Gender disparity is prominent in academia, notably in science and engineering (Ceci et al., 2014). In the United States in 2019, for example, only 16% of scientists and engineers were women (National Center for Science and Engineering Statistics, 2021). Women are also underrepresented across all professorship ranks (Chesler et al., 2010; Mason et al., 2013), especially among full professors – in 2018 only 32.5% of full professors in the United States were women (Colby and Fowler, 2020). Studies show that, compared with men, women academics produce fewer papers (Larivière et al., 2013; Paul-Hus et al., 2015), receive fewer citations (Caplar et al., 2017; Maliniak et al., 2013), and have narrower collaboration networks (Ductor et al., 2021). Women are also less likely to receive project funding (Ley and Hamilton, 2008; Witteman et al., 2019) or prestigious awards (Lunnemann et al., 2019; Meho, 2021). These gendered differences are usually correlated and mutually reinforcing, contributing to lower career satisfaction and higher attrition rates for women in academia (Xu, 2008). Understanding variables and mechanisms underlying the gendered disadvantage for women is critical for achieving gender equality in academia.

Parenting is considered highly related to gender disparities in academia (Hunter and Leahey, 2010; Kelly and Grant, 2012; Morgan et al., 2021; Powell, 2021). It contributes to the gender gap in many key research performance metrics, such as scientific productivity (Mairesse et al., 2019; Morgan et al., 2021), citations (Lawson et al., 2021), and academic collaboration (Hunter and Leahey, 2010). Evidence also shows that career satisfaction in academia is lower for mothers than fathers (Beckett et al., 2015), likely due to the reasons such as limited opportunities for tenure and promotion (Finkel and Olswang, 1996; Mason et al., 2013), lower salary (Kelly and Grant, 2012), and higher levels of work-family conflict (Martins et al., 2002). The prominent motherhood penalty could impel women to self-select away from academic careers and to leave such careers at higher rates than men (an effect known as the “leaky pipeline”: Anders, 2004). Yet, our present knowledge about parenting and gender disparities in academia has been shaped primarily by studies that are university-wide (e.g., Misra et al., 2012), monodisciplinary (e.g., Beckett et al., 2015), and thus of limited sample scales. The size, type, and mechanism of the association between parenting and academic achievements need further examination on a larger scale.

One potential reason parenting contributes to the gender disparity in academic careers is the work-family conflict. Workplace and family are both demanding in terms of time, energy, and loyalty, thus likely interfering with each other (Fox et al., 2011). Work-family conflict can be categorized into three types: (i) time-based conflict is the competition of time between different roles, e.g., mother vs. researcher; (ii) strain-based conflict happens when the strain in one role interferes with one’s ability to perform in another role; (iii) behavior-based conflict happens when the behavior requirements as a role become incompatible with the behavioral expectation of another (Greenhaus and Beutell, 1985). Studies have indicated that women usually experience higher levels of work-family conflict than men due to parenting (Martins et al., 2002; van Daalen et al., 2006; Wang et al., 2020), and academia is no exception (Fox et al., 2011). Mothers who are academics spend more time on childcare responsibilities and household tasks (Misra et al., 2012) and experience more research pressure and job stress than men (van Daalen et al., 2006). It is indicated that high work-family conflict can significantly affect women’s career satisfaction, regardless of their age (Martins et al., 2002). However, few studies have focused on how varying levels of work-family conflict across parenthood status contribute to the gender gaps in academia regarding different forms of career achievements.

