Individual, Familial, and School Risk Factors Affecting Teen Dating Violence in Early Adolescents: A Longitudinal Path Analysis Model

Round 1

Reviewer 1 Report

Page 2, lines 47-50- this is a strange transition.  I am not seeing how the previous sentence demonstrates the scarcity of studies on TDV.  The last sentences show how varied the data is.

Overall I think this study is interesting, but I would suggest reworking the paper and dropping out the parts on TDV.  I am concerned about a few things:

1. The age of your participants is quite young for a study on teen dating violence.  What does it mean to an 11-year-old to be dating?  What does that even look like?  Do children this young even know what teen dating violence is to answer honestly on a self-report measure?

2. What was your operational definition for teen dating- did you provide one to the participants, or did they self-generate a definition?  Again, what does dating mean to an 11-year-old?

3. The CADRI was validated on older adolescents- is this an appropriate measure for kids of this age? 

Do any of the studies you reference talk about witnessing and normalizing TDV in peer groups are an associated variable?  This has been found in past research as a highly associated variable. However, this is also found in teens and not children.  I would argue that you have a blend of children and young teens in your study.

I recommend reworking the paper and dropping out the piece on teen dating violence since your sample is not solely made of teens.  If you elect to keep the TDV in, you should explain your use of the CADRI, what dating is like for children, and their definition of dating.  

The English is fine.

Author Response

Replies Reviewer 1

Q1: Page 2, lines 47-50- this is a strange transition.  I am not seeing how the previous sentence demonstrates the scarcity of studies on TDV.  The last sentences show how varied the data is.

Authors’ reply: We thank the reviewer for his/her suggestions. Accordingly, with his/her suggestion we rearranged the sentence (lines 56-60).

Overall I think this study is interesting, but I would suggest reworking the paper and dropping out the parts on TDV.

Authors’ reply: We thank the reviewer for his/her positive comments about our manuscript and for his/her valuable suggestions.

I am concerned about a few things:

Q2: The age of your participants is quite young for a study on teen dating violence.  What does it mean to an 11-year-old to be dating?  What does that even look like?  Do children this young even know what teen dating violence is to answer honestly on a self-report measure?

Authors’ reply: We thank the reviewer for his/her comment, which helped us to better clarify the participants’ age issue. In this regard, very few studies investigated the prevalence and patterns of TDV among the so-called pre-adolescents. Among these, all agreed about the need to deepen and explore the developmental trajectories of such partners of aggressive behaviors, in particular considering that early adolescents that are just beginning to be involved in romantic relationships showed higher rates of TDV involvement both as perpetrators and victims (Viejo et al.,2020; Goncy et al., 2017; Taylor et al., 2017) (lines 62-68). That said, we are sensitive to this critical issue. In the discussion section (lines 484-487), we have added that future research should replicate our model by adopting a long-term longitudinal study also involving adolescent participants. We have further clarified our sample age by changing the manuscript title to ‘Individual, familial and school risk factors affecting Teen Da-ting Violence in early adolescence: A longitudinal path analysis model’.

Q3: What was your operational definition for teen dating- did you provide one to the participants, or did they self-generate a definition?  Again, what does dating mean to an 11-year-old?

Authors’ reply:  We did not provide any definition of TDV. Still, in line with the CADRI instruction, we asked only participants in dating situations to rate if they acted and/or suffered some behaviors with his/her boyfriend/girlfriend while arguing. Then, we operationalized TDV following the CADRI scoring indication.

Q4: The CADRI was validated on older adolescents- is this an appropriate measure for kids of this age? 

Authors’ reply: As far as we know, the CADRI has been validated for early adolescents by Stroever (2019). Furthermore, other authors adopted the CADRI in their studies involving early adolescents (Taylor et al., 2017; Niolon et al., 2015).

Q5: Do any of the studies you reference talk about witnessing and normalizing TDV in peer groups are an associated variable?  This has been found in past research as a highly associated variable. However, this is also found in teens and not children.  I would argue that you have a blend of children and young teens in your study.

