[Futuregen] Widowhood paper: Rejected by Journal of Gerontology

[Stefania Ilinca]

Dear all,

I keep bringing bad news after important holidays - sorry about that. 
Unfortunately, the widowhood paper was rejected by the Journals of 
Gerontology, but this time both the reviewers and the editor have 
provided detailed and useful comments - see copied below. None seem 
insurmountable at a first read.

I look forward to discussing this once you would have had a chance to 
read through the comments. In the meanwhile, I will prepare a plan for 
re-submission (including planned changes to the manuscript and the 
analyses, as well as a few target journal proposals).

My best regards, Stefania

From:

JGeronSoc at Geron.org

To:

ilinca at euro.centre.org

CC:

Subject:

Journal of Gerontology: Social Sciences - Decision on Manuscript ID 
JGSS-2021-012

Body:

Dear Author,

On 12-Jan-2021 we received your  manuscript entitled Gender Differences 
in Access to Community-Based Care: A Longitudinal Analysis of the 
Effects of Widowhood and Living Arrangements for possible publication in 
the Journal of Gerontology: Social Sciences. The thoughtful comments of 
the reviewers are enclosed for your consideration.

The reviewers are very accomplished in their research and knowledgeable 
of the topic of your manuscript.  I am quite impressed by the careful 
and constructive reviews that they have submitted.

Although the reviewers value the topic of your study, they raise a 
number of fairly serious concerns about your manuscript. The reviewers' 
comments are thorough and quite explicit.  After carefully considering 
the judgment of the reviewers, I am sorry to write that we will not be 
able to publish your paper in the Journal of Gerontology: Social Sciences.

To put this editorial decision into perspective, let me add that the 
Journal accepts less than 20% of all manuscripts submitted.

When I reread the paper with the reviewers' comments in mind, I had many 
of the same reactions: the topic is very interesting, and the paper has 
great potential. Probably the most significant limitation derives from 
the study design.

This is a study using SHARE to examine the risk of unmet care needs 
among widowed men and women. Findings indicate that women who transition 
into widowhood are more likely to receive care than men.  Living alone 
and widowhood have independent influences on likelihood of receiving 
care.  As population ages in Europe, the degree to which functional 
needs are being met in the community is an important line of inquiry.

The reviewers expressed enthusiasm for the general question, but pointed 
out a number of weaknesses with the present manuscript, both 
conceptually and methodologically.  In my own reading of the manuscript, 
I concur. I will not reiterate all of their points here but will note 
some of the key ones that drove this decision. Foremost, the study is 
grounded in a very delimited question, attempting to parse the influence 
of widowhood from living arrangement on receiving care.  Conceptually, 
bereavement, widowhood, and living arrangement are not clearly 
separated. Methodologically, the analyses test simultaneously for 
effects of widowhood, health needs, living alone, and household size 
with little attention to their measurement overlap. Second, the analyses 
did not provide formal contrasts by gender even though the results were 
interpreted in this way.  Third, the statistical model may be a novel 
contribution to the literature but this needs further elaboration for 
the reader as to its strengths over traditional fixed and random effects 
models.  It seems that indicators of model fit are missing, as well, 
making it difficult to evaluate impact of the findings.  These are not 
insurmountable critiques but careful attention to them would likely 
yield a completely different manuscript.  I hope that the feedback is 
helpful as you revise your manuscript for another journal such as 
Research on Aging.



I wish that this letter could have brought you better news, and I thank 
you for considering the Journal of Gerontology: Social Sciences. I hope 
that you will continue to submit your work to us in the future.

I also encourage you to exploit the thorough and constructive comments 
of the reviewers. They felt that the paper has promise, and I am sure 
that their comments will help you in revising the paper to fulfill that 
promise.

Best regards,
Jessica Kelley,  Editor



Reviewer(s)' Comments to Author:

Reviewer: 1

Comments to the Author
This manuscript examines gender differences in getting help among older 
adults, and trying to separate out the effects of widowhood and living 
arrangements.

The premise: Although this topic of older adults needing care is an 
important one, I'm not sure I understand this way of approaching the 
question.  Why do we need to separate out the effects of living alone 
and widowhood? How does this help us make sure that people who need care 
get it?

There are some vague terms used and I'm not sure that I followed. For 
example, "fairness" in European long-term care systems.

I'm not sure if it might be useful to look at why people are not 
receiving care?  Is it because they don't want it, or because they can't 
get it, or can't afford it?  If you're interested in issues of access, 
then affordability might be a dimension to examine.

