difficulty suppressing attention toward cues of threat (Cohen-Gilbert & Thomas, Reference Cohen-Gilbert and Thomas2013), and are worse at discriminating threat cues from safety cues (Grasser & Jovanovic, Reference Grasser and Jovanovic2021; Lau & Waters, Reference Lau and Waters2017). Put differently, adolescents may sometimes be perceiving more negative affect in their environment, believing environments to be potentially threatening or hostile. In their homes, youth could potentially be seeing certain aspects of hardship as more negative or adverse, giving youth a perspective of hardship that is unique from that of their adult caregivers.
Externally, adolescents are reorienting from parents to peers, beginning to face the external world, and contemplating facets of their future personal and professional identities (Brown & Larson, Reference Brown and Larson2009; Flanagan & Gallay, Reference Flanagan, Gallay and Benson2014; Hagquist, Reference Hagquist2007; Schoon & Heckhausen, Reference Schoon and Heckhausen2019). During this time period, social networks significantly expand, and peers may have a greater influence on youth behavior and self-evaluations, as adolescents more frequently engage in social comparisons and deeply internalize the views of others (Albert et al., Reference Albert, Chein and Steinberg2013; Jacobs, Reference Jacobs2003; O’Brien & Bierman, Reference O’Brien and Bierman1988; Rivenbark et al., Reference Rivenbark, Arseneault, Caspi, Danese, Fisher, Moffitt, Rasmussen, Russell and Odgers2020; Wrzus et al., Reference Wrzus, Hänel, Wagner and Neyer2013). Moreover, adolescents begin to interface with different social structures and institutions. Social roles and norms become particularly salient for youth, and at the same time adolescents are gaining more knowledge of their own economic circumstances (Blakemore & Mills, Reference Blakemore and Mills2014; McLoyd et al., Reference McLoyd, Kaplan, Purtell, Bagley, Hardaway and Smalls2009; McLoyd, Reference McLoyd1998, Reference McLoyd, Carlo, Crockett and Carranza2011). As youth gain this knowledge and become more attune to evaluations by peers and others, this could increase stigma and heighten concerns about unfavorable appearances of their family’s economic position and lack/quality of material goods (i.e., clothing, housing conditions), which can cascade into increased problem behavior (Silbereisen et al., Reference Silbereisen, Walper and Albrecht1990). Indeed, studies have documented that adolescents’ perception of their social status was a unique predictor of their internalizing and externalizing problems at age 14 (Rivenbark et al., Reference Rivenbark, Arseneault, Caspi, Danese, Fisher, Moffitt, Rasmussen, Russell and Odgers2020, Reference Rivenbark, Arseneault, Caspi, Danese, Fisher, Moffitt, Rasmussen, Russell and Odgers2020; Russell & Odgers, Reference Russell and Odgers2020). Examining these elements collectively, perceptions of economic hardship could be a major driver of psychological distress and behavioral adjustment during adolescence. Given the host of changes occurring during adolescence, youth may be uniquely sensitive to information or cues that convey their socioeconomic standing, like economic hardship (Somerville, Reference Somerville2013). Moreover, as adolescents take more active roles in their development and make gains in cognitive abilities–for example, more advanced planning, problem solving, and perspective taking–their own views of life experiences play an increasingly important role in predicting their behavior (Crone & Dahl, Reference Crone and Dahl2012; Keating, Reference Keating2004; Weimer et al., Reference Weimer, Warnell, Ettekal, Cartwright, Guajardo and Liew2021). Importantly, adolescence is also the time when youth begin to assess and plan for their own economic futures (e.g., making educational and career choices) and, thus, may pay closer attention to their families’ economic circumstances (Brown & Larson, Reference Brown and Larson2009; Flanagan & Gallay, Reference Flanagan, Gallay and Benson2014; Hagquist, Reference Hagquist2007; Schoon & Heckhausen, Reference Schoon and Heckhausen2019). Centering adolescents’ awareness or perceived understanding of life events may be necessary to fully understand the effects of stressful events on them (Compas, Reference Compas1987; Keating, Reference Keating2004; Rutter, Reference Rutter1983).
Research has generally focused on caregivers’ perceptions of hardship when looking at how economic hardship relates to youth adjustment, but there are a few notable studies that have looked at how youth reports of hardship relate to their own behavior, including the work by Delgado et al. (Reference Delgado, Killoren and Updegraff2013) described above. Prior to the Delgado et al. (Reference Delgado, Killoren and Updegraff2013) study, Lempers et al. (Reference Lempers, Clark-Lempers and Simons1989) found that youth reports of economic hardship predicted their levels of distress. Seminal work by McLoyd et al. (Reference McLoyd, Jayaratne, Ceballo and Borquez1994) found that African American adolescents’ perceptions of economic hardship predicted their concurrent depressive symptomology, anxiety levels, and self-esteem. While these studies included reports of youth perceptions of hardship, they relied on cross sectional data to test associations between adolescent perceptions of hardship and behavior. More recently, longitudinal work has found that adolescent-reported hardship predicted increases in anxiety and depression at later time points, even when controlling for prior hardship and behavior problems (Mistry et al., Reference Mistry, Benner, Tan and Kim2009; Wadsworth & Berger, Reference Wadsworth and Berger2006; Wadsworth & Compas, Reference Wadsworth and Compas2002). For example, Wadsworth and Compas (Reference Wadsworth and Compas2002) found that youth reports of hardship were positively related to anxiety/depression and aggressive behavior in a sample of rural adolescents. These studies did not focus on the variability and changes in these constructs, instead using measures from a single timepoint. This overlooks potential intra-year variability in both hardship and behavior problems. Moreover, it is unclear whether youth perceptions relate to behavior over and above caregivers’ experiences of hardship since these studies did not consider youth and caregiver reports simultaneously.
