We are the Association for Child and Adolescent Mental Health, or ACAMH for short. And this is ACAMH learn. So welcome to Mind The Kids. I'm Dr. Jane Gilmour, honorary consultant clinical psychologist and Child Development Programme Director at UCL, And I'm Umar Toseeb, Professor of psychology at the University of York, focusing on children and young people's mental health and special educational needs. In each episode we select a topic from the research literature and in conversation with invited authors, sift through the data dilemmas and debates to leave you with our takeaways. Today we'll be discussing intimate partner violence and aggression and their effects on children. A note here, this paper focuses on lower levels of violence. And so we use phrases like minor or less severe. This phraseology, of course, does not imply that we consider them acceptable. In this episode we ask, does minor aggression have a major impact? So in the research that I've usually come across or even done myself I'm interested in violence between children within the home, so sibling to sibling. That can be physical, emotional, et cetera, et cetera. But there are other types of domestic violence or violence within the home. So we can do child to child, parent to child, child to parent maybe, minus parent to parent, so between the adults. And that forms of violence, when it doesn't involve the child, the child can either witness it, or they might not witness it but it's happening within the home. And I think that distinction is made in the literature about whether you see the violence or you don't. And it is interesting about the witnessing, being able to observe it, if you like, and being witness to it, or it's nonetheless happening around you. And the effects are still apparent. I mean, I'm interested, Umar, about when you're saying let's think about the different participants, let's say, and observers with those contexts of violence happening in a family setup, if you like, because here we're thinking about young people witnessing violence between parent figures, potentially. So it would be interesting to think about it in the context of attachment theory because at its core, what we would be describing on person witnessing that violent interaction is a violation of that child's right to a safe place, a lack of security, and concern, probably, over their attachment figure. And we actually know that those young people witnessing violence at home are more likely to be described as insecure in terms of their attachment. Pasco Fearon has done a really interesting systematic review that also shows a link between insecure attachment and subsequent externalising behaviour. So there's a lot of links there. But you can see why these domains of literature have an interaction. And I think the other thing to say from an attachment literature point of view is that if one attachment figure is experiencing violence from another, no matter how minor that might be, then those parents or parent figures can't be emotionally available to their child to moderate the stress of witnessing that moment. So there's a lot of attachment theory within that, which of course has relevance for the other individuals you were talking about, so for example, a parent to parent violence or a young person experiencing violence from a sibling. But it's particularly relevant to the context of the paper that we're going to discuss today. That said, I hadn't thought of attachment theory when I was trying to read around this topic. But actually, yeah, you're right. It's actually very relevant. And when I've taught children and young people's mental health and behavioural difficulties, we do talk about attachment theory as one of the core theories which influences lots of later difficulties. And actually, it makes sense here as a mechanism through which intimate partner aggression then impacts upon later outcomes. But one of the things that I did want to just talk about before we get the guest in, which is when I was-- I don't know if it's still the case. But when I did A-level psychology, one of the first things that you learn is the social learning theory of aggression. So you have this-- I think it's the Bobo doll experiment, Bandura, where the child sees some sort of aggression. And then they test whether they've learned aggression, et cetera. Is that still a predominant-- not predominant. But yeah, is that still a dominant theory in the field of aggression? Because I feel like a lot of social psychology that we learned 20 years ago has now been questioned and reevaluated. Is that still a thing? I mean, the short answer is I'm the wrong person to ask. We have an expert here to talk about that. Certainly, if we consider that social learning mode theory, it certainly has high relevance for this data set, in part, because there's a very strong link between this interpersonal violence or intimate partner violence and externalising behaviour specifically. So that certainly supports the hypothesis that social learning is a key mechanism. But if we were to pull apart the theory a little bit more, we would consider the importance of the model. So in the theory we would predict that if a boy was witnessing male to female violence, they would be more likely to show higher levels of violent behaviour than girls because they identify more closely with the violent model, if you like. But my understanding of the data is that there