The use of the Internet has grown rapidly in the past two decades, especially amongst the adolescent population. As the Internet becomes pervasive and increasingly indispensable in young individuals' lives, Internet addiction (IA) has also emerged as a serious public health issue due to its close relationships with adolescent health problems, risk behaviors, and social functioning problems (1–3).

Preventing IA requires efforts taken in various sectors such as family, school, and other social institutions. Family factors, such as positive family functioning, parental monitoring, and healthy parental–child relationship, play a vital role in this process (4–6). Among different parenting characteristics, parental control was intimately related to problematic adolescent behaviors [e.g., (7, 8)].

Parental control includes behavioral and psychological control (9). Behavioral control pertains to the use of disciplinary strategies and supervisory functions to regulate children's behavior, while psychological control attempts to shape children's behaviors through strategies like guilt or anxiety induction, and love withdrawal (10). Research findings have consistently shown that the behavioral control of parents is positively related to favorable developmental outcomes in children, while psychological control harms adolescent health (7, 8, 11). As Barber et al. (11) concluded, higher parental behavioral control is related to higher levels of adolescent competence, self-discipline, and school performance as well as lower levels of problematic behaviors. When parents provide clear regulations regarding children's Internet use or monitor their Internet use in an appropriate way, adolescents have a lower chance to show symptoms of IA. In contrast, parental psychological control tends to harm children's self-esteem, and increase developmental problems and maladjustment, because psychological control hurts adolescents' emotional functioning and the sense of self (12). For example, in a sample of 5,806 seventh graders, Wang et al. (8) found that the psychological control of parents predicted students' dampened emotional functioning in China and the United States.

There are three issues that should be addressed when examining the influence of parents on adolescent developmental outcomes. First, while parental control has been identified as an important determinant of adolescent development, the impact of the parent–child relationship has not been adequately examined. According to Shek (13), adolescents' satisfaction with their parents' control and their willingness to communicate with their parents are important relational qualities that should be considered when examining adolescent adjustment. Primarily, children's satisfaction with their parents' control reflects the quality of the parent-child relationship. Whether adolescents themselves regard parental control as reasonable or not is an important factor to be considered when understanding the influence of parental behavioral and psychological control. Pomerantz and Eaton (14) argued that some adolescents might regard parental control as an expression of care. Hence, these adolescents may gain more positive influence from parental control. This argument was supported by Kakihara (15), who found that adolescents interpreted strong parental behavioral control in specific domains as indicative of competence or intrusiveness. Thus, Kakihara drew attention to the possibility of multiple yet “contradictory” interpretations of parenting reported by adolescents. Hence, it is important to consider the quality of parent–child relationship when examining the impact of parental control.

Besides, parent–adolescent communication also influences adolescent development. Research showed that healthy parent–adolescent communication provides a safe environment for adolescents to comfortably disclose themselves (16), while problematic parent–adolescent communication is often associated with increased adolescent risk behaviors (17). Forehand et al. (18) found that families of adolescents with behavior problems experienced disagreements more frequently and had less supportive parent–child relationships than their counterparts did. Cottrell et al. (19) revealed that open parent–child communication was positively associated with parental monitoring. However, as Burk and Laursen (20) pointed out, little is known about how specific relationship attributes contribute to specific developmental problems such as IA.

The second issue concerns the differential influence of fathers and mothers on adolescent development. Historically, there has been an absence of parenting research examining the relationships between adolescents and their fathers (21). In addition, instead of separating paternal and maternal impacts, many studies just considered overall parenting characteristics (i.e., adolescent perception of their parents), which hinders the understanding of the complex dynamics in this process (22). Some recent studies have recognized the distinction between maternal and paternal factors on adolescent developmental outcomes, but the majority adopted a single perspective focusing on either mother' or fathers' influence. For example, Leung et al. (23) explored maternal control in a sample of 432 poor Chinese single-mother families. Studies on paternal involvement were reviewed in Pleck and Masciadrelli's (24) research, which showed that paternal involvement was related to positive child development. However, to have a holistic picture of how family factors systematically influence adolescent IA, it is important to include both maternal and paternal factors. Hence, it is argued that when examining parental impacts, both maternal and paternal factors should be taken into account.

