Next Article in Journal
What Are the Reasons for the Different COVID-19 Situations in Different Cities of China? A Study from the Perspective of Population Migration
Next Article in Special Issue
Workplace Health in Kentucky: A Statewide Comparison
Previous Article in Journal
Benefits of STRENOLD Program on Health-Related Quality of Life in Adults Aged 60 Years or Older. In Common Sport Study
Previous Article in Special Issue
Worker Perspectives on Incorporating Artificial Intelligence into Office Workspaces: Implications for the Future of Office Work
 
 
Article
Peer-Review Record

Work Flexibility and Work-Related Well-Being

Int. J. Environ. Res. Public Health 2021, 18(6), 3254; https://doi.org/10.3390/ijerph18063254
by Tapas K. Ray * and Regina Pana-Cryan
Reviewer 1: Anonymous
Reviewer 2: Anonymous
Reviewer 3: Anonymous
Reviewer 4: Anonymous
Int. J. Environ. Res. Public Health 2021, 18(6), 3254; https://doi.org/10.3390/ijerph18063254
Submission received: 17 February 2021 / Revised: 17 March 2021 / Accepted: 18 March 2021 / Published: 21 March 2021
(This article belongs to the Special Issue Worker Safety, Health, and Well-Being in the USA)

Round 1

Reviewer 1 Report

This paper pools all the available data from the GSS QWL modules, every 4 years, to conduct straightforward empirical tests re: the re: the relationship between four outcome variables of interest and three indicators of employees' flexible work conditions.

 

Highlights, Contributions:

1) Succinct, narrowed/focused scope

2) Pools all available data into giant cross section, latest available (up to 2018).

3) Focuses on work-related outcomes of interest to IJERPH readers: job stress, job satisfaction and two indicators of health

4) Clever application of zero-inflated binomial estimation, since # of days measures may be most frequently just zero.

5) Inclusion of "prior health" in estimations, since they predict current health and the other outcomes

6) Differentiating findings by potential gender differences.

7) Many Key findings of interest, with some surprises:

p.12: work at home exacerbates rather than alleviates job stress (see reference below, that would support that finding and why and under what type of reason for working at home);

pp.12-13: Daily Schedule flexibility improves well being indicators -- reducing stress and raising job satisfaction -- but only for men, stress is worsened for women. It is compounded by having either good or poor prior health.

Work at home has no association with health outcomes.

Flexibility during work day (time off) is more powerful than flexibility at margins of the work day.

 

Additional findings to report:

1) Year dummies, include or report if they were included.

2) Table 1, the %'s reporting "zero" # of days.

3) the significance test results of mean differences in the descriptives tables 1-4.

4) regression models explanations section 2.2.2 might be better placed by moving downward, just prior to reporting the results.

5) Results with independent variable being # of hours rather than "full time" vs. "part time." though motivated by desire to avoid heterogeneity with using only the former -- these 2 controls might enter independently or interact, for example, the # of hours might matter more for full timers, or perhaps vice versa, depending on which outcome.

References needed:

Berg, P., Kossek, E. E., Misra, K., & Belman, D. (2014). Work-life flexibility policies: Do unions affect employee access and use?. ILR Review, 67(1), 111-137. [developed 13 useful, detailed indicators of flexibility

practices in workplaces]   Kim, J., Henly, J. R., Golden, L. M., & Lambert, S. J. (2020). Workplace Flexibility and Worker Well‐Being by Gender. Journal of Marriage and Family, 82(3), 892-910. [focused on work-family and stress outcomes, distinctions between motivation for working at home, salaried vs. hourly workers, and also finds that "time off during" work has stronger positive outcomes than "daily schedule" adjustment.]  

