Child Externalizing and Internalizing Behavior and Parental Well-Being During the COVID-19 Pandemic

Thank you for the helpful reviews of this manuscript. We have pasted the reviewers’ responses in full below and responded in bold after each comment. All changes to the manuscript have been highlighted in bold font.

Review 1 

This is a well-written manuscript with good design and analysis. It is interesting to see that Wave I child externalizing behavior predicts Wave II parental depression and stress to some degree. I would like to see further discussion on the rationale for why externalizing behavior but not internalizing behavior had this impact, as internalizing behavior is usually related to maternal depression. I would also like to see the transactional model mentioned in the Introduction to be integrated more in the Discussion. Otherwise, this was a  scholarly and well-crafted piece of work

Thank you for this review and for raising these helpful points. We now discuss the internalizing behavior findings in more detail in the Discussion section. We also interpret findings in light of the transactional model throughout the Discussion section.

Review 2

The data come from 66 countries and two waves so the term prediction can legitimately be used - but with only 175 familes out of over 2000 initially recruited and the use of a convenience sample recruited via social media severely limit both the generalisability of the findings and the statistic power for the analyses.

We recognize these limitations and now discuss this in further detail in the limitations section.

A flow diagram of participant inclusion would help, because the tables also suggest signifcant attrition between wave 1 and wave 2 among the original 175 families. An analysis that compared those who participated in wave 1 and wave 2 compared to those who dropped out would be revealing and would help intepret the findings.

We have now included a participant flow diagram in Figure 1.  We have also now included an analysis comparing those who participated in Waves I and II to those who did not participate in Wave II.  Participants who dropped out did not differ in terms of Wave I parental depression, parental stress level, child externalizing behavior, child internalizing behavior, or child age group. However, participants who dropped out were significantly younger than those that participated in both waves.

Sadly, the lack of power is mentioned briefly in the limitations section of the discussion, which provides lengthy explanations for the findings in much greater detail.

We recognize that this is an important limitation. This is now discussed in more detail the limitations section.

The distribution of these families across the 66 countries is not reported, but would be important to know considering different responses and policies - might this be another reason for the authors findings?

We now report the location distribution of the 175 included participants in the Methods section.  A frequency table with a full list of countries can now be found in Supplemental Table 1.

The measures for childhood mental health, parental depression and parental relationships are valid and reliable - but I am not sure that taking the average score across the stresses, which implies that they are all equally stressful is valid - I can see why the authors did this but the limitations of such an approach should be discussed.

Participants were asked to report which stressors they had experienced.  For those stressors endorsed, the participant was then asked the level of stress caused by the stressor. We summed the level of stress for each stressor endorsed by the participant. We used this approach, as individuals may vary in the extent to which they consider a particular stressor to be problematic. For example, “noise at home” may be very stressful to some participants but may only be a minor stressor for others.  We clarify this in the Methods section.  

There is multiple testing and three outcomes, with no adjustment, and it would be useful to see the actual p values rather than report p> or p< 0.05 - given the small atypical sample, I would not report marginal findings, and would argue that p=0.08 is weak evidence indeed.

We report exact p values in Table 4. We have now added exact p values to Tables 1 and 2. For ease of presentation, we report exact p values for the bivariate correlations in the text of the results section. We now do not report p values of less than 0.10 as being marginally significant.

Quite a lot of the literature summarised in the discussion seems quite different to the pandemic - where the issue for most families was being locked down with each other - ie Hurrican Katrina and the Boston Marathon bombings are distinct traumatic events lasting minutes or hours and not months.

Thank you for pointing out the important distinction between these events. While these events differed in terms of their duration, many did have serious lasting effects for individuals and families. For example, millions of individuals were displaced as a result of Hurricane Katrina. Up to 600,000 households were still displaced a month later. Six years post-Katrina, the population of New Orleans was only about half of its pre-Katrina size (Plyer, 2016). Nonetheless, we recognize that these events are substantially different from the current pandemic and discuss these differences in the Discussion section. We now also discuss research that examined the mental health consequences of the H1N1 pandemic, though the isolation caused by prior pandemics are not on the same scale as that caused by the unprecedented COVID-19 pandemic.

Finally, the recommendation for mental health interventions seems a big leap from these weak data - the issues that covid restrictions have thrown up for families is being confined to home, which might be overcrowded, either trying to work and home school, or housing, financial, or food insecurity  - a more nuance conclusion is that schools should be the last institutions to close and the first to reopen and that social support / or small group childcare for vulnerable families might reduce the increase in parental distress reported in Table 1, with financial support for those whose employment was at risk - of course mental health interventions for those with poor mental health (child or parent) are warranted but by no means are the only response

Thank you for making this important point. We have added to our discussion of implications in the Discussion section.

In this study we surveyed families’ experiences with parental depression, stress, relationship conflict, and child behavioral issues during six months of the COVID-19 pandemic through the COVID-19: Global Social Trust and Mental Health Study. The current analyses used data collected from online surveys completed by adults in 66 countries from 17 April 2020–13 July 2020 (Wave I), followed by surveys six months later at Wave II (17 October 2020–31 January 2021). Analyses were limited to 175 adult parents who reported living with at least one child under 18 years old at Wave I. Parents reported on children’s level of externalizing and internalizing behavior at Wave I. At Wave II, parents completed self-reported measures of stress, depression, and inter-partner conflict. Child externalizing behavior at Wave I significantly predicted higher levels of parental stress at Wave II, controlling for covariates. Child internalizing behavior at Wave I did not predict parental stress or depression, controlling for covariates. Neither child externalizing nor internalizing behavior predicted parental relationship conflict. The overall findings demonstrate that child behavior likely influenced parental stress during the COVID-19 pandemic. Findings suggest that mental health interventions for children and parents may improve the family system during times of disaster.