Introduction

The national and international migrant stock has proliferated due to dramatic economic growth and socioenvironmental changes (International Organization for Migration, 2022). Millions of labor-aged adults worldwide live and work outside their places of origin, but economic pressures and structural restrictions force these adults to leave behind their school-aged adolescents in their original communities (Antia et al., 2020; Fellmeth et al., 2018). Existing literature has predominantly focused on left-behind adolescents' psychological vulnerabilities and social difficulties (Wang & Mesman, 2015; Wang et al., 2019a, b), with a much sparser body of research focusing on the group’s academic engagement (school involvement and learning commitment). Identifying the impact of parental migration on youth's academic engagement and investigating related risk and protective factors (e.g., empathy) remain open empirical questions with inconclusive evidence. Addressing these research questions is essential to facilitating healthy and successful transitions for left-behind youth and developing a support system to help them cope with any detrimental effects of parental absence, especially for the youth’s academic engagement. Additionally, the research findings generated from the current study are potentially informative for educators and practitioners who must consider the specific circumstances of left-behind youth, providing insightful instructions on developing personalized intervention or prevention programs to facilitate the academic engagement of such youth.

The large-scale left-behind adolescents in China, a country that emphasizes academic achievement and educational success, provides a suitable and important study context. In the past four decades, China has been experiencing an unprecedented volume of labor migration, with approximately 61 million school-aged adolescents being left behind in their hometowns or villages (Ge et al., 2019; Wang & Mesman, 2015; Wang et al., 2019a, b). Such a high number of left-behind youth occurs as one or both of their biological parents migrate to urban mega-cities to find better job opportunities and create a better future for the adolescents and the family (Lan, 2022; Wang & Zhao, 2022). These left-behind adolescents in the original communities often stay with surrogate caregivers, such as grandparents and close relatives, and experience several emotional and behavioral difficulties (Wang & Mesman, 2015; Wang et al., 2019a, b).

Leveraging this important sociocultural context, in the current study, I compared the academic engagement of left-behind adolescents and their non-left-behind peers, and subsequently examined the unique and interactive relationships among affective empathy, cognitive empathy, and left-behind status with academic engagement in a combined sample of left-behind and non-left-behind youth. In what follows, I provide a literature review pertaining to key study variables (i.e., academic engagement and empathy), highlight the developmental considerations, and formulate specific hypotheses leading to further empirical investigation at the end of the introduction.

Academic engagement

Academic engagement, as a multifaceted and malleable construct, is defined as involvement in school or commitment to learning (Christenson et al., 2012; Fredricks et al., 2004; Lam et al., 2012). In the current study, I focused on an aggregated definition of academic engagement as adolescents' overall evaluation of academic involvement and learning commitment across affective, behavioral, and cognitive aspects. Prior research has shown that being academically engaged facilitates students' motivation, critical thinking, academic performance, subjective well-being, and longer-term career success (Datu & King, 2018; Greenwood et al., 2002). Therefore, comparing school-aged youth’s academic engagement based on their parental migration status (left-behind versus non-left-behind) is essential to instantly tackle the potential vulnerability left-behind youth may encounter and prevent them from long-term academic underachievement.

Prior empirical research has yielded inconclusive evidence, indicating that the effect of parental migration on left-behind adolescents' academic behaviors and outcomes can be either beneficial or adverse. For instance, a qualitative study by Fu and Zhu (2020) found that left-behind adolescents can be as successful in the school setting as their non-left-behind peers, regardless of parental migration status. The positive impact of parental migration on left-behind adolescents' academic performance can also be found in Bai et al.’s (2018) findings. The authors of such studies have suggested that labor migration is regarded as economically beneficial to those who stay behind due to remittances, which can support household consumption and investment. This economic support is particularly important for school-aged adolescents to access public education and resources. Likewise, in East Asian families, for instance, Chinese parents, regardless of socioeconomic situations and migration status, generally prize education as an essential way to achieve upward social mobility and maintain family dignity (Liu et al., 2020; Wang et al., 2022).