If you want to stick with this question, try to explain what we gain 
from separating out the effects of widowhood vs living arrangements.  
I'm not sure that that case was made.

Reviewer: 2

Comments to the Author
This study examines the impact of gender, widowhood, and living 
arrangements on the use of home and community-based care among older 
adults. The study demonstrates the important influence of gender, 
particularly in the use of informal and formal health care settings 
among the widowed population. The study is well written and uses 
rigorous data analysis, but I still have questions about its approach to 
separating widowhood from bereavement.
1) On page 4 lines 26 – 40. The study defines widowhood as “a long-term 
and ongoing state” while bereavement is a short-term experience of the 
death of a spouse. However, the study also states both widowhood and 
bereavement have “immediate and long-lasting effects”. These two 
statements contradict each other. If bereavement has lasting effects, 
then it is widowhood. Could the author clarify the definition of 
widowhood and bereavement and how to use them in the study?
2) There are three independent variables in the study: sex, widowhood, 
and living arrangements. The study mentions that “A binary variable that 
identifies widowhood was generated based on self-reported marital status 
in each panel wave – married living with a spouse, married not living 
with a spouse or widowed” (page 9 line 36 – 40). It is not clear how the 
study measures “bereavement”. Since separating widowhood and bereavement 
is a significant contribution of the study to the field, the study needs 
to tell readers how to make it.
3) In tables 1 -3, the study also has variables for “transitions into 
widowhood” and “transitions out of widowhood”. How to define these two 
variables and why they are important in this study?
4) The study mentioned that due to the limited number of 
regional-specific analyses by
“the size of the longitudinal sample in SHARE” (p15, line 47-50). 
However, according to Table 1, there were 7554 women in the continental 
area, 2553 women in the Nordic region, 5874 women in the Southern 
region, and 5991 women in the Eastern region. This study seems to have 
enough people to analyze the regional impact on people's use of home and 
community care. It is well known that local health policies and 
culturally based information care systems can significantly influence 
people's behaviors of trust and use of informal and formal health care. 
As the study mentions, support for widowhood is also gender-based and 
influenced by local culture and policies. Since these regional data is 
available for this study, could the authors tell us how the region 
influences these gender-based behaviors? Maybe a moderation effects 
analysis can help.


Reviewer: 3

Comments to the Author
The present manuscript presents a cross-national analysis of thousands 
of older adults from across Europe. The manuscript aims to “explore the 
complex pattern of associations between widowhood, living arrangements, 
and use of home and community-based care”. I think this is a worthy 
topic of study and the authors effectively convinced me of the 
importance of the research questions. At the same time, I think there 
are some fundamental problems with the ways that some of the analyses 
were conducted and presented. I provide more detailed comments below in 
the hopes of helping to strengthen a future iteration of this important 
and interesting manuscript. I also want to couch my comments with the 
caveat that while I feel comfortable conducting and interpreting 
multilevel models, REWB models are new to me and I apologize for any 
misunderstandings that occurred because of my ignorance.