However, the existing research findings regarding the differential effects of fathers and mothers are equivocal. For example, Giles and Price (25) found that only maternal psychological control positively predicted problematic computer use in children. In contrast, Lansford et al. (26) revealed that only paternal psychological control accounted for specific variance in children's developmental problems. The authors reported a similar result on the unique predictive effect of paternal knowledge on boys' externalizing problems. In Xu's research (2) involving 5,122 Chinese adolescents, results showed that compared to the father–child relationship, the mother–child relationship had a stronger association with children's IA. While these studies illustrated the differences between maternal and paternal impacts, the findings were inconclusive. Furthermore, they generally focused on a single parenting factor (e.g., psychological control) rather than simultaneously considering different aspects of the parent–child subsystem.

The third issue pertains to research design, in that most of the existing findings in this field are based on cross-sectional studies. Few studies have used longitudinal data to examine the growth rate of IA related to different family factors (4, 6). For example, Wang et al. (17) research using three-year longitudinal data on 913 Bahamian students revealed that parental control during early adolescence predicted a decrease in risk behavior in middle adolescence. Yu and Shek's (27) longitudinal research on students in Hong Kong revealed that good family functioning predicted a lower probability of having IA. Regarding the impacts of parental factors on the growth rate of IA, Shek et al. (28) study involving 3,328 Chinese students indicated that stronger paternal behavioral control was associated with a slower drop in children's IA, and stronger maternal psychological control was linked to a faster decrease in children's IA. Although their findings filled the knowledge gaps in the field, the study only covered early adolescence. More efforts should be made to examine the differences between maternal and paternal parenting styles among late adolescents (29).

In short, longitudinal research focusing on senior high school students' IA is urgently needed to understand how the parent–child relationship qualities shape adolescent IA. First, the dynamics of parenting and the family relationship functioning of senior high school students of IA are different from that of junior students. Students in late adolescence are physically and psychologically mature, demanding more freedom and flexibility. Parental control may not lead to the desired effect to adolescents at this stage. For example, Rogers et al. (30) suggested a stronger association between parental psychological control and internalizing problems for senior high school students than for junior students, because older adolescents might have stronger need for autonomy and more diverse internalizing and externalizing behaviors. Second, as the patterns in which mother- and father-related factors influence adolescents may evolve differently as both parents and adolescents age, longitudinal research would enhance our understanding of the differences between the influence of maternal and paternal factors on adolescent IA.

To fill these three research gaps, the current study investigated how parent–child subsystem factors influence the levels and rates of change in adolescent IA over senior high school years. Specifically, the study addressed the following three research questions. The first question is as follows: “do parental control (behavioral and psychological) and father– and mother–child relationship qualities predict children's initial level of IA at the beginning of senior high school life?” Drawing from extant literature (7, 8, 11) and related findings in early adolescence (28), it was expected that higher paternal and maternal behavioral control would predict a lower initial level of adolescent IA in senior high school years (Hypotheses 1a and 1b), while higher paternal and maternal psychological control would be associated with a higher initial level of adolescent IA (Hypotheses 1c and 1d). Regarding the parent–child relationship, we hypothesized that better relationships between the child and both parents would predict a lower initial level of IA (Hypotheses 1e and 1f).

The second research question is as follows: “how are parent–child subsystem qualities related to the developmental trajectory of adolescent IA across senior high school years?” In this field, only one previous study found that paternal behavioral control, maternal psychological control, as well as relationships between the child and both parents exerted significant predictive effects on the rate of change in IA during junior high school years (28). However, the related findings were at odds with the general expectations that favorable parental factors would be associated with a faster drop in IA as an indicator of positive adjustment in the long run. Given that only limited evidence is available, we still formed the present hypotheses based on the general theoretical expectations that positive parenting and a better parent–child relationship would be related to children's positive development (31). Specifically, it was expected that higher behavioral control of fathers (Hypothesis 2a) and mothers (Hypothesis 2b), as well as better father–child (Hypothesis 2c) and mother–child relationships (Hypothesis 2d), would predict a faster decline in children's IA because these factors were considered as positive aspects of parental impacts. Additionally, as a form of negative parenting, higher paternal and maternal psychological control were expected to predict a slower decrease in adolescent IA (Hypotheses 2e and 2f, respectively).