4) Contrast to CPS module findings re: prevalence of work schedule flexibility, from the 2017-18 ATUS module, e.g., Table 1:

Work schedule flexibility(4)

 

Had flexible schedule

81,533 34,243 42.0 29,753 36.5 100.0 71.1 16.0 12.9

Did not have flexible schedule

62,762 7,328 11.7 5,971 9.5 100.0 45.2 48.5 6.3

5) Implications for other potential dependent variables of interest (e.g., subjective well being indicators, see recent Bell and Blanchflower, Hamermesh)

Author Response

First of all, Thank You, for your review comments. We find them very informative and constructive. We have modified the paper based on your (and others') comments. We believe that your comments were instrumental in adding value to our paper. Especially guiding us to the recent publication in this topic.

We hope we are able to address all your comments. Thank You!

 

1) Year dummies, include or report if they were included.

We have not used year dummies. We used individual years to look at the prevalence, but to add power to our statistical analysis, we pooled the data used in the rest of the analyses. The data are not longitudinal, so we have not compared variations by year. We have noted that in the limitations.

2) Table 1, the %'s reporting "zero" # of days

We are not sure we understand this point. Table 1 shows the percentage of the population distributed across the various demographic categories for each flexibility element.

3) the significance test results of mean differences in the descriptives tables 1-4.

We have added the covariance significance in the table. Thank you!

4) regression models explanations section 2.2.2 might be better placed by moving downward, just prior to reporting the results.

The paper has been modified based on all the reviewers' comments.  The regression models are explained just before the 'results' section.

 

5) Results with independent variable being # of hours rather than "full time" vs. "part time." though motivated by desire to avoid heterogeneity with using only the former -- these 2 controls might enter independently or interact, for example, the # of hours might matter more for full timers, or perhaps vice versa, depending on which outcome.

Thank you! We added hours of work per week as a covariate. The part /full-time status was not adding to the significance of the results. Our results are revised, and the associations have improved. We think this has made a significant improvement over our earlier version. The tables are all modified.

References needed:

Thank you again! We had added these references. While the first paper is not directly related to what we are doing, it pointed us to a valuable literature search. The second paper was significant to consider. Our current version referred to this in both the introduction, results and the discussion section.

4) Contrast to CPS module findings re: prevalence of work schedule flexibility, from the 2017-18 ATUS module, e.g., Table 1:

Thank you! We missed the flexible schedule numbers in our introduction part, where we are describing existing prevalence. They are now added in the current version.

5) Implications for other potential dependent variables of interest (e.g., subjective well being indicators, see recent Bell and Blanchflower, Hamermesh)

In the revised draft, we have mentioned this as a limitation of the study. The QWL has limited measures of subjective well-being outcomes. Although, in literature, previous studies find job satisfaction to be positively correlated to subjective well-being.

 

Reviewer 2 Report

Dear authors,

Thank you very much for your efforts writing this paper.

This manuscript studies work flexibility and work-related well-being in the United States of America. The topic of the paper is interesting and relevant.

The analysis is based on descriptive statistics and regression analyses of USA general social survey-quality of work life data from 2002 to 2018.

The main contribution of this paper is to demonstrate the importance of work flexibility for well-being.

The most relevant strength of this investigation are the descriptive analyses and the regression analyses.

The following minor comments aim to improve the quality of this paper. I primarily recommend addressing:

C1. Please compare your results with other similar studies related to USA and to other countries (maybe results are different depending on socio-cultural issues).

C2. Lines 12, 204, etc.: “we assessed”, “We used”, “We assumed”, “etc. Please avoid the use of first-person pronouns and apply an impersonal writing style.  

C3. It is recommended to explain the manuscript organization at the end of section 1.

C4. Please include research question/s.

C5. Please discuss your research limitations.

C6. Lines 82, 86, 174, 180, etc: Please use references instead of using links.  

C7. Line 88: Please clarify that data is related to the USA.

I congratulate the authors for this interesting investigation and wish them the most success in their research activities.

Thank you very much for your efforts and for your valuable scientific contribution.

Author Response

First of all, Thank You, for your review comments and kind words! We find them very informative and constructive. We have modified the paper based on your (and others') comments. We believe that your comments were instrumental in adding value to our paper. 

