Exploring the Relationship Between Disability and Children’s Household Chores in Selected Low- and Middle-Income Countries

Comment

Reply

1/ Scope of the paper

I recommend that the analysis focuses on childhood disability and leaves out the disability status of the caregiver for several reasons.

Based on prior research, the disability status of the caregiver is likely correlated with the educational attainment of the caregiver and the wealth status of the household. This potentially introduces multicollinearity in the model.

Also, based on prior research, the threshold for considering an adult to have a disability that may impact socioeconomic status, including work and potential household chores, should be inclusive of persons reporting“some difficulty”, which is not what the authors do here (e.g. Hanass-Hancock et al 2023). Adults’ disability status is not measured in a way that is similar to that of children and MICS is not an ideal dataset for this exercise.

Indeed, as noted by the authors, MICS only captures the disability status of selected women or men in the household and therefore does not measure adult disability in a household. It cannot capture how adult disability may impact children’s time spend on household chores.

Finally, the lack of a significant effect of adult disability status is a distraction from the result of the significant association between having a younger sibling with a disability and the time spent on household chores.

We agree that MICS is not an ideal set for the measurement of the extent of adult disability – rather, it is intended as a marker of prevalence to allow for the disaggregation of various indicators by disability status. It is likely to underestimate the impact of disability given it does not include the entire range of difficulties posed by differences in functioning, notably mental health issues.

However, we opt to keep caregiver disability status as a control in our regression models. There are three main reasons for this. First, in our pooled data set, multicollinearity between the disability status of caregivers and their education for caregiver is not high. Second, we believe that the inclusion of this admittedly imperfect measure provides at least a partial control for caregiver disability status – which in turn helps to mitigate omitted variable bias. Third, we would argue that the finding that caregiver disability status may not uniformly impact children’s domestic workloads, even if estimates are potentially biased, suggests that at least in some contexts, any additional workload may be distributed in different ways that the existing evidence leads us to expect.

As a result, we do not exclude this variable from our analysis. Instead, when we introduce the measures of caregiver disability, we discuss limitations related to the information that can be derived from the Washington Group Short Set on Functioning questions (and why we believe the inclusion of these imperfect measures is nonetheless important). We revisit this issue in the discussion of results, both with reference to our analysis and as an area in which future research is needed.

To give more depth to the analysis on children, the authors should also consider adding to their analysis if having a sibling age 5-17 with a disability correlates with doing more household chores.

We don’t have this information. MICs collects information regarding disability status as part of the questionnaire for children aged 5-17 which is administered to just one randomly selected child aged 5-17 years in each household.

2/ Models and measures

-       Model Random effects vs Fixed effects: The authors present the results of a random effects model and a fixed effect model. For instance, Table 2 has both sets of results and is difficult to read. The authors should instead figure out which one to present as the primary model and consider including the other one as part of a sensitivity analysis.

Discussion in text focuses on fixed effects model as the random effects model is likely to overestimate impact of other covariates on disability by excluding country-level differences. RE model is presented as a sensitivity test; the table and corresponding discussion has been moved to an Appendix.

-       Measurement: the authors should consider using an index that captures the severity and number of functional difficulties across domains. This introduces more variation in the independent variable of interest may better capture a correlation between disability and household chores.

One of the co-authors' prior experiences with MICS disability data suggests that constructing an index to measure child disability would not significantly contribute to this study. This is due to the very small proportion of children with more than three functional difficulties or severe disabilities. As a result, creating such an index would likely introduce substantial measurement error, leading to wide confidence intervals and reducing the reliability of the results.

-       Table 1 and Table 2 should be aligned in terms of the covariates. For instance, the caregiver’s disability status seems to be a control as per Table 2 but is not in Table 1.

Table 1: the educational attainment covariate should not be underlined and the characteristics of the 2-4 child are missing.

Table 1 has been restructured and amended – it now contains the full list of covariates and their definitions.