Limitations in the current literature
While the literature demonstrates robust links between hardship and youth behavior problems, past studies have three notable limitations. First, much of the research on economic hardship has conceptualized its effects on youth as operating through parents. Specifically, studies argue that economic hardship is distressing to parents, and this distress compromises their relationships and parenting practices, leading to conflictual interactions with their children and harsh and/or detached parenting behaviors and, ultimately, increased child behavior problems (Gard et al., Reference Gard, McLoyd, Mitchell and Hyde2020; Landers-Potts et al., Reference Landers-Potts, Wickrama, Simons, Cutrona, Gibbons, Simons and Conger2015; Neppl et al., Reference Neppl, Senia and Donnellan2016; Raver et al., Reference Raver, Gershoff and Aber2007; Simons et al., Reference Simons, Lei, Beach, Philibert, Cutrona, Gibbons and Barr2016; Zilanawala & Pilkauskas, Reference Zilanawala and Pilkauskas2012). Yet, material deprivation and financial stress may impact adolescents directly. For example, youth may get stressed or upset when they are unable to go out with friends or participate in school extracurricular activities because of limited economic resources. Importantly, most studies exploring links between hardship and development (Gard et al., Reference Gard, McLoyd, Mitchell and Hyde2020; Landers-Potts et al., Reference Landers-Potts, Wickrama, Simons, Cutrona, Gibbons, Simons and Conger2015; Neppl et al., Reference Neppl, Senia and Donnellan2016; Raver et al., Reference Raver, Gershoff and Aber2007; Simons et al., Reference Simons, Lei, Beach, Philibert, Cutrona, Gibbons and Barr2016) do not include youth reports of economic hardship, instead relying solely on caregiver reports. Even studies that do ask adolescents about hardship assume that economic hardship primarily affects youth through caregivers and caregiving practices, thereby ignoring developmental changes occurring in social and cognitive processes that make adolescents’ own understanding and perceptions of the world increasingly important in shaping their behavior. This study investigates youths’ own experiences of economic hardship, including asking them questions about material deprivation and financial stress that are directly relevant to adolescents (e.g. skipping going out with friends because of a lack of money, worrying because caregivers do not have enough money to pay for things), as critical correlates of their behavior. Our youth-report measures of material deprivation and financial stress aim to capture experiences and feelings that are proximal to youth, on which they can report accurately, and, in turn, that may have strong ties to their behavioral functioning.
Next, insufficient attention has been given to the time scale at which hardship and youth behavioral functioning are measured. Studies often use measures of hardship drawn from a single point in time, capturing a snapshot rather than a more comprehensive representation of families’ financial experiences (Devenish et al., Reference Devenish, Hooley and Mellor2017; Hostinar & Miller, Reference Hostinar and Miller2019). This is a serious issue given results from the U.S. Financial Diaries project, which tracked the finances of low- and moderate-income households continuously over a year (Hannagan & Morduch, Reference Hannagan and Morduch2015). These households faced substantial swings in month-to-month income. Further, data from the U.S. Financial Diaries also showed that expenditures varied from month-to-month, and expense spikes did not align with income spikes (Morduch & Schneider, Reference Morduch and Schneider2017). For these reasons, economic hardship is likely to exhibit major intra-year variability. Indeed, a recent post-pandemic study testing this presupposition showed that caregivers’ reports of hardship varied considerably month-to-month (Liu et al., Reference Liu, Zalewski, Lengua, Gunnar, Giuliani and Fisher2022). Yet, prior research most often relies on a point-in-time snapshot of hardship, which may underestimate links between economic hardship and youth mental health if the snapshot is not an accurate representation of a family’s true experiences of hardship over time.
Lastly, the temporal spacing of measures of hardship and problem behavior varies widely across studies, which raises concerns about whether they capture the experiences of youth and families at the time around which hardship is felt (Granger, Reference Granger1969). Measures of economic hardship often predate reports of behavior by as many as eight years (Gard et al., Reference Gard, McLoyd, Mitchell and Hyde2020; Simons et al., Reference Simons, Lei, Beach, Philibert, Cutrona, Gibbons and Barr2016). As an example, Gard et al. (Reference Gard, McLoyd, Mitchell and Hyde2020) used parent-reports of economic hardship when children were 1 year old to predict externalizing problems at age 9. These large time lags, while helpful in establishing temporal precedence, may downwardly bias the relations among these variables because they are so distally measured. Nor do they account for the fact that more than one third of U.S. households experience substantial intra-year income change (Dynan et al., Reference Dynan, Elmendorf and Sichel2012; Gottschalk & Moffitt, Reference Gottschalk and Moffitt2009), and economic hardship is volatile (Liu et al., Reference Liu, Zalewski, Lengua, Gunnar, Giuliani and Fisher2022). Thus, the large temporal windows in prior studies may not capture the acute psychological distress experienced in months where economic hardship is high. Financial resources are often received monthly, and bills and other financial obligations come due weekly or monthly; therefore, experiences of hardship are likely to be felt at or around that time. Thus, it may be important to characterize relations between economic hardship and behavior with month-to-month assessments, since there are intra-year fluctuations in income, earnings, bills, and hardship resulting from the inability to pay bills or purchase necessities. Thus, we hypothesize that changes in material deprivation and financial stress will give rise to rapid changes in adolescent behavior, and our intensive repeated measures design allows us to test contemporaneous associations (see Granger, Reference Granger1969). We believe that this temporal design will provide better metrics for examining the relations among hardship and youth behavior problems (Collins & Steinberg, Reference Collins and Steinberg2006).
The current study
Motivated by limitations in the current literature, our study investigated associations between youths’ and caregivers’ feelings of hardship as they relate to youth reports of externalizing and internalizing behaviors. Importantly, given rising volatility in household economic circumstances, it considers these associations on a shorter time scale, leveraging monthly intensive repeated measures of caregiver and youth reported hardship and behavioral outcomes to explore within- and between-youth–caregiver dyad associations among variables. With this powerful design, we addressed two aims. First, given the lack of research exploring fluctuations in hardship and behavior problems on a sub-annual time scale, we calculated month-to-month variability of caregiver-reported material deprivation, youth-reported material deprivation and financial distress, and youth externalizing and internalizing outcomes over a nine-month period. Second, we tested whether variations in monthly caregiver perceptions of material deprivation and youth perceptions of material deprivation and financial stress were associated with youths’ reports of their externalizing and internalizing symptoms. As a notable contribution to the literature, we tested this aim using a multilevel modeling framework that allowed us to examine associations within- and between-dyads. In doing so, within-dyad associations reduce omitted variable bias by isolating the variance in behavior problems explained by change in hardship over time within-dyads, with each dyad serving as its own counterfactual. With respect to the first aim, we expected significant monthly variability in these variables of interest, though we predicted that caregivers’ reports of material deprivation would be more stable. For the second aim, we hypothesized that youth perceptions of hardship, both material deprivation and feelings of financial stress, would predict internalizing and externalizing behaviors, with more perceived hardship predicting increased problem behaviors. Many developmental psychopathology scholars argue that specific aspects of economic disadvantage are not likely to be uniquely associated with psychopathology; rather, the mechanistic pathways through which a unique dimension of disadvantage impact psychopathology may be different (e.g., Miller et al., Reference Miller, Sheridan, Hanson, McLaughlin, Bates and Lansford2018; Smith & Pollak, Reference Smith and Pollak2021). Accordingly, we had no specific hypotheses regarding unique associations between material deprivation or financial stress and behavior problems. Lastly, we predicted that youth reports of hardship would have stronger links to their behavior than caregiver reports.