In contrast, several empirical studies have identified the opposite trend. For example, Song et al. (2018) showed that left-behind adolescents exhibit lower academic achievement than their non-left-behind counterparts. Similar findings can also be found in a more recent study by Shen et al. (2021). These studies indicate that left-behind adolescents endure several academic challenges due to the lack of immediate parental involvement and direct supervision. This detrimental effect of parental absence coincides with developmental tasks during adolescence as salient biopsychosocial changes, school transitions, and academic demands create significant challenges to adolescents’ engagement (Waddoups et al., 2019). Further research on comparing academic engagement between left-behind and non-left-behind youth is warranted to clarify these divergent findings.

Such disparate findings in the literature may partially stem from individual differences concerning affective and cognitive characteristics. One important but less studied factor that deserves further academic scrutiny is empathy, as elaborated upon below.

Affective and cognitive empathy

Empathy refers to the capacity to feel or understand the actual or expected emotional states of someone else (Baron-Cohen & Wheelwright, 2004; Davis, 1983; Hoffman, 2000). As a multidimensional construct, empathy has been conceptualized by two main components: affective and cognitive empathy. The dramatic biopsychosocial changes that occur during adolescence help youth develop refined affective and cognitive capacities and enable the youth to simultaneously view themselves and others, providing the developmentally practical meaning of studying both affective and cognitive empathy (Allemand et al., 2015; Baron-Cohen & Wheelwright, 2004; Eisenberg et al., 2006; Hoffman, 2000). Specifically, affective empathy means to compassionately and sympathetically understand others’ thoughts and feelings in response to the others' misfortunes, while cognitive empathy means to spontaneously understand the others' points of view and mental states using perspective-taking (Davis, 1983; Eisenberg et al., 2006). Prior theoretical and empirical evidence has consistently shown that the development of both empathy components generally plays a vital role in maintaining close relationships and promoting prosocial tendencies (Ding & Lu, 2016; Zaki, 2020).

Apart from its relations with socioemotional outcomes, empathy has also been proven to be highly relevant to academic behaviors and outcomes (Feshbach & Feshbach, 2009). Prior empirical research has predominantly centered on the role of teachers’ global empathy in facilitating students’ learning and academic outcomes (Zhang, 2022). In the school setting, emotionally empathic teachers enable students to experience greater affective understanding and facilitate effective communication between teachers and their students (Feshbach & Feshbach, 2009). Effective teacher-student communication is critical not only to establish positive relations between teachers and students but also to enable the co-construction of academic knowledge (Smogorzewska et al., 2022). Despite these findings, comparatively less empirical attention has been directed toward the association between adolescents’ perceived empathy and their academic engagement. Additionally, although conceptually interrelated, affective and cognitive empathy are distinct constructs; they have different neural underpinnings (Fan et al., 2011; Shamay-Tsoory et al., 2009) and relate to different socioemotional and behavioral outcomes (Huang & Su, 2014; Jiang et al., 2021; Van Lissa et al., 2017a, 2017b; Yan et al., 2021). Simultaneously investigating the separated form of empathy subcomponents in adolescent’s academic engagement is therefore vital to developing a comprehensive understanding of this relation.

In addition to the unique effects of affective and cognitive empathy on academic engagement, the interactive role of these distinct forms cannot be ignored. Prospective longitudinal studies have provided empirical support concerning the developmental order of affective and cognitive empathy, with affective empathy often preceding cognitive empathy (Heyes, 2018; Van Lissa et al., 2014). Due to developmental incongruence regarding two components of empathy, an interactive pattern may exist between them, explaining the variance of academic engagement. Although few in number, existing empirical research has also provided support for examining this interactive pattern, but engendered inconsistent findings (Ang & Goh, 2010; Bos & Stokes, 2019; Weisz & Cikara, 2021). For instance, Ang and Goh (2010) identified an “additive” interactive pattern between affective and cognitive empathy in cyberbullying; adolescents with congruently low affective and cognitive empathy reported the highest levels of cyberbullying. More recently, Bos and Stokes (2019) discovered a “cross-over” interaction of affective and cognitive empathy on autistic adolescents’ well-being. Specifically, adolescents with high cognitive empathy were sensitive to how affective empathy influenced well-being, for better and for worse. In the presence of low affective empathy, adolescents with high cognitive empathy exhibited the lowest level of well-being. In contrast, in the presence of high affective empathy, adolescents reported the highest (Bos & Stokes, 2019). This cross-over effect has also been noted in prior research (Tone & Tully, 2014; Van Lissa et al., 2017a, b), highlighting the emotional cost of high empathy. For instance, adolescents with high empathy tend to be sensitive to negative stimuli and cues in an unfavorable context. Given the complexity, these inconsistent interactive patterns still merit further scholarly investigation. Building on and extending this body of work requires addressing the interactive effects of affective and cognitive empathy on adolescents' academic engagement.