-       The paper should have tables of descriptive statistics for dependent and independent variables, pooled and by country. The table at the country level should be in an appendix.

Descriptive results are reported for all dependent and independent variables for the pooled dataset and presented in a new table (Table 2).

3/ Writing and Editing

The authors need to discuss their results in an intuitive way in the discussion section so that the readers grasp the extent of the results more easily. For example, the authors should be able to figure out how on average, having a younger sibling age 2-4 with a disability is associated with doing X more hours of household work per week for the pooled sample and then give a range for the country level estimates.

We agree with this comment. Rather than report the marginal change in log hours or raw hours, we have reported simply the percentage increase in hours associated with a change in each covariate.

The discussion section should provide suggestions for further research. For instance, there could be research using time-use survey data that have the Washington Group Short Set or the CFM to study this issue or there is a need to collect such data if not already available.

We have added a section on future research directions. Key elements are the following:

1.      We are hamstrung by having only a partial household disability profile – all women of reproductive age, all children aged 2-4, a subsample of men typically 15-49, one child aged 5-17 per household. To better understand intrahousehold dynamics connected to household chores, it would make sense for some population surveys to collect disability profiles of all household members (including older adults), alongside time use data, at least for a subset of households. Richer adult disability data would also add value.

2.      More work is needed on the impact of involvement in different levels of domestic work – following on YL work on four countries showing that any level of engagement in domestic chores can be harmful to children if it crowds out time spent on education. The MICS 7 inclusion of time diaries for children should provide data allowing untangling time spent on UCDW v other aspects of life (schooling / leisure etc.) for a large number of countries, thereby permitting a holistic understanding of how UCDW fits into other activities children perform.

3.      The diversity of country experiences (and the importance of country-fixed effects in the pooled regressions), highlights a need to understand better what factors within countries are leading to greater and lesser engagement in UCDW for children.

More work is needed on the number of young carers in LMICS, the level of care they provide, and the impacts of providing such care on various markers of their wellbeing.

Comments

Reply

While manuscript is written in comprehensive way, it lacks positioning into the field (literature review), and it lacks discussion with the field in terms of what results actually mean in reference to previous studies. 

The paper is now better positioned in the relevant literature in two respects. First, the literature review in the introduction has been strengthened to include discussion of research, principally for HICs, on ‘young carers’ and effects of having a PWD in the household on their education, health and career prospects. There is limited research on young people providing care – including by performing domestic chores – outside of HICs, so this paper makes an important contribution in this respect.  Second, the literature review expands on why this study might contribute to the literature on the impacts of time spent on domestic chores for child wellbeing. It includes both references to the literature on ‘extensive involvement’ harming educational and health outcomes (on which the threshold for child labour is based) and on newer research which argues that the impact of involvement in household chores at lower levels too might harm child outcomes.

Also it has no outlet for country cultural or care systems differences or groupings. While unpaid domestic work of children is considered a cultural norm in most countries, we lack knowledge at what hours of domestic work this becomes harmful to a child. No hours of domestic work would mean that children are guests in their homes, not being responsible even for making their own beds, clearing after themselves after meals etc. In many countries complete lack of engagement of a child into daily functioning of families is considered culturally unacceptable. 

This is an important point. The latest research on levels of engagement in household chores and its impacts on children (from Young Lives) challenges the consensus that only ‘excessive’ levels of engagement are harmful, finding instead that any engagement becomes harmful when it crowds out either school or study time. We explain our focus on all levels of domestic work on this basis. We further justify our focus by arguing that we need to understand better factors conditioning levels of domestic work, amongst them the disability status of household members – and that inequalities in the distribution of household chores associated with markers such as sex (e.g. higher engagement by girls than boys) or the disability status of household members is inequitable.

Descriptive results are missing completely, so we have no way of understanding how many hours on average do children actually work? Either in all countries altogether or in separate countries. Please, present the descriptive results across the dependent and all independent variables.