Method
Participants
The current study draws data from the Family Income Dynamics Study (FInD). FInD is a nine-month longitudinal study designed to densely sample economic circumstances and family processes in a racially and economically diverse sample of youth and caregivers in the greater metropolitan area of Pittsburgh, PA. The sample included 104 youth (55% female) and caregiver (85% biological mother) dyads. Youth were 14–16 years old (M = 14.8, SD = 0.83) at the time of the initial survey, and 4% reported their race as Asian, 37% Black, 6% Latinx, 10% multiracial, and 43% White. Based on caregivers’ reports of prior year’s income and household size during the screening process, 51% of the dyads were considered low-income (<2xs the federal poverty threshold) and 49% were middle-income (generally between 2xs and 5xs the federal poverty threshold). Among the low-income sample, approximately 27% qualified as poor, that is, had annual incomes below the federal poverty threshold. Highest level of caregiver education varied as well: 10% of caregivers had a high school degree, 36% had taken some college courses or possessed an associate’s degree or professional/trade certification, and 54% had a bachelor’s or graduate degree. Almost half of the caregivers were married and living with their spouse (46%). Of the caregivers who were not married and cohabiting, approximately 10% were cohabiting with a partner, while 20% reported being single. Full descriptive statistics are reported in Table 1 (these reflect the full stacked data set over the nine-month period).
Descriptive statistics for sample
Procedures
Families were recruited into the study from participant registries, community organizations, schools, advertisements, and referrals. Recruitment began in November of 2019 and the final cohort was recruited in December of 2020; data collection spanned November, 2019 through August, 2021. Initial interviews were conducted in person, but in March 2020, we began conducting them over the phone after widespread shutdowns due to COVID-19. After the initial interview, caregivers and youth completed monthly surveys electronically via computer or smart phone for the next eight consecutive months. The surveys included questions tapping a range of sociodemographic, psychological, financial, caregiving, and family functioning measures. Participants were asked to complete the surveys between the first and fifth of the month. A few days prior to the follow up surveys and during the five-day window if necessary, caregivers and youth were reminded to complete their surveys via email and text. This led to a very high retention rate; no participant missed more than three waves of data, and the majority missed no waves. The total percentage of missing responses across waves for all variables of interest totaled six percent. To adjust for missing data in our analyses, we relied on Mplus’ default approach of maximum likelihood using missing-at-random assumptions (Muthén et al., Reference Muthén, Muthén and Asparouhov2017). Analyses of missing data patterns supported this assumption, that is, missingness was not predicted by our variables of interest.
Measures
Youth Externalizing and Internalizing Behavior Outcomes
Every month, youth reported on their externalizing and internalizing problem behaviors by completing the Strengths and Difficulties Questionnaire (SDQ; Goodman, Reference Goodman1997). The SDQ is comprised of 25 items assessing emotional problems, conduct problems, hyperactivity, peer problems, and prosocial behaviors. The youth were presented with questions such as “I get very angry and often lose my temper,” “I am often accused of lying or cheating,” “I am often unhappy, depressed or tearful” and responded on a 3-point scale (“not true” to “certainly true”). Consistent with prior approaches, we created internalizing and externalizing composite variables based on validated subscales from the SDQ (Goodman, Reference Goodman1997). The externalizing composite reflects the sum of conduct problems and hyperactivity subscales (10 items,
$\overline{\alpha }$
= .72−.81), and the internalizing composite reflects the sum of emotional and peer problems subscales (10 items,
$\overline{\alpha }$
= .66−.74). Higher scores indicate more externalizing and internalizing behaviors.
Economic hardship
Both caregivers and youth reported on economic hardship monthly. While caregivers’ questionnaire tapped primarily material deprivation, the youth questionnaire included questions tapping both material deprivation and feelings of financial stress. The scoring and distribution of the economic hardship measures used in the present study are consistent with other studies including families from economically diverse backgrounds (Barrera et al., Reference Barrera, Caples and Tein2001; Zilanawala & Pilkauskas, Reference Zilanawala and Pilkauskas2012). The full list of items included in all hardship measures is listed in Table 2.
Hardship measures
Caregiver Material Deprivation
Each month, caregivers completed a 12-item questionnaire assessing material deprivation adapted from Gershoff et al. (Reference Gershoff, Aber, Raver and Lennon2007) and Sullivan et al. (Reference Sullivan, Turner and Danziger2008). Eight dichotomous items (1 = Yes, 0 = No) asked caregivers if hardships occurred in the past month. Items included questions like “Did you have utilities cut off for non-payment?” “Did someone need to go to the doctor/hospital, but did not go?” and “Did you have a toilet, bath/shower, hot water heater, or other plumbing that did not work?” Three items were answered on Likert scales (0 = not at all difficult to 3 = very difficult or 0 = never to 5 = all the time). These items asked, “How difficult was it to pay bills (e.g., utilities, rent/mortgage)?,” “Did you put off buying something you need -such as food, clothing, medical care, or housing- because you don't have money?,” “How often can your household afford to do things for fun like going to the movies or eating out?” A final item asked about participants’ food situation (“1 = We can always afford to eat good nutritious meals; 2 = We can always afford to eat but not always the kinds of food we should be eating; 3 = Sometimes we cannot afford to eat enough; 4 = Often we cannot afford to enough to eat”). The latter four items were dichotomized for consistency by coding responses of “somewhat difficult” or “occasionally” or “not always” and higher as “1” (Gershoff et al., Reference Gershoff, Aber, Raver and Lennon2007; Sullivan et al., Reference Sullivan, Turner and Danziger2008). We created a caregiver material deprivation measure ranging from 0-1 by averaging across all 12 dichotomous items (
$\overline{\alpha }$
= .75−.78), where higher values indicate more hardship.
Youth Material Deprivation
The youth material deprivation measure is a composite of 12 items adapted from Sullivan et al. (Reference Sullivan, Turner and Danziger2008). These items asked how often youth experienced hardships that may be especially relevant to adolescents like skipping going out with friends because of a lack of money, not asking for something they wanted/needed because caregiver(s) couldn't afford it, missing out on something because of a lack of transportation money, and inability to participate in a school-sponsored activity because of lack of money. All items were answered based on the past month using a 5-point scale (1 = No problem at all; 2 = 2; 3 = 3; 4 = 4; 5 = A very serious problem or 1 = Never; 2 = 2; 3 = 3; 4 = 4; 5 = Always). Items were reverse coded and averaged so that a higher value on the composite reflects more material deprivation (
$\overline{\alpha }$
= .88−.91).
Youth Financial Stress
This was measured using five items drawn from Conger et al. (Reference Conger, Conger, Matthews and Elder1999). The items included questions such as “How often do(es) your caregiver(s) argue about not having enough money?” and “How upset or worried are your caregiver(s) because they do not have enough money to pay for things?” These were answered on a 1–5 scale, with higher scores signaling more distress. Scores on the five items were averaged to create the financial stress composite measure (
$\overline{\alpha }$
= .80−.91).