Based on the extant research mentioned above, academic engagement is conceptually acknowledged to be contingent on both unique and interactive relations of affective and cognitive empathy. The specific influence, varying with other characteristics (e.g., the left-behind status of the individual), however, remains underexplored. Studying empathy involves unique values for left-behind youth to foster their social interactions and integration by allowing the youth to understand themselves and others' emotions and helping the youth employ better emotion regulation strategies (Eisenberg et al., 2006; Main et al., 2017). Empathic adolescents may better place themselves in the role of their migrated parents and exhibit better understanding regarding the migrated parents' intentions to lift family income and wellness by working away from home. These shared thoughts and feelings may underline and reinforce what the left-behind youth have experienced, resulting in better academic engagement as a reward for their parents' rearing and hard work. This line of argument has been backed up by empirical research (Van Lissa et al., 2016), showing that both empathy dimensions are positively associated with reduced parent–child conflict and increased parent–child cohesion. Despite such plausible evidence, the unique and interactive relationships between the multidimensional perspective of empathy and academic engagement in left-behind youth, as compared with their non-left-behind peers, remains unclear. In that regard, the present study aimed to address these knowledge gaps in the extant research to highlight the developmental commonality and specificity in adolescence (Malti & Cheah, 2021)—a developmental phase deserving close scholarly attention.

Looking through the lens of adolescent

During adolescence,Footnote 1 youth experience dramatic biological growth and cognitive advances, major social role transitions, and pressing educational tasks (Lerner & Steinberg, 2004; Wang & Fredricks, 2014). Prior scholarship also indicated that adolescents often exhibit decreased academic engagement and marked increases in emotional and behavioral difficulties (Eisenberg et al., 2006; Li & Lerner, 2011; Wang & Eccles, 2012). Among these difficulties, academic disengagement is particularly concerning because, during this life period, most adolescents must be actively engaged in the learning process to acquire sufficient knowledge and skills, equipping adolescents with the capacity to pursue higher education or find a decent job (Wang & Eccles, 2012). During this “storm and stress” period, youth with left-behind experiences may carry additional burdens due to parental absence. Therefore, investigating the correlates of academic engagement in a combined sample of left-behind and non-left-behind youth is important since the critical role of active academic engagement serves as not only a facilitator of academic achievement but also a protective factor in a broad spectrum of positive youth development (Li & Lerner, 2011; Wang & Eccles, 2012).

During adolescence, sociocognitive advances foster the development of self-awareness, emotional understanding, and theory of mind (Eisenberg et al., 2006). These advancements collectively enable adolescents to develop a greater capacity to reflect on their own internal states and think from the perspective of others (Eisenberg et al., 2006), and ultimately a greater ability to develop affective and cognitive empathy (Allemand et al., 2015; Stern & Cassidy, 2018). Therefore, addressing the unique and interactive relationships between distinct empathy subcomponents and left-behind status with academic engagement during adolescence is developmentally appropriate and meaningful.

The present study

The objectives of the present study were twofold. First, this study compared the academic engagement between left-behind youth and their non-left-behind peers. Second, this study investigated the unique and interactive patterns of affective empathy and cognitive empathy on academic engagement. In addition, to highlight the developmental commonality (similar) and specificity (different) of study associations, a combined sample of left-behind and non-left-behind youth was analyzed regarding the left-behind status as a moderator in the previously mentioned associations. Figure 1 presents an illustrative figure of these two ambitions.

Fig. 1
figure 1

An illustrative figure displaying two research objectives

Regarding the first objective, I hypothesized that left-behind youth might report higher (Hypothesis 1a) or lower (Hypothesis 1b) levels of academic engagement than their non-left-behind counterparts, given the divergent findings in the extant literature.

