Perpetrators of the Criminal Victimisation of Children: A Longitudinal Study

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The article is clear and well-structured.
I would suggest making the passage about prenatal predictors (parental mental health) that correlate with outcomes more explicit, but without distinguishing the mechanisms (e.g., reduced surveillance), which could risk spuriousness despite the longitudinal design.
Furthermore, it would be helpful to clarify that the absence of mental health is not the factor that leads to victimisation. This avoids the risk of bias that could lead people to believe that those with a mental illness are also violent.
It would be helpful to add notes on missing data and multiple test corrections.

Author Response

socsci-3981828

Reviewer 1

1. More explicit reference to parental mental health.

We have added some sentences to this discussion acknowledging previous research reporting an association between parental mental illness and child experiences of behaviour problems and violence. We have extended our discussion of the evidence from previous studies that parental mental health is associated with a wide range of child mental health and antisocial behaviours.

 

2. Clarify whether the absence of mental health (presence of mental illness) leads to victimisation.

This is now addressed in more detail in the discussion (see above).

 

3. Add note on missing data and multiple test corrections.

Detail has been added – we use listwise deletion of missing data largely because some of the multivariate models involve many variables and we prioritise retaining as many cases as possible. We have noted in the limitations that we have not corrected for multiple test comparisons.

Reviewer 2 Report

Comments and Suggestions for Authors

The following points highlight areas that would need improvement to meet the journal's standards:

Abstract:

• Explicitly state in the abstract that the study is a secondary analysis of data from an existing independent longitudinal cohort (MUSP). Provide details about the MUSP study: year, sample, etc.

Analysis:

• Clearly describe how the multinomial logistic regression was structured. Specify the outcome categories and reference group (e.g., no victimization vs. victimization by specific perpetrator types) so that readers can understand what the odds ratios in Table 3 represent. Also state whether the odds ratios for each predictor are adjusted for other variables or from separate bivariate models.
• In Table 4, the number of victimization incidents is treated with a linear regression (standardized beta coefficients). Justify the use of a linear model for this count outcome or consider using a more appropriate count model (e.g., Poisson or negative binomial regression). If linear regression is retained, confirm that its assumptions (e.g., normality of residuals) are reasonably met for the count data.
• Ensure consistency and transparency in reporting results. For example, report confidence intervals and p-values for all key estimates (as done in Table 3) in the text as needed, and clarify how missing data were handled in the analyses (if any cases were dropped or imputed, specify this).

Interpretation:

• Avoid language that implies causality when interpreting the findings. For instance, use terms like “is associated with” rather than “predicts” to describe relationships between parental factors and child victimization. Emphasize that these are correlations in a longitudinal context, not proven causal effects.
• Provide interpretation for notable findings that are currently under-explained. For example, the result that children of parents with multiple past partners have higher victimization risk is stated but not discussed. Explaining potential reasons (e.g., family instability or exposure to more unrelated adults may increase risk) so the reader understands why this association might occur.
• When discussing results, note that factors related to the child or environment (beyond parental characteristics) could influence victimization. For instance, older children’s greater independence and peer interaction could contribute to their risk profile. Acknowledge that the child’s own behaviors or peer context were not measured in this study, which may limit the interpretation of parent-focused effects. Similarly, the null findings (e.g., no association for parental aggression or delinquency at 21 years) should be briefly addressed with possible explanations or noted as areas for further study.

Conclusions

• Revise the Conclusions section to emphasize the key findings of this study and their implications, rather than restating general knowledge. For example, the opening sentence currently notes that children commonly experience victimization (a known fact); instead, start with what this study specifically revealed (e.g., the identification of perpetrator profiles and parental pre-birth risk factors for child victimization).
• Ensure that any recommendations or broad statements in the conclusion are directly supported by the study’s results. For instance, if suggesting parenting programs or interventions for sibling conflict, explicitly tie it to the findings (e.g., “given that a large portion of victimization is perpetrated by siblings, interventions to address sibling conflict should be considered”). Avoid unsupported generalizations like the current statement on “enabling children to be more resilient to traumas,” which is outside the scope of the data. Such statements should be removed or reframed as broad implications with a clear connection to the findings.

 

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

I enjoy reviewing this manuscript. However, please further sharpen the logic and the presentation in the following three areas:

One of the most compelling aspects of your paper is how you navigate the definition of "criminal victimization." However, I noticed a bit of a logical tug-of-war between the Introduction and the Discussion. If the Intro excludes abuse, but the Discussion acknowledges that some reported acts might just be sibling rivalry or standard family friction, we run the risk of diluting your findings.

I’d suggest running a sensitivity analysis that specifically excludes sibling perpetrators. By narrowing the focus to clear-cut criminal acts by non-family members or adults, your argument about the ripple effects of parental mental health becomes much more robust. It removes the "noise" of normative family conflict and lets the actual criminal victimization data shine.

I was really struck by your "surveillance" hypothesis in the Discussion—the idea that parental mental health directly dictates how children are monitored. It’s a powerful mechanism, but it feels a bit hidden where it is currently. I’d love to see you pull that logic forward into the Introduction. If you introduce those mechanisms early on, it helps the reader understand the "why" behind your research questions from page one. It frames the study perfectly: we aren't just looking at a correlation; we're looking at how a parent’s internal state influences a child’s external environment, including their interactions with neighbors and teachers.

To make sure your readers are interpreting the data exactly how you intend, a little more "signposting" in the tables would go a long way. Specifically:

• Precision in Table 3: I noticed the Confidence Intervals for the "Father Perpetrator" group at 18 years are quite wide. That usually signals some instability due to sample size. A quick note in the text acknowledging this will show the reader you’re being rigorous and cautious with those specific figures.

• Directionality in Table 4: Regarding that family income beta weight (-.13), it’s worth a quick sentence to explicitly state that lower income is tied to higher perpetration. It’s a small detail, but it prevents the reader from having to do "mental math" to confirm the direction of the relationship.

The biggest asset of this study is the temporal ordering—the fact that G2 factors influence G3 outcomes. To really drive home that "causal" flow, I’d recommend including the specific birth years of the G3 cohort in your Methods section. It provides a concrete timeline that proves the parental assessments at age 21 happened well before the children were even born, let alone victimized. It’s a simple addition that makes your longitudinal design look much more "bulletproof."

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 4 Report

Comments and Suggestions for Authors

Dear Authors, I read your manuscript with great interest and appreciate the effort of collecting data in the span of 40 years. Before publication, I would like for you to make some small changes: 

- the hypothesis n.3 (lines 74-77) is unclear and should be reworded;

- the sentence at line 79 should be deleted.

 

Author Response

socsci-3981828

Reviewer 4

1. The hypothesis n.3 (lines 74-77) is unclear and should be reworded.

We have reviewed this section and believe it is clear.

 

2. The sentence at line 79 should be deleted.

This has been done as requested.

Round 2

Reviewer 3 Report

Comments and Suggestions for Authors

Dear Author(s),

Thanks for the effort to revise the paper. I have no further comment, and I think it is ready for publication.

Congratulations!