Sociodemographic covariates
We controlled for income based on caregivers’ monthly reports of total income received in the prior month. Specifically, they were given a 17-item list of sources of income and asked to report the after-tax amount received from each source. Sources included income from formal employment, income from informal work arrangements (e.g., “gig” work like ride share, food delivery), alimony and child support, income from benefits/assistance programs like unemployment compensation, social security benefits, or government assistance programs (e.g., Temporary Assistance for Needy Families), and income from other sources like lottery winnings, pawning goods, and gifts and loans (Fox & Burns, Reference Fox and Burns2021). The sum of all sources of income was tabulated for caregivers and they were asked if the total looked correct (they could make edits if they indicated total was incorrect). We controlled for the natural log of this sum to capture nonlinearities in income’s associaitons with development (Dearing et al., Reference Dearing, McCartney and Taylor2006; Duncan et al., Reference Duncan, Duncan and Strycker2006; Mayer, Reference Mayer2002; Votruba-Drzal, Reference Votruba-Drzal2006). Specifically, the natural log transformed income measure adjusts for the well-documented trend that the wellbeing of children whose families are at the lower end of the income distribution tends to be more responsive to income changes than that of children in higher income families (Duncan et al., Reference Duncan, Kalil and Ziol-Guest2017; Votruba-Drzal, Reference Votruba-Drzal2006).
We also controlled for three important youth and family characteristics that are known to covary with youth behavioral outcomes (Merikangas et al., Reference Merikangas, He, Burstein, Swanson, Avenevoli, Cui, Benjet, Georgiades and Swendsen2010). These included youth mean age and household size over the course of their nine-month participation. We also controlled for caregivers’ marital status, adolescent’s sex assigned at birth (0 = male, 1 = female), and whether the child identified their race as non-White (youth were able to select from multiple racial/ethnic categories or self-identify, but given limited sample size, we used a dichotomous indicator where 1 = non-White race/ethnicity), all of which were assessed at baseline. Lastly, an indicator for the wave of data collection (ranging from 0 through 8) was included in models to control for any unobservable variables that affect behavioral functioning and are highly correlated with time in study.
Analytic approach
All analyses were conducted in Mplus version 8.6. For the first aim, we examined the monthly variability in caregiver reports of material deprivation, youth reports of material deprivation and financial stress, and youth behavior outcomes. To do this, we examined intraclass correlation coefficients (ICCs) to assess the level of variation that caregiver and youth dyads experienced from month to month over the nine months of data collection. ICCs assess the strength of the correlation between a variable reported by a rater over time. As the size of ICCs increases, this indicates decreasing variability, with an ICC of one indicating no change over time.
We examined aim two in a multilevel structural equation modeling (MSEM) framework using a maximum likelihood estimator with robust standard errors (Muthén et al., Reference Muthén, Muthén and Asparouhov2017; Preacher et al., Reference Preacher, Zyphur and Zhang2010). This specification allowed us to simultaneously examine associations within (level one) and between (level two) caregiver and youth dyads across the nine months. One strength of this approach is that it addresses omitted variable bias by isolating the variance explained by change within dyads over time, since each dyad serves as its own counterfactual at level one. Specifically, contemporaneous associations in the within portion of the model reflect fluctuations in variables from month to month as they rise and fall from the participant’s average responses. These estimates are stronger than traditional cross-sectional models when it comes to drawing causal inferences because threats to internal validity posed by time-invariant omitted variables is greatly reduced. Further, by estimating associations at level two, we were also able to examine differences between dyads’ means over time, which provided valuable information on the unique contributions of caregiver perceptions of material deprivation versus youth perceptions of material deprivation and financial distress in predicting youth behavior outcomes.
One MSEM modeled direct associations of caregiver and youth hardship on youth externalizing, and the other modeled these associations on youth internalizing. Both models control for income and wave at level one, and control for income, youth mean age, youth sex, and caregiver marital status, race/ethnicity, and household size at level two. Both models also control for the correlations between income and youth perceptions of material deprivation and financial stress, income and caregiver perceptions of material deprivation, and caregiver material deprivation and youth material deprivation and financial stress at levels one and two. These models provided insight on the unique relations between the different measures of hardship and behavior outcomes as well as how much variance the caregiver and youth hardship measures shared.
We ran three alternate models as sensitivity checks to our main models. First, we two ran models to address concerns about the potentially confounding effect of the COVID-19 pandemic on hardship and youth behavior. The first one controlled for a categorical indicator for whether the data collection occurred after March 1, 2020 (which corresponds to the date when things started shutting down locally). The second included a continuous control variable reflecting the number of months since March 2020 that had passed at the time of the data collection. Second, in an effort to test the best temporal timescale at which to model associations between hardship and behavior problems, we ran iterations of models with a one-month time lag where youth and caregiver hardship predicted behavior problems in the following month. Finally, to help rule out simultaneity bias, we ran a similar lagged model, but with youth behavior problems predicting their reports of hardship one month later to address concerns that adolescents’ internalizing and externalizing behaviors may be causing teens to report higher levels of hardship, and not vice versa as our theoretical model hypothesizes.
Results
Month-to-month variability of constructs
The first aim examined the monthly variability of youth and caregiver hardship and youth behavior outcomes by looking at ICCs (Table 2). The ICCs are as follows: caregiver-reported material deprivation (0.69), youth-reported material deprivation (0.72), youth-reported financial stress (0.73), externalizing (0.78), and internalizing (0.73). These values demonstrate that the constructs in our study are relatively stable, but do have some month-to-month variability.
Associations between hardship and behavior outcomes
The second aim used MSEM to examine associations between variation in caregiver and youth hardship and youth reported behavior outcomes. The externalizing results are presented in Table 3 and the internalizing results in Table 4. Since youth and caregiver hardship are measured on different scales, we present fully standardized results for ease of interpretation and comparing effect sizes between variables.
Two-level model of associations between hardship and youth externalizing behaviors
*p < .05, **p < .01, ***p < .001
Two-level structural equation model of associations between caregiver and youth hardship and youth internalizing behaviors
*p < .05, ** p < .01, ***p < .001
Externalizing
Our results show a positive association between youth reports of financial stress and externalizing within individuals over time (Table 3). A one standard deviation (SD) increase in youth financial stress is associated with a 0.18 SD increase in youth externalizing within individuals (p < .001). In other words, when youth reports are being compared to their own averages over time, more financial stress predicted higher externalizing problems. When comparing across dyads, youth reports of material deprivation positively related to between-individual variation in externalizing problems. Specifically, a one SD increase in material hardship was linked to a 0.41 SD increase in youth externalizing when looking between individuals (p = .015). Notably, neither income nor caregiver reports of material deprivation predicted externalizing within or between dyads. Moreover, within-youth variation in material deprivation and between-youth variation in financial distress was unrelated to externalizing. Similarly, none of the covariates were related to youth externalizing with the exception of a marginally significant relation between sex and externalizing, with female youth reporting 0.17 SD lower externalizing problems than males (p = .072).