are conflicting data on this point. We're holding constant the idea that boys in general are more likely to show externalising behaviour. But there isn't a clear outcome here. And that might be something that we could raise a little bit with our expert to explore in more detail. I think the other thing to say is that there are other models that sit well with a social learning theory. They're not necessarily competing models, but they may be augmented. So one example would be the cognitive contextual model. And that shows that the individual perception of a given behaviour will influence the outcome. And that's Grinch and Fensham's data. And there's some model, rather. And there's some interesting data to support that. So I think it's true to say that A-level psychology-- it's the classic. It never goes out of fashion. But I think there's also ways of looking at that model as still being highly relevant, but thinking about some new data that will add nuance to it. Certainly, I think that's the case. You've mentioned nuance here. One of the things that we learn as-- I mean, I think we learn it in A-level psychology as well, which is around, is it nature? Is it nurture? It probably is a combination of both. And it would be reasonable here for me to suggest genetic propensities towards aggression. I think that there has been-- I think that I'm aware of a genome-wide association study where we looked at genetic-- well, somebody's looked at genetic variants associated with externalising behaviours. So I imagine an added nuance to what you've just said is that some children might have a higher genetic propensity towards aggression which then manifests in a given environment. And that given environment might well be intimate partner aggression. If your parents are engaging in aggressive behaviour and you see that behaviour, or even if you don't see that behaviour, that might be the environment in which your genetic propensity for aggression manifests. And I think on that, I think that we know that there are some existing data looking at callous and unemotional traits, for example. So those young people who have those traits and witness violence are relatively more likely to show violent behaviour in the future. But the mechanism of transmission here is debated. So in other words, is it an environmental experience and observation? Are the genetic impacts? So I think you're right in putting a pin in that particular data set. I think that will be an area that will really add to our understanding of these processes. So I think it's a pertinent point to add. It's really useful to consider. Let's introduce the guest. Let's go. I think we should OK. So we're joined by Dr. Hedwig Eisenbarth from Victoria University of Wellington in New Zealand. Hedwig is the lead author of the paper, "The impact of less severe intimate partner aggression on child conduct problems," which published in the JCPP Advances. Welcome, Hedwig. Thank you so much for having me. This is amazing. And I'm looking forward to talking with you. Thank you. So you've heard our little chat and our musings and hypotheses, and what we think is going on. So based on what we've described in terms of our understanding of what intimate partner aggression is, can you tell us about what you mean by intimate partner aggression in the context of the conversation that we're going to have? Yeah, I think that's a really important point to define the space. And intimate partner violence is the word or the concept has been used predominantly in the literature, also in terms of outcomes for those experiencing that. And that is usually oftentimes really related mainly to violence more literally. And that would be sexual harm or physical harm. Intimate partner aggression would encompass beyond that. So it encompasses both physical harm, sexual harm, but also, what sometimes is called psychological harm, so anything that has to do with threats and that is not leading directly to a physical harm but could potentially lead to something. So it would be shoving somebody around, would be shouting at each other, but also, the more the threatening of showing any sort of violent behaviour. And we use that concept in this current study because we wanted to be broad, but also, actually, really look into what are the effects of something that is actually less severe because people talk a lot about. And it's of course important to consider what happens when people experience the physical violence and the sexual harm. But we were also wondering what happens if that is actually not of that severe type, and what would be the outcome for children growing up in this kind of context. That's so helpful and I think highlighting the psychological message, if you like, around those actions, although in one level they are physical in their expression. Actually, what you're getting at is that the emotional experience or the emotional indication of how I respond to a partner or how I regard that partner may be of great import. That's a really useful definition. That's very helpful indeed. Can you outline a little bit in broad terms the links between in IPA and child outcome? Yeah. So what we found in this study-- and that's based on the growing up in New Zealand-- data set, which is an amazing longitudinal study run outside of the area around Auckland in the North Island of New Zealand. And they follow up children who they have been recruiting, basically, through