Major comments:
1) The second objective, it seems to me, is concerned with the 
interaction between a) sex and bereavement (within effects) and b) the 
interaction between sex and widowhood (between effects), but my 
understanding of the analyses is that no interactions were tested. 
Instead, statistically significant effects for one sex (usually females) 
along with statistically non-significant effects for another sex 
(usually males) were treated as if they indicated a sex difference. 
Likewise, the third objective states that it wants to, “…reflect on how 
disadvantages in [financial and human capital] can intersect with sex to 
influence patterns of [HCBS] use”. Here interactions of interest appear 
to be those between sex and the stated dimensions of capital.
1a. Here are some other statements that lead me to think the manuscript 
was concerned with, and interpreted results as if there were, 
interactions, despite not testing interactions: “Our results further 
confirm previous findings of a differential effect of socio-economic 
status indicators across sexes”; “We next turn our attention to the 
possibility that the above results are moderated by country specific 
characteristics and institutional factors that are not fully captured by 
country specific dummy variables.”; “This suggests older widows are 
better able to access care resources following bereavement, whereas 
older widowers are less likely to do so”; “We found socio-economic 
status indicators were associated with care use only for older women, 
suggesting a modifying effect of sex.”; “More specifically, we find 
gender differences in the effect of widowhood and living arrangements on 
probability to receive long-term care are attenuated in countries that 
emphasize defamilialization (of care) through public provision, shifting 
responsibility for care from the family towards the state (Saraceno, 
2016: 317).”
2) It seems that the analyses for males in Table 3 are almost definitely 
underpowered (because the 168 bereavement events are spread across 4 
countries) and that this may also be the case for females (especially if 
there is disproportionate sampling or disproportionate observation of 
bereavement events across the welfare clusters). I do not believe there 
was a discussion of power in the manuscript but given that there were 
some large effects in Table 3 that did not achieve statistical 
significance, this seems to support the contention of insufficient power.
3) In this study the primary outcome is a binary indicator comprising 
any care participants have received, combining informal and formal care 
as well as different quantities of care. I think the authors adequately 
address the lack of quantity information in the discussion. I am less 
satisfied with the rationale for combining formal and informal care. 
This seems like an important question that future readers of the 
manuscript will be curious about. Perhaps a future iteration of the 
manuscript might disaggregate formal and informal care. I am sensitive 
to the need to fit within restrictive word limits, so I would completely 
understand if such a sensitivity analysis ends up in supplemental 
materials along with a stronger rationale for the aggregation of formal 
and informal care in the main text.
4) “both widows and widowers have a higher probability of receiving care 
than older individuals with partners” Is it possible that this suggests 
that there is a problem in measurement wherein people underreport the 
care provided by their spouses? If one has needs, and someone is in the 
home with them, are they not likely to receive the necessary support? 
According to the manuscript, it seems to be the case that live-in 
partners provide support, especially when carers are female (e.g., 
“Finally, the surviving spouses in our study (i.e. the widows and 
widowers) were likely previously providing informal care to their 
deceased spouse. There is evidence that households where the wife is the 
informal carer tend to receive less formal care service due to 
persistent gender stereotypes of women's roles as caregivers (Schmidt 
2017, Larsson et al. 2014). I would like to acknowledge that I am not 
saying there is a measurement problem, but I would be interested to hear 
the authors’ thoughts on the matter and why they think there is evidence 
for/against such an error.
5) Overall there is extensive focus on statistical significance with 
relatively little focus on effect sizes. This seems particularly 
problematic because the analyses for males are so underpowered in some 
circumstances (e.g., the rather sizeable effect of bereavement for males 
in Eastern countries, which is the third largest effect in Table 3 and 
is in the opposite direction of most of the findings). I’m not 
advocating for interpreting null effects, but I do think statistical 
significance and effect size should carefully be considered in concert.
6) The consistency of the findings across the four models in Table 2 and 
the four types of countries in Table 3 was neat to see! I was especially 
surprised to see that was the case in model 4 since the variables 
included in that model were likely highly correlated with 
bereavement/widowhood effects. Bravo to the authors for testing and 
reporting that model (and also presenting the others, which likely have 
fewer issues with collinearity).


Minor Comments:

1) “The care literature has overwhelmingly focused on intergenerational 
support patterns and caregiving by children to older parents,” this 
seems like a tough statement to backup. There is a vast literature on 
caregiving focused on spouses. Indeed, in the bereavement-care 
literature, I would argue that spouses are focused on to a larger extent 
than adult-children because some studies suggest (and lay-wisdom 
overwhelmingly suggests) that the adverse consequences of bereavement 
for former caregivers is greater than that for adult-children.
2) The manuscript mentions excluding waves 3 and 7, but it was unclear 
why. Did these waves not include necessary data?
3) I believe Table 1 presents (for the most part) observation N’s as 
opposed to participant N’s (I believe transitions into and out of 
widowhood are likely participant N’s, though I guess they may not 
necessarily be). I wonder if there might be an interesting way to 
incorporate participant N’s in this table along with observation N’s. 
For instance, it seems like an interesting data point to include how 
many participants received care at any point in the study versus the 
number who never received care. Based on my reading of the results 
section, it seems that there is interest in this level of analysis and 
that much of the first paragraph of the results section interprets the 
findings of Table 1 as if it presented results at the individual level. 
It seemed especially surprising to me that variables like “low 
education” were presented at the level of observations, given that such 
a variable is probably not very likely to change over time for this 
older sample (though it would be neat to know if it did change for the 
sample).
4) Terms like gender and sex seem to be used interchangeably in this 
manuscript, though my impression based on the method section is that 
“sex” is what was actually measured. I therefore tried to confine my 
terminology to sex and the associated categories of it that I believe 
were measured here (except for quotes from the manuscript).