In terms of the estimated correlations between income and hardship, income and youth reports of financial stress were negatively correlated within dyads (r = −.15, p = .028). Between dyads, income was negatively correlated with all hardship measures. The correlation was strongest between income and caregiver reports of material deprivation (r = −.58, p < .000), with between-individual negative correlations between income and youth reports of financial stress and material deprivation being more than half that size (r = −.20, p = .044 and r = −.24, p < .008, respectively). The two youth reports of hardship were significantly and positively correlated within-dyads (r = .47, p = .000), while interestingly only the youth reports of financial stress (and not material deprivation) correlated with caregiver reported material deprivation at the within-level (r = .11, p = .004). All hardship measures were significantly and positively correlated between dyads (r = .45-.78, p < .000).
Internalizing
The results for internalizing (Table 4) differed from externalizing in that it was youth reported material deprivation that predicted within-individual variability in behavior problems. A one SD increase in youth reported material deprivation was associated with a 0.19 SD increase in youth internalizing within individuals (p = .002). Youth reported material deprivation was also positively related to internalizing when looking at between-individual variability (β = .40, p = .018). Thus, when youth reports of material deprivation are compared to their own average over time as well as when comparing between adolescents, more material deprivation predicted higher reports of internalizing. Youth reported financial stress did not predict internalizing at either the within- or between-levels. Similar to externalizing, neither income nor caregiver reported material deprivation were related to internalizing within or between dyads. Additionally, none of the covariates significantly predicted internalizing. The correlations between hardship measures are the same as the correlations in the externalizing model reported above.
Sensitivity analyses
The results of the sensitivity analyses are available in the online appendix. To summarize, first, all observed associations were robust in the two model specifications that controlled for the timing of data collection during the COVID-19 pandemic (Table A of appendix). Second, the within-individual associations between youth-reports of hardship and behavior problems were no longer significant when we lagged the measures of hardship by one month (Table B of appendix), which we believe is strong evidence that behavior problems occur contemporaneously to when youth hardship is experienced. Indeed, methodological and theoretical work (Granger, Reference Granger1969) suggests that when variables in an intensive longitudinal design fluctuate substantially over the time scale they are assessed, it is crucial for the time scale of associations in the models to map on to the pace of change. Between-individual associations remained significant when lagging reports of hardship by a month, and parent-reported material deprivation was positively related to youth externalizing when looking within-dyads, but this association was not replicated when looking between-dyads or when predicting internalizing. Finally, adolescents’ behavior problems did not predict their reports of hardship reported in the following month (Table C of appendix). This helps bolster our argument that variability in adolescents’ experiences of hardship predict their behavioral functioning, and not vice versa.
Discussion
By examining longitudinal data on perceptions of household economic hardship and youth behavioral outcomes collected monthly over a period of nine months, this study advances the current literature in two major ways. First, in a diverse sample of caregivers and youth, it documented the month-to-month variability of caregiver- and youth-reported hardship and behavior. Second, this study provided evidence that youth-reported perceptions of material deprivation were related to both externalizing and internalizing behaviors and youth reports of financial stress were linked to externalizing, over and above their caregivers’ reports of material deprivation. Indeed, accounting for youths’ own perceptions of material deprivation and financial stress (which could function as a pathway by which economic hardship might impact youth), caregiver-reported hardship did not predict youth behavior. These findings provide valuable information to researchers interested in the role of economic hardship in adolescent development, and in particular how youth perceptions of hardship influence their behavioral development.
Hardship and behavior vary month-to-month
The present study is the first to test whether caregiver and youth feelings of hardship, as well as behavior problems, vary significantly on a monthly basis. Our study confirmed that there was month-to-month variability observed in our measures of hardship, externalizing, and internalizing (as seen with the ICCs in the analyses for Aim 1). While the ICCs for the three hardship measures would certainly be considered large, indicating stability within dyads over the nine-month period, none of the values exceed 0.75 and thus would only be considered moderately stable (Koo & Li, Reference Koo and Li2016). This is further evidence that experiences of hardship are somewhat volatile, just as income has been shown to be volatile (Hannagan & Morduch, Reference Hannagan and Morduch2015; Liu et al., Reference Liu, Zalewski, Lengua, Gunnar, Giuliani and Fisher2022). Indeed, we ran post hoc analyses of the income reports to compare within-year variability of our hardship measures with income; monthly income showed similar stability in this sample (ICCs = 0.75). Variability in youth and caregivers’ perceptions of economic hardship observed in this study is consistent with broader trends in increasing economic volatility for U.S. families (i.e., month-to-month, or year-to-year fluctuations in family income), particularly low-income families (Dynan et al., Reference Dynan, Elmendorf and Sichel2012; Gottschalk & Moffitt, Reference Gottschalk and Moffitt2009). Our study illustrates that hardship and behavior exhibit enough variation to estimate intra-individual changes from month-to-month, and albeit small, within-individual deviations in hardship did predict variation in behavioral functioning, which makes them meaningful and weakens presumptions of stability for these constructs that are a hallmark of many prior studies exploring links between hardship and child development (Gard et al., Reference Gard, McLoyd, Mitchell and Hyde2020; Simons et al., Reference Simons, Lei, Beach, Philibert, Cutrona, Gibbons and Barr2016).
Youth perceptions of financial stress and material deprivation are unique from caregiver perceptions of material deprivation
Additionally, this study highlights the importance of measuring both caregiver and youth perceptions of economic hardship, including financial stress and material deprivation. Indeed, within dyads, variation in caregiver reported material deprivation and youth reports of financial stress and material deprivation were barely correlated (r = .04−.11), and across families, there were modest correlations among the hardship measures (r = .45−.47). This is relatively consistent with prior literature showing average correlations around r = .25 (Delgado et al., Reference Delgado, Killoren and Updegraff2013; McLoyd et al., Reference McLoyd, Jayaratne, Ceballo and Borquez1994; Mistry & Elenbaas, Reference Mistry and Elenbaas2021); notably, these prior studies did not disaggregate within-dyad and between-dyad correlations. Of course, our measures of hardship differed across caregivers and youth. High scores on the caregiver-reported material deprivation measure reflect arguably more grave living conditions, like food insecurity and going without utilities and/or adequate medical care. On the other hand, our youth measures of hardship were intended to capture hardships particularly relevant in adolescence, like impediments to their social lives or status. In this way, it is not surprising that the reports are not highly correlated. Nonetheless, the significant variance in youth reports of financial stress and material deprivation that is unique from caregiver reports of material deprivation (and the fact that youth reports predict youth behavior) indicates that these are important constructs to take into account when studying economic contexts and adolescent development.