the mothers when they were still pregnant, through the life. They are now about 16 years of age. And we selected a subsample of them. And the ones where they had reported partner at the time of pregnancy, plus the first nine months. And what we found there is that this intimate partner aggression that is reported by the mother during pregnancy and up until the first nine months of the age of the child had a positive impact in terms of it had led to an increase of child conduct problems at a later age. So the child at age four and a half and eight. While the partner reported intimate partner aggression, they get the same questions. And it's also not directional, those questions. That was linked only with conduct problems in the child in early childhood, so two years of age, but not later on. It's interesting. Yeah, there's a lot to think about. One of the things that we were discussing earlier was taking a very broad view of what we know about the literature for in terms of other risks and protective factors. I mean, you've done extensive work on, for example, bullying in the broader sense. But Hedwig, I'm wondering if there is-- could you take us a tour around the literature about some other risk and protective factors that we might cue into, thinking potentially about mental health and parental style and so on? Can I supplement that question-- Yes, you can. --and add slight bit to it? Which is, to what extent is the measure of intimate partner aggression just a proxy for the rest of the home environment? Is what you're measuring intimate partner regression? But actually, it's just telling us a lot about what's happening in that home in general. Yeah, I think this is a really good point. I mean, apart from that, we always need to stick with what these data and the results are based on. So basically, this is mothers during pregnancy reporting on how they are interacting with their partner and the partner the same. And this might definitely only reflect a broader picture of a context that they are living in. And so the strain that they experience that might be reflected then in intimate partner aggression might come from different other sources. So there might be financial strains. There might be health strains. There might be other issues that people are facing that actually look then or that they also are apparent in this intimate partner aggression that they report. So I do also think that we never say this is an isolation, only one thing that we're measuring there. And this brings us to the other risk and protective factors. We know a lot more about risk factors unfortunately. And protective factors are not only the absence of risk factors. But the risk factors that we know about for conduct problems-- as you mentioned earlier, oftentimes they're referred to as externalising behaviour in young people-- are a lot related to parenting behaviour. There is a little bit of genetics involved. Umar, you pointed out earlier. But parenting behaviour, in terms of harsh parenting, inconsistent parenting, and the lack of warmth, maternal warmth specifically, and other variables are then linked to the person themselves. So things like cognitive abilities, they are most likely not driven by this context, but potentially also more genetically driven. And so these kind of factors play a strong role for the outcome. Plus there might be some biologically driven factors. And so this brings us back to the question of, how do we think-- what is the mechanism? Why is that observing-- or being in the context, because we didn't ask. We don't know if the children were present when this happened, what the mothers reported and the partners. And actually, it's quite interesting that it seems to not matter hugely. The children don't need to be in the same room. It's the context again. And that's why it seems to be almost like a proxy measure of that context that they grow up in. Plus, the first part of the data set-- these were pregnant mothers. I mean, you could say, well, there's some voice that you can hear. You can hear a raised voice. But there is something going on in the mother that is potentially linked to stress in this moment of experiencing that aggression, not only them experiencing themself but also them being aggressive. Again, we don't actually know because it can be both ways. And that's also quite important. So there is something going on that creates a state that will have an impact. And we do know that, for example, also components within the mother in terms of traits and low stress reactivity, for example, will have an impact on later externalising. So all of these variables are definitely part of that. And that's also why we brought them into our analyses. And we thought we need to look at also, for example, mental health of the mother and the context that-- again, part of a bigger context. And this could be mental health problems in the mother or the partner which arise from the intimate partner aggression. They can be independent of that. They can be there before the intimate partner aggression happens. And so for that reason, that was really important for us to incorporate that as a potential mediator in this analysis. And that's where we need this longitudinal kind of studies where we can actually look into this at different time points and to discern what is first and what comes later, and to control for all these variables over time. So I want to just