The third question is as follows: “what is the concurrent and longitudinal influence of paternal and maternal factors on adolescent IA during senior high school years?” While there is support for the relatively stronger influence of paternal parenting (26), other findings suggest a stronger influence for maternal parenting (2, 25). Besides the inconclusive picture, very few studies have distinguished different aspects of parenting practices (e.g., behavioral and psychological control and the relationship between parents and children) and examined the long-term impact using longitudinal data (28). Due to these factors, the following two general competing hypotheses were advanced: fathers are more influential than mothers in influencing adolescent IA (Hypothesis 3a), and mothers are more influential than fathers in influencing adolescent IA (Hypothesis 3b). All the above hypotheses are summarized in Table 1.

The present study also considered important demographic characteristics including student gender, family economic status, and family intactness. Specifically, boys often report a higher level of Internet use and a higher interest in online games than girls do (4). Some studies found that parenting effects vary among boys and girls. For example, Shek (32) examined parenting functions among Chinese adolescents with disadvantageous background and found that paternal parenting exerted more influence on boys' mental health and problem behavior, while maternal parenting played a major role in affecting girls' mental health and problem behavior. In contrast, Rogers et al. (30) reported that fathers' psychological control was more influential on daughters than on sons regarding externalizing behavior. Shi et al. (4) examined the relation between family functioning and children's IA and found that the path from family function to IA (via loneliness) was significant only for girls. By including student gender, this study attempted to contribute to the ongoing discussion about differential parenting influence on IA amongst boys and girls.

The present study examined parental influence on the development of Internet Addiction (IA) among Chinese adolescents. Three research questions were focused upon in this study. First, by separating paternal and maternal factors, our study examined the impacts of multiple parenting factors and quality of the relationship between parents and children on the initial level of children's IA. Second, the predictive effects of parental factors on the growth rate of children's IA were also investigated. Third, we also examined and compared the influence of different parental factors on adolescent IA at a single time point and over time. While a recent study has addressed these questions in early adolescents (28), the present study involved late adolescents in senior high school. In addition, gender-based analyses were conducted to further explore the effects of children's gender.

While most of the present findings are similar to the observations reported on early adolescents, some are different from these previous findings (28). For the first research question, male adolescents showed a higher initial level of IA than did females, which is in line with previous findings (28, 43). In addition, we observed a decline trend in adolescent IA. It is plausible that students in the higher grade will be less devoted to activities related to the Internet as they must prepare for the university entrance examination. Additionally, the magnitude of decline in IA was greater among male adolescents than that among female students. This gender difference may be attributed to female adolescents' relatively low level of initial IA, which was unlikely to decline substantially (i.e., floor effect). Furthermore, we did not observe a significant effect regarding children's gender and parental impacts on IA levels, which is inconsistent with the previous findings that parental psychological control was more influential on girls' than on boys' externalizing behaviors (26, 30). Nevertheless, as no previous research has tested gender effect regarding change in IA over time, more research is needed to portray a conclusive picture.

Regarding the parental influence on the initial level of IA that was addressed by the first research question, four hypotheses (1a, 1d, 1e, and 1f) were supported, whereas Hypotheses 1b and 1c were not. Specifically, when each aspect of the quality of the parent–child subsystem was examined separately in individual growth curve (IGC) models based on the full sample, higher paternal but not maternal behavioral control, and better relationships with both parents, were significantly associated with lower levels of initial adolescent IA. Additionally, higher maternal but not paternal psychological control predicted a higher level of initial adolescent IA. Gender-based IGC models showed similar findings for male and female adolescents. On the one hand, these findings suggested that parents' behavioral control and better parent-child relationships are positive parenting factors, whereas psychological control is a negative factor, regardless of children's gender. This conclusion echoes the previous findings observed in both Chinese and Western contexts (7, 28, 44).

On the other hand, the observations shed light on the differential influence of fathers' and mothers' factors. Specifically, paternal parenting had a closer association with adolescent IA via behavioral control, while mothers exerted stronger influence through psychological control. These findings seem to be different from the results reported in Shek et al.'s (28) research, which found that paternal and maternal factors tend to function in the same way in predicting the initial level of adolescent IA in early adolescence. However, as indicated by the estimated coefficients of predictors in Shek et al.'s (28) research, paternal behavioral control in comparison to maternal behavioral control, and maternal psychological control in comparison to paternal psychological control, tended to have a stronger association with the initial level of adolescent IA. Taken together, these findings suggest that fathers and mothers might contribute to the development of adolescent IA through different parental behaviors at different stages of adolescence. Such an assumption can help explain the existing mixed findings on paternal vs. maternal impacts on adolescent development. This will be discussed with reference to regression analyses below.