Besides asking about different hardships, another reason youth and caregiver reports were disparate may be due to caregivers shielding their children from the economic hardships that come with financial struggles (Mistry & Lowe, Reference Mistry, Lowe and owe2006; Quint et al., Reference Quint, Griffin, Kaufman, Landers and Utterback2018). Past research suggests that caregivers’ decisions to disclose financial information to their children involve a careful balancing between the rewards and risks of such disclosure (e.g., being honest versus causing burden or worry for their adolescents), as well as cultural norms of sharing financial information (Romo, Reference Romo2011, Reference Romo2014). This is consistent with general findings of variability in the depth of financial information caregivers share with their children (LeBaron et al., Reference LeBaron, Marks, Rosa and Hill2020; Quint et al., Reference Quint, Griffin, Kaufman, Landers and Utterback2018). At the outset of this project, we held preliminary focus groups to hone our research questions and measures. During these conversations with families, caregivers often discussed thoughtfully choosing which household financial information to share with their adolescents. Specifically, caregivers disclosed information to help youth think about their futures or explain family financial situations, while sometimes withholding information to lessen worry and anxiety. This is also supported by our finding of a stronger correlation between caregiver reports of material deprivation and monthly income (r = −.58) than youth reported hardships and monthly income (r = −.20 − −.24) between dyads, which shows that caregivers’ perceptions of economic hardship more closely map on to the actual income coming into the household. Another reason for the small correlation between caregiver and youth hardship may relate to differences in personality or coping mechanisms (Wadsworth & Berger, Reference Wadsworth and Berger2006) or reflect adolescents’ developing ability to identify and distinguish their own perceptions from the mental states of others (Demetriou & Bakracevic, Reference Demetriou and Bakracevic2009; Weil et al., Reference Weil, Fleming, Dumontheil, Kilford, Weil, Rees, Dolan and Blakemore2013). Future research should aim to understand the mechanisms leading to disparate perceptions of hardship between caregivers and their adolescents (like, for instance, shielding), and the sequalae of these differential perceptions.
Pioneering work on hardship by Gershoff et al. (Reference Gershoff, Aber, Raver and Lennon2007) in a younger cohort (∼6 years of age) found that hardship worked through caregivers, with those experiencing material hardship reporting stress related to making ends meet. While seminal, the sample did not include adolescents, so investigators could not examine youth perceptions of material deprivation or financial stress. It may be that hardship felt by parents and hardship experienced by youth operate through different pathways in shaping development and/or act as different intermediate mechanisms bridging documented associations between economic disadvantage to heightened problem behavior (Costello et al., Reference Costello, Copeland and Angold2011; D’Onofrio et al., Reference D’Onofrio, Goodnight, Van Hulle, Rodgers, Rathouz, Waldman and Lahey2009; Larsson et al., Reference Larsson, Sariaslan, Långström, D’Onofrio and Lichtenstein2014; McLoyd et al., Reference McLoyd, Kaplan, Purtell, Bagley, Hardaway and Smalls2009). Longitudinal work connecting these two sets of findings would be critical, but it is clear from our study that valuable information is lost without including youth-reported measures.
Interestingly, while few studies have incorporated youth reports in considering the effects of hardship on development, youth perceptions of contextual factors have been considered in other fields of study. In fact, emerging work suggests that self-perceptions of experiences (rather than more objective measures, or measures derived from other reporters) are potentially more powerful predictors of outcomes. For example, there is an abundance of research showing that youths’ feelings of discrimination relate to their socioemotional wellbeing (Bogart et al., Reference Bogart, Elliott, Kanouse, Klein, Davies, Cuccaro, Banspach, Peskin and Schuster2013; Brody et al., Reference Brody, Chen, Murry, Ge, Simons, Gibbons, Gerrard and Cutrona2006; Seaton et al., Reference Seaton, Neblett, Cole and Prinstein2013). Similar patterns have been found for peer victimization and neighborhood violence (Bouman et al., Reference Bouman, van der Meulen, Goossens, Olthof, Vermande and Aleva2012; Goldman-Mellor et al., Reference Goldman-Mellor, Margerison-Zilko, Allen and Cerda2016; Gromann et al., Reference Gromann, Goossens, Olthof, Pronk and Krabbendam2013). Related to economic circumstances, the journal Developmental Psychology, devoted a special issue to highlight the importance of youth perceptions of economic inequality in relation to developmental outcomes (Ruck et al., Reference Ruck, Mistry and Flanagan2019). Yet, when it comes to experiences of economic hardship, the vast majority of studies only used parent reports of the material hardships and financial stress that economic disadvantage places on families. Thus, research needs to acknowledge that the occurrences of hardship and strain related to economic disadvantage are not experienced identically across individuals within a family. Indeed, our work illustrates that it is necessary to include youth perceptions in studies of their development.
Associations between hardship and youth behavior
Perhaps the most striking takeaway from this study is that youth reports of material deprivation and, to a lesser extent, financial stress are associated with their externalizing and internalizing behaviors, while caregiver reports of material deprivation are not. Specifically, consistent with prior research (Delgado et al., Reference Delgado, Killoren and Updegraff2013; Schenck-Fontaine & Panico, Reference Schenck-Fontaine and Panico2019; Wadsworth & Compas, Reference Wadsworth and Compas2002; Zilanawala & Pilkauskas, Reference Zilanawala and Pilkauskas2012), our results show that lower reported levels of youth hardship, particularly material deprivation, are associated with fewer behavior problems. This is true both when looking at youth compared to their own reports over time (within-associations) and youth compared to peers’ reports (between-associations). This finding solidifies the case for interviewing youth themselves about economic hardship (and specifically hardships that are most impactful to their daily lives), as well as other important processes, when conducting research on adolescent development.
Notably, looking within-individuals, adolescents’ reports of experiences of material deprivation, like missing out on school trips or outings with friends due to lack of funds, being unable to get necessary supplies for school because they/their family did not have money, lack of space for privacy at home, and failure to see a medical or mental health professional because they could not afford it, predicts their internalizing, while their financial stress, like fighting with caregivers about money and worrying about caregivers inability to pay for things, predicts externalizing behaviors (material deprivation, but not financial stress, predicted both types of behavior problems across youth). Perhaps during adolescence, when youth are increasingly focused on peer acceptance and social comparisons and more frequently internalize the views of others (Albert et al., Reference Albert, Chein and Steinberg2013; Jacobs, Reference Jacobs2003; O’Brien & Bierman, Reference O’Brien and Bierman1988; Rivenbark et al., Reference Rivenbark, Arseneault, Caspi, Danese, Fisher, Moffitt, Rasmussen, Russell and Odgers2020; Wrzus et al., Reference Wrzus, Hänel, Wagner and Neyer2013), the lack of these material goods and experiences, which are outwardly apparent to others, affects youths’ image and self-concept and, in turn, leads to internalizing symptoms. On the other hand, our measure of financial stress includes interactions with caregivers that may be conflictual (e.g., arguing about money). These conflicts may prime adolescents to be confrontational or aggressive in other areas of their lives, which may manifest as externalizing behaviors. This supports developmental psychopathologists’ theories that the mechanisms through which different aspects of adversity or economic disadvantage affect psychopathology vary (e.g., Miller et al., Reference Miller, Sheridan, Hanson, McLaughlin, Bates and Lansford2018; Smith & Pollak, Reference Smith and Pollak2021).