unpack something where in your paper you talk about less severe or not as severe intimate partner regression. And you discuss that in the introduction where you give some context. And this is not what you tested. But I just want your view on this. I would hypothesise that no aggression is associated with the best outcomes. The most severe aggression is associated with the least favourable outcomes. And then there's a gradient in between. So you focused on the less severe aggression. And you found that those children fared worse than those with no aggression. But is it just a quantitative thing? Would you hypothesise that the more severe the aggression, the worse the outcomes would be? And what you're demonstrating here is that even at the lower end of the severity it does still have an adverse effect. So you're talking about the dose response curve basically, so the more, the worse. Yeah. We couldn't tell from our data set because we don't know the severe types. That's one of the things with longitudinal data sets sometimes-- that it's hard to ask these questions. And these specific questions have been asked. And so we cannot really control for the severe forms. And so we wouldn't know. And I would be not inclined to make assumptions. And I can just say it has an impact. It has also an impact. But we can't compare it. And we need to keep that also because we're in a different context. So much of the research of this type has been happening in different cultural, other contexts. So I wouldn't want to really say like, of course, that's the same, just because the effect size is the same or something like that. So I wouldn't want to do that. And I appreciate it as an expert ever would say, I'm not prepared to make assumptions. This is an empirical question. I also wonder if there might not simply be-- yeah, we don't know the question about dose effect. Coming back to these competing theories or complementary theories we were discussing at the beginning, would there be individual differences? Whether that's the age of the child-- if I'm in my teenage brain phase of life, my brain is particularly receptive to environmental experiences. Might the experience of witnessing or that aggression being around me have potentially greater impact than potentially at another stage of my development? Or are there genetic vulnerabilities in the way I process information that mean I experience that at a greater intensity, holding constant the degree of severity of the violence? And I totally applaud your caution in making assumptions. But it's a very interesting set of hypotheses to explore as another step. And your data set is so valuable in just raising the question that less severe forms of aggression have significant impact, so let's think about why. Yeah. I just want to also link in this [AUDIO OUT]. I'm very grateful that I'm able to work with the data set which is collected by a set of researchers that I'm just associated with. Really, I'm also not part of the owner. I don't own that data set. But I was really, really lucky to be able to work with it. It is special, and it will allow us to ask those questions, though, because I do agree that this will be really important to look into what happens over time. And when does it matter most, or more, or less? And coming back to your initial ideas or thoughts around the theories that might play a role here. And I do strongly think that attachment theory is definitely one theory that we need to think of. And we do know that retrospectively-- so when we ask adult people about their attachment experiences. And it can be debated if that's the right way to do this research. There's a strong link here between attachment and externalising in adult people. And so this is definitely playing a role. On the other hand side and talking about the biological underpinnings, some of these traits or some of the behaviours are put in motion really early on. And then there is an interplay of different variables that make it also protective variables. And we need to understand those better. That might actually be reducing the chances that they are getting carried out or as it's called, they are sleeping or something like that. And we do know from some nice longitudinal work in other parts of the world that some children would show behavioural problems. They would show also these colours on emotional traits that you were referring to. But then they don't show that later on-- they fade. And some things are there that can contribute to that. And so when we come back to this protective idea, some parenting and parenting help might help, might be actually be quite helpful. But it's never going to be the only thing. There might be different other variables as well. So I don't think we fully understand that yet, at least I think so. But it will be really, really key to discern a little bit. I mean, we always want to reduce harm and this intimate partner aggression anyways. It's not good for any of the parties. So there's actually a no-brainer to just say, let's reduce it. But as you point out, [INAUDIBLE] time points in the development where it's much more important to really make sure that the child has protection, that will be quite key. And just from a police point of view, from really the prosecutor's point of view, what they say oftentimes is that this is forgotten. So they show up at the doorstep when they get a call for domestic violence. And they