Authors’ reply: We thank the reviewer. As reported in previous replies we have now stressed across the manuscript that our sample included early adolescents (10-14 years old).

Q6: I recommend reworking the paper and dropping out the piece on teen dating violence since your sample is not solely made of teens.  If you elect to keep the TDV in, you should explain your use of the CADRI, what dating is like for children, and their definition of dating.  

Authors’ reply:  We thank the reviewer for bringing attention to such an important issue as the age of the participants. We decided to keep the TDV, considering that this variable was central in our path model when we designed the study. However, because we are sensitive to the reviewer's comment, we made it clear since the title that our study is aimed at early adolescence. We have also reported studies that used an age sample similar to ours with some of our predictor variables. We believe that an intervention in the early adolescence stages may allow for a more effective preventive strategy. Nevertheless, it is desirable that future studies could replicate our model by considering adolescents, and we have added this to the Discussion section.

Reviewer 2 Report

Comment to the Authors:

The reviewer carefully assessed the submitted paper titled "Individual, Familial, and School Risk Factors Affecting Teen Dating Violence: A Longitudinal Path Analysis Model" and found it to be of great interest. The paper delved into the causal relationship between various risk factors—individual, familial, and school-related—and teen dating violence. The analysis was based on short-term longitudinal data collected in Italy. Notably, the paper focused on exploring causal links between witnessing intimate partner violence and teen dating violence. Although the authors did not directly observe significant effects of witnessing intimate partner violence on teen dating violence, they effectively elucidated the indirect influence through school bullying and victimization.

The reviewer recognizes that the insights revealed by the authors hold valuable implications for addressing the complex issue of teen dating violence and its mitigation. Consequently, this contribution could enrich the body of literature within the realm of teen dating violence studies. However, the reviewer also harbors certain concerns, predominantly pertaining to the theoretical framework and methodologies employed by the authors. Addressing these concerns adequately is essential for the reviewer's recommendation of the paper's publication.

Firstly, the reviewer raises a query regarding the appropriateness of the data employed in the study, which is based on a short-term longitudinal survey. The mere six-month interval between T1 and T2 prompts skepticism about the data's ability to accurately capture causal relationships between witnessing intimate partner violence and subsequent teen dating violence. There exists a possibility that instances of teen dating violence observed at T2 may have already manifested at T1. Should this conjecture hold true, the reviewer contends that the authors cannot assert a causal connection between witnessing intimate partner violence at T1 and teen dating violence at T2. To mitigate this concern, it is imperative for the authors to account for potential autocorrelations by controlling for teen dating violence at both T1 and T2.

Secondly, the reviewer posits that the theoretical foundation of the submitted paper appears underdeveloped. While acknowledging the importance of investigating the interplay between individual, familial, and school-related risk factors influencing teen dating violence, the reviewer deems the diagram presented in the Introduction section (Figure 1) to be overly exhaustive and lacking in theoretical precision. Consequently, the interpretation of the analytical results depicted in Figure 2 seems somewhat disjointed and lacking in a cohesive theoretical framework. As a consequence, the principal findings of the authors' study remain ambiguous to the readers. Enhancing the theoretical underpinning of the study is crucial, and the reviewer urges the authors to articulate specific hypotheses that align with theoretical explanations concerning the interrelationships among individual, familial, and school-based risk factors influencing teen dating violence.

Thirdly, the reviewer highlights an omission in the authors' analysis—the absence of gender as a controlled factor when assessing the effects of witnessing intimate partner violence, school bullying (or victimization), cyberbullying (or cybervictimization), violence against teachers, and deviant behavior on teen dating violence. Anticipating potential gender-based discrepancies in the impact of these factors on teen dating violence, the reviewer emphasizes the need for the authors to incorporate gender as a controlled variable. At the very least, the reviewer recommends that the authors analyze the interplay of individual, familial, and school-related risk factors on teen dating violence stratified by gender and subsequently examine gender-based differences in these effects based on the outcomes.