Regarding the second objective, I hypothesized a positive main effect, namely that both empathy components would be uniquely and positively related to academic engagement (main effect; Hypothesis 2a). Additionally, I hypothesized two plausible interaction patterns (i.e., additive effect and cross-over effect) regarding the interactive patterns of affective and cognitive empathy due to the dearth of consistent empirical evidence. Specifically, I hypothesized that adolescents with congruently low affective and cognitive empathy would report the lowest levels of academic engagement, or perhaps adolescents with congruently high affective and cognitive empathy would report the highest (additive effect; Hypothesis 2b). I also hypothesized that adolescents with high cognitive empathy, in the context of low affective empathy, might report the lowest levels of academic engagement; in contrast, in the context of high affective empathy, might report the highest (cross-over effect; Hypothesis 2c). Finally, as both empathy components contain especial values on left-behind adolescents, I hypothesized that either additive or cross-over effect would be pronounced in left-behind youth (versus non-left-behind youth) (three-way interaction; Hypothesis 2d).

When testing these hypotheses, I considered several sociodemographic covariates—including age, gender, parental education background, and family affluence—based on prior research. For instance, academic engagement tends to decrease with age (Wang & Eccles, 2012), and female adolescents and those with high family socioeconomic status (e.g., high parental education and affluent family economic situations) are likely to have high academic engagement (Li & Lerner, 2011; Wang & Eccles, 2013). To ensure the robustness of focal study associations, these variables must therefore be accounted for statistically. Additionally, when testing the second hypotheses, I adjusted for left-behind youth’s length of separation from parents as youth separated from their parents during the initial years often encounter significant challenges (Wang & Liu, 2020).

Method

Participants and procedure

The current study was based on an ongoing prospective research project to systematically investigate the contextual and individual correlates of youth's socioemotional and academic functions. Prior to data collection, ethical approval was granted by the second author’s affiliation. Informed consent forms or assent were obtained from all the participants involved in the current investigation. The answers for this investigation were highlighted as neither right or wrong; I intentionally used this instruction to potentially decrease the influence of socially desirable responding.

In total, I distributed approximately 1300 questionnaires, and a final valid sample of 1060 was collected, with a response rate of 81.5%. This high response rate was partially due to well-established school collaborations on an annual basis; in that regard, school authorities, teachers, and students were highly motivated to participate in this feedback-oriented investigation. During the assessment, participants attended public primary and middle schools in China. All participants came from low-to-medium family income backgrounds, as evidenced by the family affluence scale (Mean = 3.79; SD = 2.00; scores ranged from 0 to 9), and most of their parents completed secondary school education. Among these participants, 323 (Mean age = 13.02; SD = 1.77; age ranged from 10 to 18; 44.6% females) were identified as left-behind youth, with 40.6% both-parents migration, 3.4% mother-migration only, and 56.0% father-migration only. The average separation duration was 3.93 years (SD = 3.80). Recruiting from the same classrooms, 737 non-left-behind adolescents (Mean age = 13.07; SD = 1.60; age ranged from 10 to 18; 51.7% females) living with two biological parents were identified and used in subsequent analyses.

Measures

Academic engagement

Academic engagement was measured by the Student Engagement Questionnaire (Lam et al., 2012). This questionnaire was further revised by Ma et al. (2015) to suit Chinese adolescents. This 16-item questionnaire contains three dimensions: affective engagement, behavioral engagement, and cognitive engagement. One item example is, “I like what I am learning in school (affective engagement)”. Items were rated from 1 (completely disagree) to 5 (completely agree). All items were summed to a composite score, and higher scores indicated higher engagement. Prior research on Chinese adolescents has exhibited good internal consistency in this questionnaire (e.g., Wang et al., 2022).

Affective empathy and cognitive empathy

Both aspects of empathy were assessed by the Chinese validation of the Basic Empathy Scale (Geng et al., 2012; Jolliffe & Farrington, 2006). Sample items include, “I get caught up in other people’s feelings easily (affective empathy)” and “I can often understand how people are feeling even before they tell me (cognitive empathy)." Response options ranged from from 1 (strongly disagree) to 5 (strongly agree). Summed scores were calculated for each dimension, with higher values indicating higher affective/cognitive empathy. Past research on Chinese adolescents has shown good internal consistency of this scale (Fang et al., 2020).