For the second research question, all the hypotheses based on general theoretical models were not supported. However, the findings are consistent with the previous observations among early adolescents (28). It was found that higher paternal behavioral control and better mother–child relationship predicted a slower decline in adolescent IA during senior high school years. Meanwhile, higher psychological control was linked with a faster decrease in adolescent IA. In short, these findings suggest that—similar to the situation in early adolescence—parental impacts also gradually diminish over time during late adolescence. These observations coincide with the decreasing amount of variance in adolescent IA explained by parental factors, as revealed by regression analyses. For example, the percentage of adolescent IA accounted by all concurrent parental factors decreased from 4.4% at Wave 4 to 2.9% at Wave 6. As suggested by Shek et al. (28), this finding may be attributable to adolescents' decreasing dependence on parents and increasing devotion to other types of social relationships (e.g., peer relationships). Additionally, at later stages of senior high school, all students in Hong Kong must invest most of their time and energy into preparing for the public examination. Thus, adolescents would generally be less different from each other in terms of their IA behaviors. Furthermore, those with IA problems may have dropped out from the study. Therefore, parental influence on adolescent IA may become less significant. Nevertheless, as the present study is a pioneering attempt, these findings need to be confirmed in future replication studies.

Regarding the contribution of fathers and mothers to children's IA levels (i.e., the third research question), the present study revealed that paternal factors accounted for a higher percentage of adolescent IA than did maternal factors, both concurrently and longitudinally. The general greater paternal impact vs. maternal impact on children's IA is in line with previous studies involving different developmental outcome measures. For example, using global treatment style, responsiveness, and demandingness to index parenting qualities, Shek (45, 46) reported that, relative to maternal parenting qualities, paternal parenting qualities had a greater impact on Chinese adolescent morbidities such as hopelessness and psychological well-being indexed by self-esteem, purpose of life, and life satisfaction. One recent study also identified that paternal, but not maternal, expectations regarding children's future significantly affected children's well-being (47). Generally speaking, fathers are less involved in parenting tasks than mothers are, especially in Chinese society (48). In the present study, fathers were less controlling and had a poorer relationship with the children as compared to mothers. However, as fathers occupy a superior and controlling role in Chinese families, fathers may be more influential in their children's developmental outcomes although they are less involved and more detached. This interpretation is consistent with other findings in early adolescents (49).

Different from the present study, Shek et al. (28) found that paternal and maternal factors had equal impacts on children's IA levels. One possible explanation is that the age of the adolescents may have served as a moderator. Specifically, Shek et al.'s (28) study involved early adolescents (mean age = 12.59 years old) who were younger as compared to the present participants. A recent study in Korea supported the potential moderating effect of children's age on parental impacts (50). The study showed that, while parental affection negatively predicted children's problematic mobile game use in the elementary school group, parental monitoring lead to less problematic mobile game use in the high school group. Future research could directly compare paternal impacts on adolescent IA with reference to different age groups to test the possibility that paternal factors might have generally greater impacts as compared to maternal factors among older adolescents.

Differential parental impacts should also be discussed with reference to specific aspects of parenting. As mentioned earlier, when each dimension of the parent–child subsystem was considered in the IGC models, paternal behavioral control but not maternal behavioral control, and maternal psychological control but not paternal psychological control, were significantly linked with the initial level of children's IA. Basically, these findings are consistent with the concurrent and longitudinal predictive effects derived from the regression analyses. When paternal and maternal factors were examined simultaneously in regressions, paternal behavioral control and maternal psychological control were the two most robust predictors of adolescent IA, both concurrently and longitudinally. The consistent findings obtained from different analytical approaches informed the reliability of these findings.

Based on the findings, it can be conjectured that the impact of paternal factors is greater in terms of behavioral control, while maternal parenting is more influential in terms of psychological control over time. In previous studies, while some findings suggested stronger paternal impacts on adolescent development, some others supported stronger maternal impacts. One possible explanation for the mixed findings is that most of these studies focused on overall parenting without considering different dimensions of parenting or only referred to one dimension of parenting. For example, in studies that identified greater impacts of maternal psychological control on adolescent development outcomes, behavioral control or other parental factors were not examined simultaneously (25, 51). Chen et al. (52) identified stronger influence of fathers' indulgence, but they did not take psychological control into account. In fact, the results in a few studies that considered different domain-specific parental factors showed that both parents' influence differed from each other on specific aspects. For example, while maternal control showed a greater impact on children's achievement motivation and positive youth development than did paternal control, paternal sacrifice exerted a stronger influence than did maternal sacrifice on adolescents' positive youth development (53, 54). Thus, our findings suggest that it is important to distinguish between different processes of parenting in examining and comparing paternal and maternal influences.