Lastly, the reviewer conveys an issue concerning the resolution of Figures 1 and 2, which are crucial to comprehending the results derived from the authors' analyses. Due to the low resolution of these figures, the reviewer encounters difficulty in discerning the finer details. In light of the significant information encapsulated within Figures 1 and 2, the reviewer urges the authors to present these figures in a higher resolution to ensure clarity and facilitate readers' understanding.

The reviewer extends gratitude for the opportunity to evaluate the submitted paper. It is the reviewer's earnest hope that these comments and suggestions contribute positively to the refinement of the paper's quality and impact.

Author Response

Replies to Reviewer 2

Q1: The reviewer carefully assessed the submitted paper titled "Individual, Familial, and School Risk Factors Affecting Teen Dating Violence: A Longitudinal Path Analysis Model" and found it to be of great interest. The paper delved into the causal relationship between various risk factors—individual, familial, and school-related—and teen dating violence. The analysis was based on short-term longitudinal data collected in Italy. Notably, the paper focused on exploring causal links between witnessing intimate partner violence and teen dating violence. Although the authors did not directly observe significant effects of witnessing intimate partner violence on teen dating violence, they effectively elucidated the indirect influence through school bullying and victimization. 

The reviewer recognizes that the insights revealed by the authors hold valuable implications for addressing the complex issue of teen dating violence and its mitigation. Consequently, this contribution could enrich the body of literature within the realm of teen dating violence studies. However, the reviewer also harbors certain concerns, predominantly pertaining to the theoretical framework and methodologies employed by the authors. Addressing these concerns adequately is essential for the reviewer's recommendation of the paper's publication.

Authors’ reply: We thank the reviewer for her/his positive feedback on our manuscript. We have integrated his/her suggestions and run additional analyses. The reviewer's comments have been fundamental in improving the manuscript. We hope that the reviewer will appreciate this revised version. We thank the reviewer again for his/her valuable feedback on our manuscript.

Q2: Firstly, the reviewer raises a query regarding the appropriateness of the data employed in the study, which is based on a short-term longitudinal survey. The mere six-month interval between T1 and T2 prompts skepticism about the data's ability to accurately capture causal relationships between witnessing intimate partner violence and subsequent teen dating violence. There exists a possibility that instances of teen dating violence observed at T2 may have already manifested at T1. Should this conjecture hold true, the reviewer contends that the authors cannot assert a causal connection between witnessing intimate partner violence at T1 and teen dating violence at T2. To mitigate this concern, it is imperative for the authors to account for potential autocorrelations by controlling for teen dating violence at both T1 and T2.

 Authors’ reply: We understand the reviewer's concerns and performed additional analyses to control for TDV even at time 1. First, we examined the correlation between the same form of adolescent attendance measured at T1 and T2. Since violence is usually a persistent phenomenon over time and, unfortunately, is never an isolated phenomenon, it is not surprising that the same forms of TDV correlate with each other at T1 and T2. However, this analysis showed that the maximum shared variance between the same constructs at different times was 28%, suggesting that the two measurements were (at least partially) independent.

*p = .01 ** p < .01 *** p < .001

Considering that a significant correlation emerged between TDV at T1 and T2 anyway, following the reviewer's suggestion, we performed a series of Durbin-Watson tests between TDV time series, checking for potential autocorrelations. The Durbin-Watson tests showed the absence of autocorrelations with all values between the cut-off of 2 and 2.5 (DW Verbal Emotional TDV Acted: 2.4; DW Physical TDV Acted: 2.22, DW Verbal-Emotion TDV Suffered: 2.03, DW Physical TDV Suffered: 2.04). We have added the Durbin-Watson tests to the manuscript.