Left-behind status

I asked each adolescent to report their current left-behind status based on the following question with four categories: did your father and/or mother migrate to other cities working continually for a long period (at least six months)? (a) only your father migrated to another city for work, (b) only your mother migrated to another city for work, (c) both your parents migrated to another city for work, and (d) both parents are currently living together with me in the original communities. For those responding to (a), (b), or (c), an additional question was asked to indicate the length of separation from their parents (years). These items were adapted from prior research (Ma et al., 2022a, b).

Sociodemographic characteristics

Sociodemographic information—including age, gender, parental education, and family affluence levels— were also collected from the participants. Two items with three categories were used to gather the information regarding the youth’s parental education background: (a) secondary school or lower, (b) high school, and (c) undergraduate education or higher. A summed score of these two items was used to represent parental education background, and higher scores were indicative of higher parental education. Additionally, family affluence was assessed by a 4-item Chinese-validated family affluence scale (Boyce et al., 2006; Liu et al., 2012). Summed scores of these items were used, with higher values indicating higher family affluence.

Data analysis

Data analyses were performed in SPSS 28.0 (IBM Corp., 2021) and R (R Core Team, 2021).Footnote 2 Missing values were first estimated by the Little’s Missing Completely at Random test (Little, 1988), and subsequently replaced using the expectation–maximization algorithm.

Prior to the analyses that aimed to address research hypotheses, descriptive statistics, including means, standard deviations, values of skewness and kurtosis, were first estimated for study variables. Their bivariate correlations were calculated by Person’s correlations. The internal consistencies of key measurements were estimated by McDonald's Omega (Hayes & Coutts, 2020). Independent t-tests or Chi-square tests were used to estimate group differences (left-behind youth vs. non-left-behind youth) in study variables.

In terms of the first research objective, ANCOVA was used—after adjusting for age, gender, parental education, and family affluence—to compare the academic engagement between left-behind and non-left-behind youth.

In terms of the second research objective, a four-step hierarchical linear regression model was used to examine the unique and interactive effects of affective empathy, cognitive empathy, and left-behind status on academic engagement.Footnote 3 The regression model was established based on a sequential manner: in the first step, only confounding variables were entered into the model; in the second step, main effects were additionally considered; in the third step, the two-way interactions among study variables were entered; in the final step, a three-way interaction was additionally added. Interactive patterns of significant two- or three-way interactions were further probed by simple slope analyses and figures, following Aiken and West's (1991) approach.

Results

Descriptive statistics and bivariate correlations

Means, standard deviations, values of skewness and kurtosis, and internal consistencies, as evidenced by McDonald’s Omega, are shown in Table 1. All variables showed acceptable ranges of skewness, kurtosis, and internal consistencies. Independent t or Chi-square tests indicated no significant differences in key study variables and sociodemographic characteristics between left-behind and non-left-behind youth, in addition to a significant gender difference, with more females in the left-behind group.

Table 1 Descriptive statistics

Bivariate correlations among study variables are presented in Table 2. Specifically, at the bivariate level, both affective and cognitive empathy were positively related to academic engagement in left-behind and non-left-behind youth; both types of empathy were also positively correlated in both groups of youth. In terms of covariates, higher age was significantly related to decreased academic engagement, and higher parental education was significantly related to higher academic engagement in left-behind and non-left-behind youth. Finally, for non-left-behind youth only, higher family affluence was significantly correlated with higher academic engagement.

Table 2 Bivariate correlations

Group differences in academic engagement

The results based on ANCOVA showed no significant differences in academic engagement between left-behind youth and their non-left-behind counterparts (F = 2.14, p = 0.14), after adjusting for age, gender, parental education, and family affluence. As visualized in Fig. 2, observed data points and distribution in academic engagement between left-behind and non-left-behind youth were similar. Neither hypothesis 1a nor hypothesis 1b was supported.

Fig. 2
figure 2

Academic engagement between left-behind youth and non-left-behind youth. N = 1060. Central red dot represents mean, bar/line corresponds to median, band indicates interquartile range, bean indicates data distribution, and scattered dots correspond to observed data

Unique and interactive relations of affective empathy, cognitive empathy, and left-behind status with academic engagement

The results of hierarchical linear regression are reported in Table 3.