We hope we are able to address all your comments. Thank You!

C1. Please compare your results with other similar studies related to USA and to other countries (maybe results are different depending on socio-cultural issues).

We have revised the discussion section to address this point. We hope you will like it.

C2. Lines 12, 204, etc.: “we assessed”, “We used”, “We assumed”, “etc. Please avoid the use of first-person pronouns and apply an impersonal writing style.

  

Our organization (NIOSH) strongly recommends using active voice in an overall effort to improve "plain language." To address your comment, we tried to make some changes throughout by referring 'to the study' rather than 'us.'

C3. It is recommended to explain the manuscript organization at the end of section 1.

Thank you for the recommendation. We have added a sentence explaining the organization at the end of section 1.

C4. Please include research question/s.

We added our research questions in the second last paragraph of our introduction section. 

C5. Please discuss your research limitations.

We had included a paragraph on limitations in the discussion section. In the current version, we have expanded on the section.

C6. Lines 82, 86, 174, 180, etc: Please use references instead of using links.  

We have replaced the links with the appropriate references.

C7. Line 88: Please clarify that data is related to the USA.

Done. We have mentioned it in our current version.

Reviewer 3 Report

Dear Authors!

Thank you for taking up a very interesting topic of  Work Flexibility and Work-Related Well-Being.
It is worth exploring this topic, especially in today's difficult pandemic times.


The article is valuable overall but suffers from several shortcomings. The list of issues to be improved is presented below:

  • The aim of the research should be clearly defined both in the abstract and in the text itself - in the introduction section. As it stands, the text looks more like a research report than a research paper
  • There are too many keywords. Some of them do not even appear in the text itself (e.g. total worker health; organizational design; technology)
  • It is worth setting a clear hypothesis or hypotheses and then confirming or rejecting it as a result of research. If the authors do not want to put forward hypotheses, it is worth emphasizing the descriptive nature of the study and indicating it. Currently, the authors point to the hypothesis / lines 206 and 468 / but it is not clear what they really mean. This is a bit confusing for the reader
  • As IJERPH is a scientific journal, it is worth referring to scientific sources to a greater extent. Many of the indicated source items are only research reports. I suggest extending the literature review


I encourage you to improve the article because it is an interesting voice in research on work flexibility.

In the future, it is also worth continuing this interesting study basing on newer data (from years 2019-2020), including data from the period of time of which many workers were ‘forced’ to conduct remote work during the era of COVID-19 pandemic.


Best regards,
The reviewer.

Author Response

Thank You, for your constructive comments. Yes, we agree with you mostly, and have revised the paper including a broadened literature search. Our aim is to carry this research forward with 2022 data when available. However, we are undertaking similar COVID related research with BLS data from 2020. 

We hope that we have addressed your comments properly and that you will like our current version.

Sincerely

 

  • The aim of the research should be clearly defined both in the abstract and in the text itself - in the introduction section. As it stands, the text looks more like a research report than a research paper

We have added research questions in the text. We did not make changes in the abstract. The abstract format is consistent with that of abstracts in other related publications that conducted regression analyses. We have added our research aim/questions in the second last paragraph of our introduction section.

  • There are too many keywords. Some of them do not even appear in the text itself (e.g. total worker health; organizational design; technology)

We deleted total worker health and technology and revised organizational design to work organization.

  • It is worth setting a clear hypothesis or hypotheses and then confirming or rejecting it as a result of research. If the authors do not want to put forward hypotheses, it is worth emphasizing the descriptive nature of the study and indicating it. Currently, the authors point to the hypothesis / lines 206 and 468 / but it is not clear what they really mean. This is a bit confusing for the reader

We added our research questions in the second last paragraph of our introduction section. Using research questions instead of hypotheses is consistent with other related publications that conducted regression analyses. For example, see Ray, T.K.; Kenigsberg, T.; Pana-Cryan R. Employment arrangement, job stress, and health-related quality of life. Safety Science, 2017, https://doi.org/10.1016/j.ssci.2017.05.003

We have changed 'hypothesis' to 'assumption' (current line 222) and deleted the reference to 'hypotheses' in line 468 in our previous version.