Overall, looking at both within- and between-variability in hardship, adolescents’ feelings of material deprivation were more predictive of their behavior problems than feelings of financial stress were, which is contrary to our hypothesis that both measures would predict internalizing and externalizing. This is important information for researchers moving forward; in a literature that tends to overlook adolescents’ own experiences and, even when studies do ask youth to report on hardship, they do not focus on material needs and wants that may be acutely triggering for adolescents, instead focusing on more “adult” hardships like arguing about money, being stressed about finances, having plumbing or electrical problems, and bill payment. In the future, studies should include some of these “adolescent” issues in measures of hardship when trying to understand adolescent mental and behavioral health.
While teens’ behavior problems were robustly predicted by their perceptions of hardship, and particularly their perceptions of material deprivation, remarkably caregivers’ reports of material deprivation were unrelated to youth behavior problems. This is notable since actual household income was more closely correlated with caregiver-reports of hardship than with youth-reports, and household income is more connected to “actual” family finances. Moreover, compared to youth reports of material deprivation, the parent material deprivation measure captures experiences that seem more dire and potentially harmful to children’s well-being, like food insecurity and going without utilities and/or adequate medical care. However, our study suggests caregivers’ experiences of hardship (like plumbing problems, food insecurity, or inability to pay bills) may not be as impactful for youths’ behavior as their own experiences of hardship that affect their daily lives, like interactions with friends, their personal space and things, and school experiences. Indeed, these are likely to be more salient to adolescents.
Of course, it is important to recognize that, since youths’ perceptions of hardship and behavioral functioning were both reported on by youth themsleves, associations between youth reported material deprivation and financial stress and behavior problems could be artificially inflated by single-reporter bias. Future research could use third party reports, like administrative records of school discipline, teacher reports of behavior, interactions with the juvenile justice system, and mental health diagnoses to more thoroughly substantiate these findings (De Los Reyes & Ohannessian, Reference De Los Reyes and Ohannessian2016). However, those metrics also have considerable biases (Auguste et al., Reference Auguste, Bowdring, Kasparek, McPhee, Tabachnick, Tung and Galán2023; Skiba et al., Reference Skiba, Horner, Chung, Rausch, May and Tobin2011, Reference Skiba, Chung, Trachok, Baker, Sheya and Hughes2014). Additionally, it is unlikely that single-reporter bias is the sole explanation for the predictiveness of youth reports of hardship. Adolescence reflects a period of social cognitive development when thoughts become more complex and the links between thoughts about experiences and behaviors become stronger (Crone & Dahl, Reference Crone and Dahl2012; Keating, Reference Keating2004; Weimer et al., Reference Weimer, Warnell, Ettekal, Cartwright, Guajardo and Liew2021). Hence, associations between youth perceptions of hardship and their externalizing and internalizing behavior likely reflect the general strengthening of associations between thoughts and behavior that occur during adolescence. Further, as children age, they become more attuned to their economic circumstances (Brown & Larson, Reference Brown and Larson2009; Flanagan & Gallay, Reference Flanagan, Gallay and Benson2014; Hagquist, Reference Hagquist2007; Schoon & Heckhausen, Reference Schoon and Heckhausen2019).
Of note, despite findings between youth perceptions of hardship and externalizing and internalizing, we did not find significant associations between monthly income and behavior outcomes. This is contrary to the robust literature on direct associations between income and youth behavior outcomes (Akee et al., Reference Akee, Copeland, Keeler, Angold and Costello2010; Blau, Reference Blau1999; D’Onofrio et al., Reference D’Onofrio, Goodnight, Van Hulle, Rodgers, Rathouz, Waldman and Lahey2009; DeSilvey, Reference DeSilvey2021; Dearing et al., Reference Dearing, McCartney and Taylor2006; Duncan et al., Reference Duncan, Kalil and Ziol-Guest2017; Hao & Matsueda, Reference Hao and Matsueda2006; Woods-Jaeger et al., Reference Woods-Jaeger, Hampton-Anderson, Christensen, Miller, O’Connor and Berkley-Patton2022). However, it is consistent with other research that finds more modest correlations between hardship and other characteristics of socioeconomic status, like income (Gershoff et al., Reference Gershoff, Aber, Raver and Lennon2007). Nor did we find links between caregiver reported material deprivation and youth behavioral outcomes in contrast to much of the literature on hardship and behavior problems using a similar deprivation measure (Gershoff et al., Reference Gershoff, Aber, Raver and Lennon2007; Schenck-Fontaine & Ryan, Reference Schenck-Fontaine and Ryan2022). Since it could be argued that youth perceptions of hardship act as a pathway through which family income or caregiver perceptions of hardship impact children, we ran post hoc analyses dropping both measures of youth perceptions of hardship from the models to assess whether income or caregiver perceptions of material deprivation predicted youth behavior without controlling for youth perceptions. Neither predicted even when youth perceptions of hardship were dropped from the model. Thus, our study supports that adolescents’ experiences with their income level and economic context, as measured through material and financial hardship, can be just as or more consequential than actual income (McLoyd et al., Reference McLoyd, Kaplan, Purtell, Bagley, Hardaway and Smalls2009; McLoyd, Reference McLoyd1990, Reference McLoyd, Carlo, Crockett and Carranza2011; Schenck-Fontaine & Panico, Reference Schenck-Fontaine and Panico2019). This further emphasizes the importance of measuring both income and feelings of hardship when examining the role of economic disadvantage in youths’ development (Boushey et al., Reference Boushey, Brocht, Gundersen and Bernstein2000; Gershoff et al., Reference Gershoff, Aber, Raver and Lennon2007; Sullivan et al., Reference Sullivan, Turner and Danziger2008).
Limitations and future directions
Our work leveraged an intensive, repeated measures design which importantly allowed us to look at within-dyad associations on a temporally valid timescale. However, elements of our study had some limitations that are critical to highlight. First, we followed 104 youth–caregiver dyads, which is a relatively small sample size compared to some other studies of financial contexts and youth behavior using very large, nationally representative samples like the National Longitudinal Survey of Youth or the Early Childhood Longitudinal Studies (Gershoff et al., Reference Gershoff, Aber, Raver and Lennon2007; Miller et al., Reference Miller, Betancur, Whitfield and Votruba-Drzal2021; Miller & Votruba-Drzal, Reference Miller and Votruba-Drzal2017). Nor was this sample representative of any particular population of interest. Additionally, we were only able to follow dyads for a period of nine months. This leaves us unable to look at long-term implications of youth perceptions of hardship. This will be an important next step to more deeply understand relations between hardship and behavior development.