Table 3 Hierarchical regression analysis predicting academic engagement

As shown in Table 3, in the first step (covariates only), age was negatively related to academic engagement, whereas family affluence was positively correlated with academic engagement. The other covariates were not significantly related to academic engagement. Altogether, the covariates explained 4% variance of academic engagement. In the second step (main effect), affective empathy and cognitive empathy were each positively related to academic engagement (Hypothesis 2a was supported), whereas there was no significant correlation between left-behind status and academic engagement. The main effects additionally explained 13% variance of academic engagement. In the third step (two-way interactions), the interactions between affective empathy and cognitive empathy, and between affective empathy and left-behind status were positively related to academic engagement, whereas the interaction between cognitive empathy and left-behind status was negatively related to academic engagement. The three significant two-way interactions additionally explained 2% variance of academic engagement. Although the two-way interactions between affective empathy and left-behind status, and between cognitive empathy and left-behind status were significant, I did not interpret them further, as these interactive terms were established in the linear regression as a precondition to high-order interaction (i.e., three-way interaction), but were not considered as the current research objectives.

The significant interaction between affective and cognitive empathy was evaluated and interpreted by simple slope analyses and visualized figures (see Fig. 3). Results showed that the positive association between affective empathy and academic engagement remained significant for both high (b = 0.90, SE = 0.13, t = 7.21, p < . 001) and low (b = 0.27, SE = 0.13, t = 2.13, p = 0.03) levels of cognitive empathy (see Fig. 3). From a descriptive perspective, adolescents reported the highest levels of academic engagement when affective and cognitive empathy were congruently high, thus supporting the additive effect (Hypothesis 2b).

Fig. 3
figure 3

Interaction effect of affective empathy and cognitive empathy on academic engagement. N = 1060. Cognitive empathy was divided into three levels based on mean and standardized deviation. Scattered dots in the figure correspond to observed data

In the final step (three-way interaction), the three-way interaction was positively related to academic engagement, additionally explained 1% variance of academic engagement (Hypothesis 2d was supported). Simple slope analyses have shown that, for left-behind youth, the positive relation between affective empathy and academic engagement was significant at high cognitive empathy (b = 1.25, SE = 0.21, t = 5.90, p < . 001), but not at low cognitive empathy (b = 0.34, SE = 0.21, t = 1.60, p = 0.11). The same pattern has been shown in non-left-behind youth at high cognitive empathy (high cognitive empathy: b = 0.56, SE = 0.13, t = 4.17, p < . 001; low cognitive empathy: b = 0.21, SE = 0.14, t = 1.44, p = 0.15). From a descriptive point of view (see Fig. 4), a cross-over effect was supported for left-behind youth, whereas an additive effect was supported for non-left-behind youth. Specifically, in the presence of high affective empathy, left-behind youth with higher cognitive empathy (vs. lower cognitive empathy) tend to report the highest academic engagement. However, in the presence of low affective empathy, left-behind youth with higher cognitive empathy (vs. lower cognitive empathy) were likely to report the lowest levels of academic engagement. By contrast, in the presence of high affective empathy, non-left-behind youth with higher cognitive empathy (vs. lower cognitive empathy) tend to report the highest academic engagement. In the presence of low affective empathy, non-left-behind youth were likely to report similar levels of academic engagement, regardless of the levels of cognitive empathy.

Fig. 4
figure 4

Interaction effect of affective empathy, cognitive empathy, and left-behind status on academic engagement. N = 1060. Cognitive empathy was divided into three levels based on mean and standardized deviation. Scattered dots in the figure correspond to observed data

Discussion

The unprecedented number of migrants worldwide has radically shifted the composition of societies, resulting in millions of left-behind adolescents remaining in their original communities. Given the significance of this societal issue, relatively few endeavors have sought to investigate how parental migration (left-behind versus non-left-behind) impacts youth's academic engagement and systematically estimate the unique and interactive links of affective and cognitive empathy with academic engagement in left-behind and non-left-behind youth. The current study added to the extant research by addressing these knowledge gaps, and the findings of this study are discussed below with respect to the specific hypotheses developed.