take care of whatever's happening there. And it's forgotten that there's this child sitting in another room in the corner. And that's the thing that I think even-- this is probably severe cases that get where the call happens. But this is where we need to get better with taking care of that super early, much earlier than we think, much earlier before we think that they can observe, and see that, and copy that behaviour. That would be the social learning theory type of modelling. So the messaging is not simply to the parents in terms of understanding that or indeed to practitioners, but also other agencies that are involved in child protection and safeguarding. I think that's a really important point. Can I make an observation and then ask a question? I can go ahead and hypothesise about the developmental stage. So when in our work with sibling bullying we find that sibling bullying, so violence between children in the home, decreases as they go into adolescence and as they grow older. And we explain that in terms of, as children become adolescents they spend more time away from the home, and then navigate towards friendships and peers, and spend less time at home. So there's less opportunity for bullying between siblings. I would say that what you'd probably expect is that the effect of intimate partner aggression on the child is less severe in adolescents because they navigate away from the home. So even though there is a high end time, I would think that maybe just because they're spending less time at home, the influence on their outcomes is less. So I would say it might be less of a problem, but still a problem. The question that I was going to ask you was-- we've talked a lot about intimate partner aggression. And I know that there's been a change in society in the last 10, 20 years in terms of violence against women and girls. And I suppose there's lots of questions here. So when we're talking about intimate partner aggression in the context of your study, do we have any indication of whether it's male to female? So is it from mum to dad, predominantly? Or do we have no indication of that? And then in terms of intimate partner aggression, how have these trends changed over the last decade or two? Is it more of a problem now? Is it less of a problem now? Do we not know. Great question. So we don't know in our data set which direction, which in some ways is sad. But also, I think it's probably quite protective for the two of them answering the question. And for that reason, I think there's a little bit of like, oh, that's actually might be a good thing in this kind of context of asking people who are in a longitudinal study those questions. So we wouldn't know from our data set. And we just know that it's between them. From the research side of view, there was a predominant focus on the mothers being affected most times. And so the mother or the woman, if you want to say gendered, being the perpetrator is actually something that has been considered only more recent-- some time as well. And it's been definitely also much more under-reported by men to be subject to violence and aggression in the relationship than women would under-report. So there's under-reporting overall. But the under-reporting is even stronger for men no matter in which type of relationship they are. And so that has strong impact on all the stats that we look at, really. And also, we do know that there are cultural differences in reporting and help seeking for that type of experience. And so again, that's something that's just in the mix that we need to consider. So it matters how was that research done. How have people been asked? Was this completely anonymous? Did they show up somewhere? Where they asked on the phone? This really is, I think, not easy to discern, really. I cannot speak to the development over time. The prevalence is high, if you want to say so. Across the globe, if you look at the WHO statistics, for example, that they put together, it's a lifetime experience of prevalence of 24% approximately and a past year prevalence of 4%. And there are some countries that go way beyond that and other countries that go way below that. It varies quite a lot across the globe as well. But I think we need to consider it high. And Europe-- I think there was a paper that looked at specifically European countries. And that lifetime prevalence was, I think, 48%, so really, really high. And I think that there's some reflection. It's always this difficulty of there's a reflection of more reporting, more acknowledgment of this is what I experienced. This is what this means when I experience that compared to the past. It's difficult then to discern, I would say. And I don't know about research. There might be great research out there that actually looks at that to discern if that's a reporting thing or if that's something to do with actually changes to the way that it's addressed publicly. And people do seek more help. I think I'm struck by two things, one is the potential cohort effect, depending on the acceptability of coming forward and the other is that cultural shift. And I'm just thinking that New Zealand is home to two very valuable longitudinal data sets. And presumably, I think the Dunedin Study certainly was selected-- the community was selected because they weren't transient. They were likely to stay where they are. And it's easier to catch up in the various waves of data collection. And