• When the media reports on science, continue to be skeptical

    There’s a new study on child development making the rounds. “Longitudinal Links Between Fathers’ and Mothers’ Harsh Verbal Discipline and Adolescents’ Conduct Problems and Depressive Symptoms“:

    This study used cross-lagged modeling to examine reciprocal relations between maternal and paternal harsh verbal discipline and adolescents’ conduct problems and depressive symptoms. Data were from a sample of 976 two-parent families and their children (51% males; 54% European American, 40% African American). Mothers’ and fathers’ harsh verbal discipline at age 13 predicted an increase in adolescent conduct problems and depressive symptoms between ages 13 and 14. A child effect was also present, with adolescent misconduct at age 13 predicting increases in mothers’ and fathers’ harsh verbal discipline between ages 13 and 14. Furthermore, maternal and paternal warmth did not moderate the longitudinal associations between mothers’ and fathers’ use of harsh verbal discipline and adolescent conduct problems and depressive symptoms.

    The WSJ trumpets, “Study Says Yelling Is As Hurtful as Hitting“. Is that what they found?

    I went to the paper. Here’s their measure of yelling:

    These items were adapted from the Conflict Tactics Scale (Straus, 1979). Items were: “In the past year, after your child has disobeyed you or done something wrong, how often have you: a) shouted, yelled, or screamed at the child, b) swore or cursed at the child, and c) called the child dumb or lazy or some other name like that?” Items were rated on a 5-point scale, ranging from 1 (never) to 5 (always). The three items were averaged together to create the constructs of mothers’ and fathers’ harsh verbal discipline. The constructs demonstrated good internal consistency at each time point (mother: α = .77 and .78; father: α = .75 and .77).

    OK, that’s a little odd to me. Full disclosure: I’ve yelled at my kids. I hate it when I do, I regret it afterwards, and I almost always try and apologize later. But you know what? Kids yell at each other sometimes. They yell at me sometimes! My parents yelled at me sometimes. It’s part of being human. So anyone who didn’t answer yes in some way to (a) would be odd in my book. But that’s way different from calling a child dumb, lazy, or some other name. I don’t think I’ve even sworn at my children ever. But these three things are treated equivalently for this study. Is that valid? I have no idea.

    But if you get past that, I give you the major finding of this paper, the relationship between harsh verbal discipline and conduct problems:

    ResultsDo you understand that? Me neither. I think they are showing that there are statistically significant relationships between these factors. I don’t doubt that. But does that mean that yelling at your kids is as harmful as hitting them? No. As always with the media and research, take what you read with a grain of salt.



    • There’s yelling and there’s yelling. Do you yell at your kids because to get down here right now and pick up this mess they made and left? Or do you yell at them because you had a stressful day and their normal kid behavior is getting on your nerves?

      Then, of course, there’s levels of spanking, from a firm swat on the behind of a toddler who refuses to sit in a time-out to whipping a kid for being less than impossibly perfect. According to current theory, both are going to damage your kid for life.

      • I hope you don’t think I’m arguing with you. I agree that there’s nuance, and I’m not convinced this study picked up on that.

    • The paper studied “yelling” (and it’s variants). It doesn’t appear that the paper say anything about “hitting”. WSJ (unfortunately, as usual) goes off the deep end and makes stuff up.

    • You might actually know and just being facetious, but to the other readers out there. This is a path analysis (subset of Structural Equation Modeling). The cross-lagged idea is that you are using an experience at one time to predict something in the future while trying to adjust for confounders, i.e. trying to get evidence towards causality. As with most correlational/regression research you have the potential for omitted variable bias.

      In path diagrams double headed arrows are correlations which are non-directional, and single headed arrows are regression coefficients.

      What this diagram is showing is at age 13 Paternal Harsh discipline is correlated with conduct problems (.19) and with Maternal Harsh discipline (.15), the latter two correlated at (.15). This correlational pattern is roughly the same at age 14. [I am guessing that higher conduct scores are bad].