Although these analyses seem encouraging regarding our model's ability to predict TDV at T2 net TDV at T1, we decided to perform a new control path analysis by including the difference between TDV at T1 and TDV at T2 as the dependent variable in the model. This analysis allowed us to examine the contribution of our model in predicting an increase or decrease in TDV, overcoming any problems associated with the initial and final levels of TDV at T1 and T2. The pathway model with the difference in TDV perfectly replicated our results with only TDV at T2 (the difference was only of one or two decimals in beta), suggesting that, even considering a potentially high level of TDV at T1, our model was able to explain a further increase (or decrease) in TDV at T2. While the set of analyses mentioned above seems to encourage us about the capability of our model in predicting TDV at T2, controlling for TDV levels at T1, we were very sensitive to the reviewers’ comments. For this reason, in the Discussion section, we added that the short-term interval of measurements between T1 and T2 was a limitation of our study. Future studies can replicate our results with a long-term design.

Q3: Secondly, the reviewer posits that the theoretical foundation of the submitted paper appears underdeveloped. While acknowledging the importance of investigating the interplay between individual, familial, and school-related risk factors influencing teen dating violence, the reviewer deems the diagram presented in the Introduction section (Figure 1) to be overly exhaustive and lacking in theoretical precision. Consequently, the interpretation of the analytical results depicted in Figure 2 seems somewhat disjointed and lacking in a cohesive theoretical framework. As a consequence, the principal findings of the authors' study remain ambiguous to the readers. Enhancing the theoretical underpinning of the study is crucial, and the reviewer urges the authors to articulate specific hypotheses that align with theoretical explanations concerning the interrelationships among individual, familial, and school-based risk factors influencing teen dating violence.

Authors’ reply: We thank the reviewer for these suggestions. Accordingly, we have revised our hypothesis to align with theoretical explanations concerning the interrelationships among individual, familial, and school-based risk factors influencing teen dating violence (lines 111 – 151).

Q4: Thirdly, the reviewer highlights an omission in the authors' analysis—the absence of gender as a controlled factor when assessing the effects of witnessing intimate partner violence, school bullying (or victimization), cyberbullying (or cybervictimization), violence against teachers, and deviant behavior on teen dating violence. Anticipating potential gender-based discrepancies in the impact of these factors on teen dating violence, the reviewer emphasizes the need for the authors to incorporate gender as a controlled variable. At the very least, the reviewer recommends that the authors analyze the interplay of individual, familial, and school-related risk factors on teen dating violence stratified by gender and subsequently examine gender-based differences in these effects based on the outcomes.

Authors’ reply: We thank the reviewer for this important comment, which allowed us to run additional analyses about gender differences. As hypothesized by the reviewers, the multi-group path model stratified for gender showed differences between males and females. We have added a new section in the manuscript about gender analyses (lines 337 -395). We have further discussed gender differential paths in the Discussion section (lines 445-467). 

Q5: Lastly, the reviewer conveys an issue concerning the resolution of Figures 1 and 2, which are crucial to comprehending the results derived from the authors' analyses. Due to the low resolution of these figures, the reviewer encounters difficulty in discerning the finer details. In light of the significant information encapsulated within Figures 1 and 2, the reviewer urges the authors to present these figures in a higher resolution to ensure clarity and facilitate readers' understanding.

Authors’ reply: We have revised both figures increasing resolution.

The reviewer extends gratitude for the opportunity to evaluate the submitted paper. It is the reviewer's earnest hope that these comments and suggestions contribute positively to the refinement of the paper's quality and impact.

Authors’ reply: We thank the reviewer again for this comment. We hope that the reviewer will appreciate this revised version. We thank the reviewer again for his/her valuable feedback on our manuscript.

Round 2

Reviewer 1 Report

I am pleased with the changes that were made and support your publication.

Reviewer 2 Report

The reviewer is grateful for the opportunity to review the submitted paper and is glad that his comments and suggestions could contribute to improving it. He will recommend it for publication.