  • As IJERPH is a scientific journal, it is worth referring to scientific sources to a greater extent. Many of the indicated source items are only research reports. I suggest extending the literature review

We have broadened our literature search and have added new references in the introduction and the discussion sections to address your point. However, much of the recent data and findings on this particular topic are in reports (from BLS, EUROFOUND, ILO, and etc.) that we have cited.

Reviewer 4 Report

Overall, a very interesting paper and nicely done. I know that all papers aren’t about COVID-19, but as I was reading, I was left thinking about the impacts of the pandemic on the future of flexible work and the participation of women in the workforce. May be something worth noting in the intro or discussion, though I certainly understand if the authors choose not to.

 

Line 34 – recent data to cite regarding workforce participation of women declining due to COVID?

 

Lines 68-69 – sentence does not make sense as written

 

Lines 88-99 – found this paragraph difficult to read. Would recommend that the sentence be broken up into smaller ideas to make it easier to digest that the study is examining.

 

Lines 105-106 – I believe they are more accurately referred to as NORC at the University of Chicago (I believe they dropped the name behind the acronym)

 

Throughout the indicators of flexibility and well-being, I was left wondering about the strategy of dichotomizing variables and how the breaks were decided, particularly for those utilizing a 5 point scale. In some cases, for example, the stress question, the middle value of sometimes was classified as not satisfied. For overall health, the middle option of good was classified into the positive category. Was there a justification for why they were split and if so, can you provide it?

 

Lines 167-175 – even though there was a citation for how healthy days was calculated, I was left wondering why the authors would combine the two variables. I think information is lost as to the physical vs mental health outcome. Additionally, if someone reports 15 unhealthy physical and 15 unhealthy mental days, you would assign them 30 unhealthy days. Is it possible that the poor physical days are one and the same as the poor mental health days? Might flexibility impact these differently?

 

Line 239 – what are psychosocial working conditions?

 

Line 279 and Table 1 – less than 1% of the sample was in the less than 8 years of education. Is there a case for collapsing the first 3 lines in the table under education into one category? Would that affect your results?

 

Table 1 – since you dichotomized health status, make the label correspond to how your previously described it – “good” not “good or better”

 

Table 2 – based on how it is reported here, I see that you do have physical and mental health separate. Clarify in lines 167-175 how this variable is included.

 

Lines 330-332 – when you say women were 18% more likely to report job stress, etc, is that for women who are able to work from home?

 

Lines 332-336 – can you also provide the % for the other findings you highlight to help with interpretability?

 

Line 352 – same comment as above

 

Tables 5-7 – what is the rationale for presenting the 10% significance (vs 5% and 1%)

 

Lines 362-364 – please review, sentence doesn’t make sense as written

 

Lines 390 – you discuss work from home as remaining the same and for the others report the % at the last time point. They are all pretty stable across time and I think you could call that out.

 

Line 401 – change being job satisfied to job satisfaction

 

Lines 400-420 seem to be rehashing results instead of adding context

 

Lines 437 – should this say mixed well-being outcomes associated with flexible work? It’s not just work at home

 

Line 451 – remove “of”

 

Line 467 – missing beginning of sentence

Author Response

First of all, Thank You, for your kind words. We found you review comments constructive. Based on them, and the comments from others, we have modified the paper. We believe the current version is more exhaustive and much improved. Hopefully we have addressed all your comments. Our replies are in bold faced.

Sincerely

 

 

Overall, a very interesting paper and nicely done. I know that all papers aren’t about COVID-19, but as I was reading, I was left thinking about the impacts of the pandemic on the future of flexible work and the participation of women in the workforce. May be something worth noting in the intro or discussion, though I certainly understand if the authors choose not to.