The first objective was to compare the academic engagement between left-behind and non-left-behind youth. Against the first hypothesis, these two groups of adolescents did not exhibit significant differences in academic engagement. This finding mirrored prior research (Chang et al., 2019), showing that parental migration has no significant influence on left-behind youth’s math achievement. In accordance with prior research (Fu & Zhu, 2020; Lan, 2022), the current findings have challenged the putatively vulnerable image of left-behind as it contains stigma and discrimination toward adolescents whose parents migrate to different areas for work. One possible interpretation of such a finding is related to remittance obtained from migrated parents. The improvement of the family's economic situation may relax household liquidity constraints and provide direct educational resources and investments for school-aged adolescents, which leads to active academic engagement (Bai et al., 2018). Another interpretation of such a finding is that academic engagement is highly prized and valued in Eastern societies, such as China. In such societies, active academic engagement is often perceived as a manifestation of filial piety and family glory (Wang et al., 2022). Likewise, achieving education success is still regarded as an effective way to achieve upward social mobility and obtain decent social status and high wages (Wang et al., 2018). Under such cultural influences, Chinese adolescents are, regardless of their left-behind status, strongly motivated to actively engage in learning and perform well at school. Notably, two groups of adolescents from the same schools and with similar sociodemographic backgrounds (e.g., family income and parental education background) may show somewhat of a decrease in the potential difference in academic engagement as these sociodemographic factors are highly correlated with adolescents’ academic engagement (Bradley & Corwyn, 2002). Addressing the first objective contributes to clarifying the divergent and inconclusive findings in the literature regarding how parental migration affects youth’s academic behaviors and outcomes. At the same time, focusing on academic-related outcomes enriches the extant research on left-behind youth that has predominantly focused on their psychosocial and behavioral perspectives.

The second research objective was to examine how the two distinct components of empathy uniquely and interactively contribute to academic engagement in left-behind and non-left-behind youth. In line with the second hypothesis (2a), both empathy components were positively related to academic engagement. In terms of affective empathy, adolescents with great understanding of others’ emotions are likely to form interpersonal emotional bonds between individuals and facilitate interpersonal harmony (Davis, 1983; Eisenberg et al., 2006). Harmonious peer relations and teacher-student interactions then further foster youth’s active engagement in academic activities. In terms of cognitive empathy, adolescents with great capacities for perspective-taking tend to better understand the mental states of others (Davis, 1983; Eisenberg et al., 2006). In the school setting, adolescents with high cognitive capacities for perspective-taking, compared to those with lower capacities, can better infer and critically evaluate the views of the authors in academic texts or characters, establishing a potential bridge to academic engagement (Gehlbach, 2004; Kim et al., 2018).

In addition to the main (unique) effects of both empathy components, this study also showed an additive interaction effect between affective and cognitive empathy on academic engagement, supporting Hypothesis 2b. Specifically, adolescents with congruently high affective and cognitive empathy reported the highest academic engagement. This finding aligns with the existing conceptual and empirical evidence highlighting that these empathy components work together to achieve desired outcomes (Ang & Goh, 2010; Davis, 1983). One possible explanation for such a finding is that high affective empathy (e.g., empathic concern toward others) may contribute to youth interpersonal skills improvement, whereas high cognitive empathy (e.g., perspective-taking) may facilitate adolescents’ own comprehension of characters and effectively transfer this knowledge from textbooks to actual life (Davis, 1983; Eisenberg et al., 2006). These interpersonal and intrapersonal processes tend to work in tandem to achieve the optimal outcome in school settings, such as active academic engagement.

Additionally, the current findings exhibited a three-way interaction, supporting Hypothesis 2d. The additive effect was conditional upon the left-behind status. Specifically, for non-left-behind youth, the additive pattern remained, whereas for left-behind youth, a cross-over interactive pattern was discovered. Left-behind youth with high cognitive empathy exhibited heightened susceptibility to both the beneficial effect of high affective empathy and the adverse effect of low affective empathy. This susceptibility may be partially explained by the differential susceptibility theory (Belsky & Pluess, 2009; Belsky et al., 2007), showing that left-behind adolescents are especially susceptible to both positive and negative stimuli (Ma et al., 2022a, b). Left-behind youth with low affective empathy may have interpersonal difficulties at school; in that context, those with high cognitive empathy, despite understanding their parents' good intentions for migration to achieve a better socioeconomic condition, may de-value the importance of the educational process and divert attention from academic activities to other work-related activities. Addressing the second objective adds to the literature discussing the linkage between empathy and academic outcomes in adolescence. Simultaneously focusing on the roles of affective and cognitive empathy also delineates a comprehensive picture of this relationship and clarifies the inconsistent interactive patterns in the literature. Additionally, examining these associations in a combined sample of left-behind and non-left-behind youth contributes to the discussion on developmental commonality and specificity of the association between distinct empathy subcomponents and academic engagement (Malti & Cheah, 2021).