Presentation of logs instead of hours muddles the text in terms of readability. 

Descriptive results are reported for all dependent and independent variables for the pooled dataset and presented in a new table (Table 2).

We agree with this comment. Rather than report the marginal change in log hours or raw hours, we have reported simply the percentage increase in hours associated with a change in each covariate.

As expected, there are country differences, but what they actually mean? And what do significant or nonsignificant results mean in term of previous research in the field?

We flag this as an area for future research – see response to Comment 15 below.

We highlight that this is the first cross-national analysis we are aware of that examines how children's unpaid workloads at home are affected by the disability status of their primary caregiver and younger children within the household in low- and middle-income countries. For this reason, we believe the main value of this paper is in highlighting the potential importance of considering the disability profile of household members for children’s time allocation. The revised paper makes this contribution clearer.

Comments

Reply

Please clearly present the specific results of the "Ordinary Least Squares" (OLS) method in the abstract. For example, explicitly mention what was revealed through the analysis.

Done

The introduction is very brief and does not adequately explain the rationale for conducting this research. Please provide a summary of previous studies on the relationship between disability and children's household chores in various contexts. Additionally, explain why this research is necessary and include the research questions and hypotheses. Overall, the introduction needs to be expanded and strengthened.

The paper is now better positioned in the relevant literature in two respects. First, the literature review in the introduction has been strengthened to include discussion of research, principally for HICs, on ‘young carers’ and effects of having a PWD in the household on their education, health and career prospects. There is limited research on young people providing care – including by performing domestic chores – outside of HICs, so this paper makes an important contribution in this respect.  Second, the literature review expands on why this study might contribute to the literature on the impacts of time spent on domestic chores for child wellbeing. It includes both references to the literature on ‘extensive involvement’ harming educational and health outcomes (on which the threshold for child labour is based) and on newer research which argues that the impact of involvement in household chores at lower levels too might harm child outcomes.

Independent and Dependent Variables: Provide a clearer description of how the variables were coded.

The clarification of the statement regarding children aged 2-4 is in the following sentence: ‘Children were defined as having a disability if [set of conditions]’. The theoretical rationale for the definition provided is added [UNICEF publication Seen, Count, Included, p. 20].

We agree this is confusing and have amended to provide more detail regarding the three binary measures.

This statement is unclear: "For children ages 2–4, we constructed one binary measure of disability." Please clarify. In addition, I do not find in  manuscript  how "disability status" was dichotomized. Which scale was used? What is the theoretical rationale for creating binary measures? Please further clarity.

The clarification of the statement regarding children aged 2-4 is in the following sentence: ‘Children were defined as having a disability if [set of conditions]’. The theoretical rationale for the definition provided is added [UNICEF publication Seen, Count, Included, p. 20].

The first paragraph in section "2.3 Statistical Analysis" should be moved to the end of the introduction section, as it is not related to statistical analysis.

This suggestion is adopted.

Overall, the results and tables presented are not entirely convincing. The presence of empty cells in the tables is confusing and raises concerns about the rationale for including the nations in the analysis. This, in turn, raises questions about the validity of the results, especially when combining multiple datasets.

The empty cells relate to non-statistically significant results. The results are not significant for all variables in three countries only – Guinea Bissau, Suriname or Zimbabwe. These three countries are removed from the table. Blank cells are replaced by ‘NS’ so that it is clear that the marginal effect is not statistically significant.

Results and Discussion: Currently, this section is not clearly separated. Please divide it into distinct "Results" and "Discussion" sections. The conclusion is also mixed with the discussion, and the paper needs to be restructured to ensure clarity and coherence.

This suggestion is adopted.

Comments

Reply

1. Winsorizing is not explained, although the reason for doing so is adequately stated. A plain English explanation of this process - for example it is done to mitigate the effect of outliers on the mean. It is an averaging method that involves replacing the smallest and largest (in this instance, the largest) values of a data set with the observations closest to these. 

