談戀愛對青少年認知能力與非認知能力的影響：來自中國教育追蹤調查的證據

Effects of Romantic Involvement on Adolescents' Cognitive and Noncognitive Skills: Evidence from China Education Panel Survey

在西方國家，戀愛關係被看成青少年成長的一部分；在中國，青少年的戀愛關係被稱為早戀，通常被認為是不合適宜的。談戀愛真的對中國青少年成長不利嗎？基於中國教育追蹤調查（2013—2014學年、2014—2015學年）資料，使用傾向值加權的方法，本研究系統評估了談戀愛對中學生的認知能力與非認知能力的影響。認知能力以認知能力測試得分、語數外成績班內標準分、班級名次來測量；非認知能力以自我效能、抑鬱程度、對未來的信心、教育期望、學習態度來測量。研究發現，相對於未談過戀愛的中學生，談過戀愛的中學生在認知能力和非認知能力方面的表現顯著更差。在談戀愛影響認知能力的機制中，學業投入時間是一個重要的中介變項（mediator），談戀愛會透過減少學業投入時間來降低認知能力方面的表現。研究進一步發現，談戀愛的負面影響在女學生當中表現得更為明顯。戀愛捲入程度愈深，其給青少年帶來的負面影響愈大。

In Western countries, romantic relationships are part of the typical adolescence experience, whereas in China, adolescent romantic relationships are disparaged as ＂puppy love＂ and often considered far from malapropos (Li et al., 2010; Liu & Li, 2015; Shi, 2016). In general, Chinese parents and teachers hold conservative values toward adolescent dating because they believe that adolescent romance leads to various adverse outcomes, particularly poor academic performance. Here we analyzed whether puppy love negatively affects adolescents' cognitive and noncognitive development in China. Thus far, numerous studies have focused on adolescent romantic relationships in China (for reviews, see Liu & Li, 2015; Shi, 2016). However, some relevant major research gaps and questions warrant further research. First, most studies on the consequences of adolescent romance thus far have been limited by their reliance on speculative discussion or case analysis. Although few studies have quantitatively analyzed the role of romantic relationship in adolescent development, they have used data that are not nationally representative (Chen et al., 2009; Li et al., 2010; Li et al., 2019; Zhou et al., 2000). Moreover, they have failed to examine the mechanisms linking adolescent dating to individual development. Second, few studies have used statistical analyses to explicitly resolve the endogeneity problem related to the adolescent dating-individual development association. Adolescents with dating behaviors form a specific population. Whether an adolescent engages in romantic relationships depends on his/her individual and social characteristics (e.g., age, sex, and parental marital status). The systematic differences between adolescents with and without romantic experience can confound the causal effects of romantic relationships. In other words, to estimate the causal effects of adolescent romantic involvement, the underlying selection bias should be considered. In this study, we stated and tested a series of hypotheses regarding the effects of adolescent romantic relationships on individual cognitive and noncognitive skill development. This study aims to expand the current knowledge in this area in several aspects. First, in contrast to previous studies, we evaluated the effect of adolescent romance quantitively by using nationally representative data along with a selection bias mitigation method (i.e., propensity score weighting). We also distinguished light and heavy involvement in romance among adolescents and thus presented a more detailed depiction of the consequences of romantic involvement on adolescent development than the previous studies have provided. In our analyses, we explored the role of study time in mediating the effects of romantic involvement on cognitive and noncognitive skills. Finally, we explored the roles of social context (i.e., parents', teachers', and peers' attitudes toward adolescent romantic relationships) in conditioning the relationship of romantic involvement with cognitive and noncognitive skills in adolescents. To the best of our knowledge, this is the first study to systematically estimate the causal effects of romantic relationship on cognitive and noncognitive skill adolescent development. In this study, we analyzed data from China Education Panel Survey (CEPS)-which is an ongoing, large-scale, and nationally representative longitudinal survey, conducted by National Survey Research Center (NSRC) at Renmin University of China. Currently, two waves of CEPS data are publicly available: baseline and Year 1 follow-up. In the baseline survey during 2013-2014, the CEPS interviewed two cohorts: grade 7 and 9 students. Follow-up survey during 2014-2015 tracked all the original grade 7 students in the baseline survey. We restricted our data analysis to the grade 8 students in the Year 1 follow-up survey because the CEPS did not collect information on adolescent romance in the baseline survey. After exclusion of cases with invalid or incomplete data, our analytical sample comprised 4,400 boys and 4,219 girls, of which 605 and 402 had a romantic experience, respectively. The adolescents self-reported their relationship status (ever dated, or never dated). Those with a romantic experience were further classified into two groups: those engaged in only light sexual activities (i.e., confined to holding hands) and those engaged in heavy sexual activities (i.e., kissing, other more intimate physical contacts, or both). Cognitive skills were analyzed on the basis of the scores on the cognitive ability test as well as those on Chinese language, English language, and math midterm tests, all of which were standardized within the class. Noncognitive skills were assessed on the basis of self-efficiency, depression, confidence in the future, educational expectation, and learning attitudes. To estimate the causal effects of romantic involvement in adolescents, we used propensity score weighting, controlling for selection assignment differences between adolescents with and without a romantic experience. With this approach, all adolescents in this study could be used for outcome evaluation. To explore the mediating roles of study time, we used the Karlson-Holm-Breen (KHB) method to decompose the total effect of romantic involvement into direct effects and indirect (i.e., mediating) effects (Breen et al., 2013). To explore the influence of parents'/teachers' and peers' attitudes in conditioning the relationship between romantic involvement and development outcomes, we employed multivariate statistical models with interactions between their attitudes and romantic involvement. A key finding of this research was that in China, the influence of adolescent romantic involvement is generally negative. Specifically, compared with adolescents with no romantic experience, those with at least some romantic experience had lower levels of cognitive and noncognitive skills. As adolescents become intimate and committed in their relationships, they may begin to experience the related adverse effects. These findings are consistent with evidence from studies on adolescents in Western societies (Connolly & Johnson, 1996; Furman & Collibee, 2014; Furman et al., 2009; Giordano et al., 2008; Joyner & Udry, 2000; Neemann et al., 1995; Orpinas et al., 2013; Schmidt & Lockwood, 2017; Zimmer-Gembeck et al., 2001). Moreover, regarding the association of romantic involvement with cognitive stills, study time is a crucial mediator in adolescents, through which romantic relationships undermine their cognitive skills. Regarding the process of contextual influences, parents' or teachers' disapproval can increase the detrimental effects of romantic involvement, particularly in girls, probably because it leads to additional stress in dating adolescents. In general, our findings somewhat support developmental task theory (Roisman et al., 2009): romantic relationship is not a salient task during adolescence. Romantic involvement might distract adolescents from their current essential tasks, such as schoolwork. Our findings also support social timetable theory (Elder, 1995; Rook et al., 1989), which highlights the timing of major life events. In China, dating is considered an activity performed during adulthood rather than during adolescence. Therefore, parents and teachers interfere with adolescents' involvement in romantic relationships. Finally, our results indicate that the negative effects of romantic involvement are generally more pronounced among girls than among boys, particularly in terms of academic performance. A possible reason for this result is that in a relationship, girls are relatively dependent and empathy for others, whereas boys tend to be independent and focus on satisfaction (Seiffge & Krenke, 2011; Wang et al., 2021). Another possible reason is that when confronting interpersonal stresses, girls are more likely to avoid coping strategies than boys are (Grant et al., 2006; Seiffge-Krenke & Klessinger, 2000). In girls, higher sensitivity to relationship stress, combined with their avoidance coping style, can reduce their sense of control and potentially threaten their well-being and development.