There are two additional observations which deserve attention. First, the concurrent negative impacts of both parents' psychological control on adolescent IA are consistent with findings on early adolescence (28). Researchers advocated that parental psychological control would ruin children's age-appropriate sense of autonomy, which in turn results in adolescent misbehaviors and damaged psychological well-being. However, most of the findings were derived from samples of adolescents in Western cultures, which value autonomy and independence. Some scholars argued that in Asian countries, including China, parental psychological control may not be so detrimental as cultural norms in these areas emphasize children's obligations to the family and interdependence rather than independence (8, 55). Nevertheless, recent studies showed that in Chinese adolescents, parents' psychological control was concurrently associated with children's maladjustment, such as depression and anxiety as well as hopelessness (56). Thus, the present study adds to the extant literature by showing that the negative impacts of parents' psychological controlling behavior may be universal.

Second, the present longitudinal findings suggest that, during senior high school years, maternal psychological control might be more influential than paternal psychological control is. This observation is not consistent with the general view that negative paternal parenting is especially detrimental due to the greater authority and power of fathers in the family and the related attributes of the father–child dyad (30). For example, paternal psychological control was more closely associated with early adolescents' externalizing behaviors over time than maternal psychological control was (26, 30). Likewise, in early adolescence, paternal psychological control served as a more robust longitudinal predictor than maternal psychological control did for adolescent IA (28). Among university students, paternal, but not maternal, negative parenting in terms of denying and overprotectiveness resulted in a higher level of children's problematic Internet use (57). Nevertheless, it is possible that paternal and maternal negative parenting impact different adolescent developmental outcomes. For instance, Shek (58) found that paternal—but not maternal—psychological control showed a significant predictive effect on adolescents' life satisfaction in 1 year. Meanwhile, maternal—but not paternal—psychological control was a significant predictor of changes in adolescents' self-esteem. The present findings suggest the importance of examining this issue with reference to different measures of development outcomes among different age groups of adolescents.

The present study sheds light on the differential parental impacts on the concurrent and future levels and the growth rate of adolescents' IA during senior high school years. However, several limitations should be noted. First, the present study employed a quantitative design, which was unable to reveal the subjective feelings of participants and delineate potential mechanisms behind the quantitative findings. To portray a comprehensive picture and to triangulate the present findings, it is necessary to further investigate differential parental impacts using qualitative research strategies, such as focus group interview and case study.

Second, the utilized data were obtained only from adolescent participants' self-report. Although social desirability is a major shortcoming of self-report methodology, self-report data have commonly been collected in adolescent research, especially in longitudinal studies [e.g., (59)], possibly due to ethical and practical concerns. Besides, it is efficient and cost-effective, and it can be argued that adolescents themselves know their own experiences and lives much better than others. In the present study, to reduce social desirability bias as much as possible, anonymity was emphasized, and the participants were clearly instructed to provide answers based on their true perceptions without communicating with others. Nevertheless, future research could employ individual interviews in addition to self-report or involve different informants such as parents and teachers.

Third, the present study surveyed Chinese adolescents in a single region (i.e., Hong Kong). To verify and expand the generalizability of the present findings, future studies will benefit from investigating related questions among adolescents in other Chinese communities, such as those from mainland China, and adolescents from other ethnicities. Fourth, from Wave 4 to Wave 6, 899 participants withdrew from the study. In particular, more than 500 students withdrew from the study after Wave 4. This fact might affect the present findings. However, comparisons between the matched sample and dropouts suggested that there were no substantial differences in key variables between the two groups. Thus, the present findings were unlikely to be significantly affected by systematic attrition. Finally, as a control variable, family intactness was indicated by parental marital status. Since there were other factors related to family intactness, such as family members' residential status (e.g., whether children are living with their parents) (60), future studies could consider more comprehensive measures of family intactness.

The datasets generated for this study are available on request to the corresponding author.

DS designed the project and contributed to all steps of the work. XZ contributed to the development of the idea and data interpretation. XZ and DD drafted the work and revised it based on the critical comments and editing provided by DS. All authors approve of the final version of the manuscript and agree to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.