We believe you have raised an important point. We are analyzing COVID related BLS data now. For this paper, it is beyond our scope. Also, the GSS data 2020 is not yet out. However, we cited and mentioned recent labor force statistics from BLS, without mentioning COVID directly. We also acknowledged the importance of our findings in the light of recent increase in remote work in the introduction and the discussion sections.

Line 34 – recent data to cite regarding workforce participation of women declining due to COVID?

We are not sure there is any national-level data source to substantiate that claim unequivocally. BLS numbers show that women (16years+) are working more from home than men, and a slightly lower proportion of them lost their jobs compared to men. However, for women (20+), a Gallup analysis shows there was a 0.2 percentage points gap (-ve) in labor force participation among women. We cited that in the revised text.

 

Lines 68-69 – sentence does not make sense as written

Thank you! We edited this sentence in the new version.

Lines 88-99 – found this paragraph difficult to read. Would recommend that the sentence be broken up into smaller ideas to make it easier to digest that the study is examining.

Thank You! We have broken down the long sentence into smaller ones for convenience.

Lines 105-106 – I believe they are more accurately referred to as NORC at the University of Chicago (I believe they dropped the name behind the acronym)

We added the abbreviation NORC. We had previously spelled out the name but had not included the acronym.

Throughout the indicators of flexibility and well-being, I was left wondering about the strategy of dichotomizing variables and how the breaks were decided, particularly for those utilizing a 5 point scale. In some cases, for example, the stress question, the middle value of sometimes was classified as not satisfied. For overall health, the middle option of good was classified into the positive category. Was there a justification for why they were split and if so, can you provide it?

Thank you for raising this question. We dichotomized the outcome variables and the flexibility variables to facilitate the interpretation of the results.  We ran the regressions without dichotomizing, keeping the variables in their original Likert scales to verify our results' robustness. We presented these results at international conferences. However, we decided to show our dichotomized results in the article. We followed the following strategy:

For the stress question, How stressful is your work? We converted the 5-point Likert scale responses into:

Stressed => always, often,

Not Stressed => sometimes, rarely, and never

This was based on the ongoing NIOSH practice of reporting job stress prevalence in the United States since 2002. Responses 'always' and 'often' are combined to document the prevalence of job stress during 2002 - 2018. This also correlates with positive responses to high job demand and low job control items in the survey. Further, as literature points out, e.g.,  Seery and Silver (2010)[1], certain aspects of stress, especially those of shorter duration, are beneficial for developing resilience and enhancing productivity. Hence those who reported stress 'sometimes' are considered not stressed in this and other NIOSH studies, e.g., Ray et al., 2017[2].

Dichotomizing the outcome variable, job satisfaction, is intuitive. The question asks, All in all, how satisfied would you say you are with your job?

The responses were:

Satisfied => very satisfied, somewhat satisfied

Not Satisfied => not too satisfied, not at all satisfied

We used a similar approach with the variable 'overall health status' and literally dichotomized it at 'good.' Fair is lower ordered than good, and hence we classified it as 'not good.' The question asks

Would you say that, in general, your health is excellent, very good, good, fair, or poor?

Health status good => excellent, very good, good

Health status not good => fair, poor

Also, this variable enters our regression model as a control variable, and the placement of the break would notchange the relationship between flexibility variables and the well-being outcomes. It would have changed the interpretation of this particular variable only.

 

[1] https://webfiles.uci.edu/rsilver/Seery,%20Holman,%20&%20Silver%202010%20JPSP.pdf

[2]  https://doi.org/10.1016/j.ssci.2017.05.003

Lines 167-175 – even though there was a citation for how healthy days was calculated, I was left wondering why the authors would combine the two variables. I think information is lost as to the physical vs mental health outcome. Additionally, if someone reports 15 unhealthy physical and 15 unhealthy mental days, you would assign them 30 unhealthy days. Is it possible that the poor physical days are one and the same as the poor mental health days? Might flexibility impact these differently?