Next, our caregiver- and youth-reported material deprivation questionnaires were not identical, and we did not have a parallel measure of caregivers’ perceptions of financial stress. With respect to the material deprivation measures, we decided to use questionnaires that were validated in the particular age ranges we were sampling in order to ensure strong measures of hardship. In doing so, however, we selected measures of material deprivation comprised of different items (and we did not have caregiver items assessing financial stress). Thus, while we believe the caregiver and youth measure of material deprivation capture similar underlying constructs, some of the variation in hardship scores across reporter may certainly be due to the use of different items. Additionally, items for the caregiver reports of hardship included a mix of dichotomized indicators and Likert-type indicators, which we dichotomized for consistency, which may have influenced reliability in the measure (α = .75−.78). Relatedly, the reliability of our measure of internalizing symptomology was relatively low (α = .66−.74). Inclusion of alternative assessments of internalizing psychopathology may better capture potential relations and could be an important next step.
Additionally, some of our reported effects (particularly the within-individual effects) are very modest in magnitude. It will be critical for additional work to estimate associations between youth perceptions of hardship and their behavioral functioning to contextualize these findings. Given the sampling design of the study, we believe it is still notable that there is reasonable variability on a sub-annual basis for many critical constructs, and this variability significantly influences adolescent outcomes. Furthermore, there may be some reason to believe that the observed effect sizes were, in part, a function of our sample and method. First, despite our purposive sampling to ensure at least half of the sample was low-income, our resulting caregivers may be a relatively socioeconomically privileged group (e.g., 37% had an advanced degree and a third were not in the labor force). This may have resulted in less experiences or lower intensity of hardship, which could have deflated effect sizes. However, we must note that many people with advanced degrees or who are not in the labor force (which include students and people unable to work for health reasons) still struggle financially (Baum, Reference Baum2014; Carnevale et al., Reference Carnevale, Cheah and Wenzinger2021; Schofield et al., Reference Schofield, Shrestha, Percival, Passey, Kelly and Callander2011). And importantly, many middle-class families are currently experiencing economic hardship; for example, 20% of middle-income families reported receiving unemployment benefits in 2020 (Kochhar & Sechopoulos, Reference Kochhar and Sechopoulos2022; Krause & Sawhill, Reference Krause and Sawhill2018). Also, our measures of hardship were assessed near the start of the month – which is when most public benefits are disbursed in the state where participants lived. Perhaps we would have observed higher levels of and more variability in our hardship measures if we assessed it near the end of the month. To the extent that perceptions of hardship were lower and less variable in our sample due to the assessment timeframe, this could have led to reduced effect sizes. It is also important to think about the impact of our effects at scale and over time (Götz et al., Reference Götz, Gosling and Rentfrow2022; Prentice & Miller, Reference Prentice and Miller2016). Associations between hardship and behavior problems may be stronger depending on the persistence and chronicity of lower income and greater hardship. Greater cumulative exposure to hardship would likely relate to greater behavioral problems. While modest in our reported results, accumulation of different negative effects could significantly shape outcomes over time.
Furthermore, this study was correlational, as opposed to experimental, in design. Thus, our results should not be interpreted as causal. Economic hardship is not randomly assigned across families. Caregivers’ economic conditions are actively shaped by larger social contexts and caregiver choices that affect their earnings, education, and employment, and factors influencing their selections also shape the proximal contexts in which their children develop (Davis-Kean, Reference Davis-Kean2005). However, using a multilevel structural equation modeling framework reduced omitted variable bias (Angrist & Krueger, Reference Angrist, Krueger, Ashenfelter and Card1999; Duncan et al., Reference Duncan, Magnuson and Ludwig2004). In particular, by estimating both the within- and between- dyad associations, this study addressed omitted variable bias stemming from time invariant characteristics of dyads, which are held constant in the within-dyad component of the model. To our knowledge, this is the first study considering links between hardship and adolescent development that has utilized this type of estimation. In this regard, this study helps to move the literature forward. At the same time, we recognize that our findings may still be biased by the influence of unobserved time varying characteristics of parents or youth that are correlated with hardship and youth externalizing or internalizing.
Finally, data collection for this study overlapped significantly with the start of the COVID-19 pandemic. The pandemic introduced unique conditions that influenced families’ circumstances and experiences related to our study (Browne et al., Reference Browne, Wade, May, Jenkins and Prime2021; Donker et al., Reference Donker, Mastrotheodoros and Branje2021; Patel et al., Reference Patel, Nielsen, Badiani, Assi, Unadkat, Patel, Ravindrane and Wardle2020). For example, family income may have been negatively affected by factors such as job loss, but hardship may have been lessened by COVID-19 safety-net policies, like more comprehensive unemployment benefits, government stimulus payments, pauses on student loan payments, etc. But families’ perceptions of hardship were almost certainly shaped by factors such as stressful in-person working/school conditions, increases in remote work/school, lack of childcare, and less in-person social interaction. Thus, it is possible that findings from our study would differ if data collection did not overlap with the COVID-19 pandemic. However, by analyzing monthly longitudinal data, our study was able to measure and control for variation in our key constructs before and during the height of the pandemic. Nonetheless, it is important to contextualize our findings in the current economic climate (i.e., post-the height of the pandemic). While the economy has largely rebounded, there are still groups of workers that have not recovered pandemic losses (Center for Budget and Policy Priorities, 2023). For instance, employment rates for workers with lower levels of education still have not reached pre-pandemic levels (Center for Budget and Policy Priorities, 2023). Coupling that with the fact that inflation peaked a couple years after the pandemic and is currently much higher than it was during data collection, we conjecture that many of the hardships experienced by families in this study that may have been precipitated by the pandemic are continuing to impact the lives of low- and middle-income families.
Conclusion
Economic instability and hardship are on the rise (Schaeffer, Reference Schaeffer2020); so too are youth mental health problems (Murthy, Reference Murthy2022). Yet, the role of youth perceptions of hardship in shaping their behavioral development has garnered little research attention. This study addresses this gap by estimating associations between variation in monthly measures of caregiver reports of material deprivation and youth reports of material deprivation and financial stress, and then linking variability in reports of hardship to youth behavior problems. We found significant variability in youth- and caregiver-reports of hardship and found that youth perceptions of hardship (but not caregiver reports) predicted youth externalizing and internalizing. Results illustrate the importance of accounting for youths’ own experiences of material deprivation and financial stress when thinking about the ways in which families’ economic conditions relate to youth behavior. Understanding the mechanisms driving youth problem behaviors is vital in efforts to fight any negative consequences of growing economic inequality and hardship on youth and families.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S0954579423001451.
Funding statement
Work was funded by a grant from the Eunice Kennedy Shriver National Institute of Child Health and Human Development (R01HD110423) and also an internal grant from the Learning Research and Development Center, University of Pittsburgh.
Competing interests
No authors report conflicts of interest.
Open access