We have added an endnote explaining the procedure.

2. The issue of whether children may have also had added UPDCW if older adults (with or without a disability) is substantive. The authors have opted, reasonably, to treat this as a limitation of the dataset but this does not address the likelihood of UPDCW by children 5-17 due to ageing caregivers. I suggest thinking through what steps could be taken to examine this substantive issue and reporting these as further research  needed.

A consideration of impact (and outcomes) of extra domestic and care work on young carers from the literature and via additional population research is urgently needed.

We have added a section on future research directions. Key elements are the following:

1.      We are hamstrung by having only a partial household disability profile – all women of reproductive age, all children aged 2-4, a subsample of men typically 15-49, one child aged 5-17 per household. To better understand intrahousehold dynamics connected to household chores, it would make sense for some population surveys to collect disability profiles of all household members (including older adults), alongside time use data, at least for a subset of households. Richer adult disability data would also add value.

2.      More work is needed on the impact of involvement in different levels of domestic work – following on YL work on four countries showing that any level of engagement in domestic chores can be harmful to children if it crowds out time spent on education. The MICS 7 inclusion of time diaries for children should provide data allowing untangling time spent on UCDW v other aspects of life (schooling / leisure etc.) for a large number of countries, thereby permitting a holistic understanding of how UCDW fits into other activities children perform.

3.      The diversity of country experiences (and the importance of country-fixed effects in the pooled regressions), highlights a need to understand better what factors within countries are leading to greater and lesser engagement in UCDW for children.

More work is needed on the number of young carers in LMICS, the level of care they provide, and the impacts of providing such care on various markers of their wellbeing.

3. The description "constructed three binary measures of disability" 102-106 is confusing. I suggest using the same description as placed on line 163 “the disability status of caregivers (a  binary variable and an indicator of severity)” which is more accurate than three binary measures.

We agree this is confusing and have amended to provide more detail regarding the three binary measures in both places.

4. More attention given to the protective factors would be helpful. For example not surprisingly given gender distribution of tasks, the number of adult women in the household could be related to older women past 49 years. Is this data available in MICS and if so could this data be used (whether disability status is available or not) to better inform the current findings.

It is certainly possible that the presence of older adults could be protective of the time children allocate to household chores. As noted in the same review, children may also allocate more time to household chores if older adults, with or without a disability, are present in the household. Older adults are, however, more likely to require themselves.

Because MICs does not give any information regarding the disability or health status of any older women in the household, we cannot predict a priori whether or not their presence is likely to substitute for the time children spend on UCDW or to increase demand for this work.

We introduced a variable that was the count of women in each household that were 50+ years old – the variable was not statistically significant. We interpret this as suggesting both that some older women may be providing care, and others may themselves require care.

5. The paper would benefit in the background and the discussion from some consideration of and reference to the literature on ‘domestic duties and care workload’ on child and young people carers of siblings or their primary caregiver/s with disability. This literature is equivocal on the impact of this on the health and wellbeing including educational outcomes for these young carers. This is a longstanding literature albeit primarily restricted to high income countries. However, an important one given that as this paper only addresses low-and middle-income countries and the protective factors may only be available for limited populations within those countries, the concluding sentence, although representing a human development and rights perspective may be unlikely to be achieved.

The paper is now better positioned in the relevant literature in two respects. First, the literature review in the introduction has been strengthened to include discussion of research, principally for HICs, on ‘young carers’ and effects of having a PWD in the household on their education, health and career prospects. There is limited research on young people providing care – including by performing domestic chores – outside of HICs, so this paper makes an important contribution in this respect.  Second, the literature review expands on why this study might contribute to the literature on the impacts of time spent on domestic chores for child wellbeing. It includes both references to the literature on ‘extensive involvement’ harming educational and health outcomes (on which the threshold for child labour is based) and on newer research which argues that the impact of involvement in household chores at lower levels too might harm child outcomes.