Our objective in this study is to assess the association of flexibility with Health-related quality of life (HRQL). There are many measures of HRQL. One such measure is the 'Healthy Days' developed by CDC. The purpose of HRQL is to measure health over multiple domains. Two such broad domains are physical and mental health. So, both physical and mental days are combined to obtain the comprehensive health status measure. Alike any HRQL measure, 'healthy days' has limitations. But, GSS-QWL provides limited information for estimating HRQLs. Hence 'healthy days' is the most suitable and feasible measure.

However, like you, we wondered about the partial effects of flexibility on mental and physical days. The results are more pronounced on mental health than physical health. Below, we present the coefficients from our linear regressions with the physical and mental days separated. In the actual paper, we used zero-inflated negative binomial regressions.

Our objective is to utilize a well-known measure of HRQL, and healthy days is such a measure. It does have its limitations[3] but also its strengths. Its internal consistency has already been tested. Studies show that it compares favorably with other measures of HRQL.

 

Healthy Days

Days with sound mental health

 

Days with sound physical health

Take time off

1.46***

1.28***

0.61***

Change schedule

0.03

0.07

0.04

Work at home

0.54*

0.59**

0.3

 

 

[3] https://www.cdc.gov/hrqol/surveillance.htm#:~:text=The%20CDC%20Healthy%20Days%20measures%20are%20an%20acknowledged,2002%20when%20only%2020%20states%20asked%20these%20questions%29.

 

Line 239 – what are psychosocial working conditions?

We added examples and a reference (that was already included on our list) for more information the first time we mention "psychosocial working conditions in new line 125. (Grosch, J.W.; Caruso, C.C.; Rosa, R.R.; Sauter, S.L. Long hours of work in the US: Associations with demo-graphic and organizational characteristics, psychosocial working conditions, and health. American Journal of Industrial Medicine, 2006, 49(11), 943-952, doi: 10.1002/ajim.20388.)

Line 279 and Table 1 – less than 1% of the sample was in the less than 8 years of education. Is there a case for collapsing the first 3 lines in the table under education into one category? Would that affect your results?

We don't anticipate that collapsing the first 3 lines under education would significantly change the results, and we kept the original breakdown.

 

Table 1 – since you dichotomized health status, make the label correspond to how your previously described it – “good” not “good or better”

Thank you! We have made the appropriate changes.

Table 2 – based on how it is reported here, I see that you do have physical and mental health separate. Clarify in lines 167-175 how this variable is included.

Please see the response to the comment above. We presented the variables separately in our descriptive analysis, but we used Healthy Days, a broadly used HRQL construct, to understand the association of flexibility with well-being.

Lines 330-332 – when you say women were 18% more likely to report job stress, etc, is that for women who are able to work from home?

No, this refers to women in our total sample

Lines 332-336 – can you also provide the % for the other findings you highlight to help with interpretability?

Yes, we have done this in our current version. 

Line 352 – same comment as above

Done

Tables 5-7 – what is the rationale for presenting the 10% significance (vs 5% and 1%)

We are presenting all three levels of significance. This will help us compare with our other studies based on QWL.

Lines 362-364 – please review, sentence doesn’t make sense as written

We have made revisions to our current version. 

Lines 390 – you discuss work from home as remaining the same and for the others report the % at the last time point. They are all pretty stable across time and I think you could call that out.

Thank you! We have added a sentence that clarifies this point.

 

Line 401 – change being job satisfied to job satisfaction

Done. We have checked for the same throughout.

Lines 400-420 seem to be rehashing results instead of adding context

We have made changes to the discussion section. We hope you will like the current version.

Lines 437 – should this say mixed well-being outcomes associated with flexible work? It’s not just work at home

We revised the sentence to reflect this.

Line 451 – remove “of”

Done.

Line 467 – missing beginning of sentence

We have changed the section and modified the paragraph by deleting the sentence.

Round 2

Reviewer 3 Report

Dear Authors,

Thank you for referring to my comments. I am pleased to recommend the text for publication.

The reviewer.

Back to TopTop