      Now they use the measures at age 13 to predict those at 14, and what we see is that harsh discipline at 13 is associated positively- to conduct problems at 14 both paternally (.11) and maternally (.12). At the same time they find that conduct problems at 13 are positively associated with harsh discipline by father (.09) and mother (.10) at age 14. lastly, if a parent yelled in the past they will yell again (.23 and .21, respectively).

      Bottom line, strongest thing found was if you yelled at your kid in the past, you are more likely to yell in the future. And conduct problems in the past predict future yelling as well as yelling predicts future bad conduct…

      1) Since you mention that conduct problems is actually an average of 3 items it really begs the question why they did not let that be a latent variable with 3 indicators (removing any potential measurement error). Why add that error back into your model- if you are comfortable using a cross lagged panel you should be ok going beyond Path Analysis to SEM.
      2) Why are they not looking at the impact of paternal harsh discipline at 13s effect on maternal’s discipline at 14? (and semi-vice versa).

      • Great comment. It’s a little being facetious, but it’s also a large extrapolation from their statistical findings to major statements about “yelling versus hitting”. The point is that this is nuanced and complicated, but the media doesn’t do that well. I appreciate your work above.

    • Typically, I agree that the media has a tendency to inaccurately report studies of these sort, especially on the topic of parenting. I am even more loathe to defend the WSJ’s reporting on any subject. However, in this specific case, I think you are mischaracterizing these research findings.

      First, this study nor the WSJ are not claiming that occasionally losing it and yelling at your kids is the equivalent to physical abuse. I think it is fairly clear that the concerning behavior is consistently using harsh verbal discipline, aka verbal abuse, as a disciplinary tactic. Although it is true one of the elements they consider is described simply as yelling, it is rated on a 5 point frequency scale. It is pretty clear to me that the authors’ argument is concerned with yelling, cursing, and degrading insults when it is a consistent and frequent method of discipline. From WSJ “Those kids whose parents used higher levels of harsh verbal discipline when their children were 13 experienced larger increases in behavior problems the next year, including fighting with peers, trouble in school and lying to parents, as well as symptoms of depression.” Interestingly enough, that is nearly a word for word echo of the paper’s conclusion.

      Second, you don’t even mention one of their most important contribution which is that these effects are consistent even when mitigated with actions of parental warmth. Likewise, this is not an isolated study, but part of larger and seemingly consistent literature on the subject.

      Third, the graphic presentations of their results are kind of terrible, but bidirectional models are really tricky to display. I found it pretty easy to follow that that a child’s depressive symptoms was not correlated with the amount of verbal abuse (figure 2), while a child’s bad behavior did increase the degree of verbal punishment (figure 1).

      Fourth, to Mark Spohr’s comment that: “It doesn’t appear that the paper say anything about ‘hitting'” – the study does directly address physical discipline – “Parental physical discipline was assessed at Wave 1 when the child was age 13 by two items completed by the mother and the father, respectively. They were asked: “In the past year, if your child has disobeyed you or done something wrong, how often did you (a) push, grab, or shove him/her and (b) slap or spank him/her?” Items were scaled using a 5-point answer format ranging from 1 (never) to 5 (always). The two items were averaged to create the physical discipline construct (r = .42, p < .001)." They go on to explain that "Most importantly, we include parental physical discipline as a covariate. The inclusion of physical discipline in the models will enable us to examine whether the effects of harsh verbal discipline on child adjustment are due to verbal discipline or are a function of its co-occurrence with physical discipline."

      Finally, it appears that your personal experience may be misleading your interpretation of these results. By your own description, you seem like a great parent. You also probably did not grow up in this type of environment. If you had one or more parent that spent your childhood yelling and verbally demeaning you, it is probable that the results would not have been very surprising. Nor would you be likely to confuse an occasional outburst with the type of verbal discipline described in both the article and study.