so those specific cultural differences in reporting or acceptability of particular behaviours and so on will be very important if we are extrapolating on two occasions from the New Zealand communities, partly because of some of the cultural differences that make them specifically such a valuable community to measure over time period. So I think there are many moving parts in these equations, not least the cultural and cohort effect. Yeah. The nice thing, actually, about New Zealand in addition-- well, there's also more longitudinal data sets. It's just amazing. But that actually also, there are different cultural groups within that. So we could even when we can potentially look at different cultural subgroups within the data set which are reflecting very, very different ways of approaching violence reporting, talking about things that happen in the family. So that's something we could look at. But I fully agree because people bring up the cultural differences a lot from also a religious point of view. And then there are some religious contexts in which-- so that's not necessarily the country necessarily, but then the religious context that relates to that reporting behaviour and help seeking behaviour as well. Can we speculate and hypothesise some more? So I wonder whether child outcomes differ depending on who the perpetrator of the violence is. So if the violence is from male to female, I imagine that that would have a different impact on behavioural outcomes or children's outcomes in general than if the violence is from female to male, so if it's mum to dad versus dad to mum. And I think it's partly because I would guess that the context in which violence would happen from dad to mum would be very different. It might be a certain type of home, or community, or culture, or set up, or whatever where that setup would be different if the violence was happening the other way, because I think violence from male to female-- would that not be? Am I wrong there? Yeah. Well, I do strongly think we don't know really. I think we underestimate what's happening in the homes. I really think so-- and not only the homes but in relationships. And clearly, this is just our stereotype. Our stereotype is, well, however we grow up in whatever-- if we say Western context. For me it's clear that if there is violence in the relationship, it needs to be the man perpetrating violence toward the woman. I mean, it might still be of dominant thing to happen. That might be true, but I don't think it's that strongly shifted. Also, I'm not sure and I wouldn't know any data that would have a comparison for cultural differences in that sense, if that happens more in some cultural contexts-- the man being the victim, basically, than in other contexts. But will be super interesting to look at. I fully agree because I think that would be super, super interesting. The only thing I can think of that goes into this direction is if you were to look at reports of self-perpetrated externalising of the mother and the father-- there's a data set that I'm thinking of. That's why I'm saying it this way-- and then the outcome in the offspring. And both the gender of the offspring-- you could look at into that. And I have this vague memory of that one data set that I'm thinking of. There were some nuances to that, yes. But again, this externalising is not necessarily the perpetrated within the relationship. So it's not close enough to what you're looking for there. I think what we need is either for someone to do this study, or if someone who's listening has done this study to be like, oh, yes, this has been done. And this is what the finding is. Yes. You need to stop stereotyping and making assumptions. Well, the only thing that remains for us to do, Hedwig, is with us to thank you so much for your time. And your consideration and thoughtful interpretation of the data is-- in general, that's a really good model for us all if we're thinking about how to interpret data, and an indication of an expert takes an expert position with the confidence of knowing that I can interpret certain data sets. And there's some questions that remain unanswered. And that is a wonderful experience and lesson for us all. But thank you so much. We have appreciated your consideration. And we look forward to hearing what happens with the data in the next data set collection. Thank you again. Yeah, that was fantastic. Thank you so much for having me. It was a pleasure to speak with you and to discuss and hypothesise together. Yes. I follow you. Let's unpack that. I just want to try and question my questioning of assumptions. Was the problem that I wasn't acknowledging the fact that female to male violence is a big thing as well? Was that the problem? I think, well, this reminded me of the period of time where we were unaware as an academic community that, for example, there was sexual abuse perpetrated by women. And there was a period of time where it was assumed that various events were only the domain of a male perpetrator. And so I think what Hedwig was raising is that although we come from-- of course, we all come from assumptions. It's inherent. It's part of being a human being-- is that the abhorrent nature or the anomaly of particular patterns of violence remains unexplored. I think the reason I said, we can't say I feel like-- it's my pet hate. You don't feel anything. You feel an emotion. You don't feel [INAUDIBLE], so