Limitations and implications

These significant findings notwithstanding, several limitations should be considered when interpreting the current findings. First, for the parsimony of the linear model, this study solely focused on a global form of academic engagement, limiting a detailed investigation into the correlates of different components of academic engagement. Further studies should consider differentiating the multidimensional components of academic engagement (affective, behavioral, cognitive, and agentic) or simultaneously include academic disengagement (e.g., school dropout), delineating a fine-grained understanding of the relationship between both empathy subcomponents with academic engagement and disengagement. Second, this study was based on self-reported questionnaires, although the measurements employed have been validated in the Chinese cultural context and exhibited sound psychometric properties. The single methodology used, however, may inflate study associations, which warrants further multi-method study to confirm. Third, the research findings were backed up by a cross-sectional design, but the directionality of study associations cannot be inferred when using this research design. Prospective longitudinal studies should be deliberately conducted to clarify this issue. Finally, the current study focused on relatively wide age ranges, which do not provide a developmentally detailed picture of the relationships between empathy subcomponents and academic engagement in left-behind and non-left-behind youth. Future studies should therefore extend the current findings by leveraging a unique developmental phase, such as early or late adolescence, to illustrate the study associations under examination.

The above limitations notwithstanding, the present research contains important theoretical and practical implications. Regarding the theory, the current study mainly extends the existing literature on the socioemotional plights of left-behind youth by adopting a strength-based approach, centering on their active academic engagement. Likewise, this study advances extant scholarship regarding empathy as a monolithic construct, particularly contributing to theoretical discussions on a multidimensional approach to empathy and how affective and cognitive empathy, in isolation (or working together), relate to academic outcomes. The current study also expands on empathy theory by examining its developmental commonality and specificity based on various conditions, such as left-behind status.

Regarding practical implications, the current study indicates that left-behind youth may not be academically disadvantaged, challenging the globally vulnerable image of left-behind populations. This study instead suggests that mainstream media and policymakers should avoid pathologizing the social issues of the left-behind population and that school authorities and teachers should facilitate educational equality at school and help, whenever appropriate, decrease negative stereotypes established among peer groups.

Additionally, a multicomponent view of empathy allows educators and practitioners to develop a fine-grained picture of how affective and cognitive empathy uniquely and jointly relate to academic engagement (Weisz & Cikara, 2021). The current study therefore provides essential insight into when and for whom targeted affective and cognitive empathy training programs tend to facilitate active academic engagement. For instance, educators and practitioners may attempt to supervise students regarding the meaning of affective and cognitive empathy, to facilitate their understanding of others' emotions, and to take the perspective of others (Teding van Berkhout & Malouff, 2016). In terms of affective empathy, educators and practitioners should use situation-based materials to improve left-behind youth’s awareness of their own perspectives in response to the protagonist's experiences and misfortunes. Educators and practitioners may encourage students to share their feelings under various contexts, and other students in the same activities should practice active listening techniques and vicariously facilitate the feelings of others during this period. In terms of cognitive empathy, educators and practitioners should work on facilitating youth's awareness regarding others' perspectives—using role-playing techniques, for instance. Educators and practitioners can motivate students to openly and frequently share how they feel about a situation and why they feel that way. In this context, other students can also be free to express when and how they feel differently in this situation. Accordingly, students can learn how to accept feedback and practice empathy in front of different viewpoints. The implementation of affective and cognitive empathy measurement during these activities seems important to gauge students' dynamic changes in terms of the two components. Notably, for left-behind youth with low affective empathy, educators and practitioners should de-emphasize the role of cognitive empathy.

Conclusion

In conclusion, the current study challenges the globally vulnerable image of left-behind youth, indicating that they may not be academically disadvantaged. This study also highlights the unique and interactive roles of affective and cognitive empathy in academic engagement, and the developmental specificity of these associations highlights the distinct role of left-behind status herein. Specifically, congruently high levels of affective and cognitive empathy benefit youth’s academic engagement, regardless of their left-behind status. Yet incongruent levels of empathy characterized by low affective empathy but high cognitive empathy are harmful to left-behind youth's academic engagement.