that's why. I think my, "I feel like," was like another way of saying, my understanding of the situation is that. It's a young person's vernacular, Omar. And I'm too old to say I feel like. But you are, so you have permission to use it. I'm not that young. But yeah, maybe nobody has that permission. Yeah, that was really helpful. I think that it's nice because having these conversations gets us to challenge some of the misconceptions. Omar, I think you're particularly open to the idea of it, of questioning something and responding to something. And I think that is the joy of being around somebody who has the capacity to observe, to question in a way that invites curiosity. So I love that question and reflecting on it is such a positive experience. I think it's so valuable. It's very generous of you is what I would say. So should we do our takeaways for us? Yes. I just wanted to say two very quick things about the clinical implications, because I came across a really good review by Shonkoff, which is an overview of the literature looking at the long term effects of being in a chronically stressful situation and the effects on the developing brain. Because I think Hedwig touched on some of these, that aroused stress response system, which means that every time I walk into a stressful situation, I may react. My stress system overshoots, if you like. But there are also important impacts in terms of neurodevelopmental issues. So for example, learning, and memory, and executive function are likely to be affected. And of course executive function includes inhibition and emotional regulation. So if we think about those two functions in the context of expression of anger or contact, they are particularly important for our discussion today. I think the second thing, just to say very briefly, is that there are evidence-based strategies for conduct disorder. And somewhat encouragingly, they are well placed to address some of the processes that we've discussed today. So depending on the age of the young person, parent training is indicated, cognitive reappraisal programmes. So considering pro-social problem solving strategies and family therapy would be indicated. Now all of those approaches will typically include psychoeducation about modelling behaviour, emotional regulation, consistency to parenting, and so on. They're highly relevant here. But perhaps our add on for a clinical implication is that practitioners must underline the importance that any level of inappropriate or violent behaviour is likely to have an impact. And I think that additional messaging-- I'm using air quotes here. So if you're watching the video, you'll know that I'm air quoting. But minor aggression or little shoves, in inverted commas, will have a major impact. That messaging will be important in all levels of conduct disorder support because it will augment the message. What about takeaways? Umar, what would you say your academic takeaway might be? So we were having a discussion earlier about the environment within which intimate partner aggression takes place. And it made me reflect on some work that we did where we looked at the perpetrators of sibling bullying, the kids doing the bullying and the impact on their mental health years later. And what we find, I think from what I remember, was that if you were not involved in any bullying, you had the best mental health. But then if you were involved in bullying as a perpetrator or a victim, as in someone's bullying you or you're doing the bullying, you had poor mental health. It was a bit different. But actually, you are disadvantaged either way. And a lot of our explanations for why the bullies have poor mental health were about the individual, so something about their characteristics or their personality or whatever. But actually, what I learned from this conversation is the reason why bullies might have poor mental health is because the context in which that bullying is happening, and the impact that has on the communication and the interaction between different members of the family, and the support that different members of the family can give each other-- that support structure might then be broken down. And that might be why bullies also have poorer mental health compared to people who don't bully. So that made me think differently about some of my own work and how we try and explain the finding that people who bully but aren't necessarily bullied themselves have poor mental health within the home. And I think that's a really important point which is generalizable to other areas of investigation. Yeah, that's really useful. From a clinical point of view I think the takeaways are clearly, these data raise the important idea that the emotional context of physical actions or implied aggression may well be at the heart of harmful transactions. So in other words, it's a social and emotional messages that go alongside aggression that are particularly pernicious. And of course, the final point is psychoeducation for parents must include a discussion of the protective factors that may allow a young person to thrive in the context of stressful circumstances, but emphasising that witnessing violence at any level is likely to have a negative impact. So these small moments matter. They have a big impact. Excellent. That's been a really nice conversation-- lots of data, lots of hypothesising, lots of questioning, which has been great. [MUSIC PLAYING]