The role of family support should also be considered. Family has become “the newest battlefront in the struggle for gender equality” (Hauser, 2012), enduring and consequential for well-being across the life course (Thomas et al., 2017). Family support, especially partner support, represents a significant form of family-to-work enrichment and has positive effects (Greenhaus and Powell, 2006; Kinnunen et al., 2006). Partner support can be instrumental or emotional (Adams and Golsch, 2021; Greenhaus and Powell, 2006) and may decrease work turnover rate and boost women’s perceptions of gains in productivity and career satisfaction (Ferguson et al., 2016; Juraqulova et al., 2015; Watanabe and Falci, 2016). Partner support can also relieve the tension inside the family and reduce the pressure from the family, mitigating work-family conflict and increasing well-being and career performance (Dickson, 2020; Ferguson et al., 2016; Thorstad et al., 2006; Wang et al., 2020). Although partner support has been found to help reduce work-family conflict – with the reduction for mothers being greater than that for fathers – the demand for partner support from mothers is often less satisfied (Adams and Golsch, 2021; Dickson, 2020). Despite previous discussions on the role of partner support in family-to-work enrichment, limited attention has been paid to the relationship between partner support and the gender disparity in academia.

Using data from the Web of Science and responses from 7,764 academics in the United States and Canada to a survey distributed in 2019 (see Supplementary file 1), this study examines the gender gaps in various measures of academic achievement (both objective and subjective), and how parenting-related work-family conflict and partner support mediate these gaps. Objective career achievements are observable, socially recognized indicators signaling one’s human capital values (Valcour and Ladge, 2008). We use bibliometric indicators of scientific productivity, citation, and collaboration to measure objective career achievements (see Materials and materials). Subjective career achievements are one’s subjective feelings of career attainments (Ng et al., 2005), including self-reported research satisfaction, career satisfaction, and the perceived recognition by scholarly communities in this study.

Based on self-reported gender identification, the sample contains 4,425 (57.0%) women, 3,311 (42.7%) men, and 28 respondents (0.4%) who self-identified as non-binary. Due to the limited sample size for the non-binary group, our subsequent analyses only focus on women and men, which we admit is a limitation of our study. To better understand the role of parenting, we aggregated respondents based on their self-reported parenthood status: the parent group (n=5,670, 73.3%) and non-parent (n=1,534, 19.8%) group. Subsequent analyses refer to respondents in the parent group who self-identified as women and men as mothers and fathers, and those in the non-parent group as non-mothers and non-fathers. It should be noted that the two groups only include respondents who have ever married or cohabited (for two years or longer), given work-family conflict and partner support being variables of significant interest in this study.

Our analysis of the parenthood status of academics provides new insights into the dilemma many women academics face: motherhood or academic career. On the one hand, mothers who are academics are significantly more likely than fathers to report higher levels of negative impact on careers due to children, reflecting the motherhood penalty on the academic career (Bonache et al., 2022; Misra et al., 2012). The perceived negative impact of parenting could intensify gender inequalities, as previous research found it was one of the major systemic barriers pushing women to self-select away from academia, contributing to the academic “leaky pipeline” (Anders, 2004). On the other hand, we found that women are less likely than men to have children and have fewer children.

The insignificant differences between non-fathers and non-mothers in their productivity, citation, and collaboration imply that parenthood contributes to the gendered differences in these three aspects for parents. Given the observed motherhood penalty, it is unsurprising that academic women have fewer children and are less likely to have children than men, which our results suggested as being related to career considerations: Our results show non-mothers are more likely than non-fathers to report their decision to be childfree was related to career considerations, and that mothers are more likely than fathers to report that the lower number of children they have was also related to career considerations.

Our study also unveils gender gaps in both objective and subjective career achievements of academics and suggests much of the gender gap only exists in the parent group. Specifically, we find gender differences in all measures of subjective career achievement and objective career achievement for the parent group. For the non-parent group, however, no significant gender difference was found in any career achievement measures. We thus argue that given many previous studies (Bendels et al., 2018; Caplar et al., 2017; Holliday et al., 2014; Larivière et al., 2013) reported overall gender disparities in academia without differentiating between parents and non-parents, these disparities may primarily derive from the differences between academic mothers and fathers. Follow-up studies on gender disparities could further examine this issue by comparing parents and non-parents.

Given the role of productivity in the academic reward system, current policies and regulations that aim to provide short-term support for childcaring and recovery right after birth or adoption are insufficient to bridge the gender gap in academia. Furthermore, the impact of parenting on the research output of mothers may last longer than asserted by previous studies (Morgan et al., 2021): We find significant gender differences in productivity across all career stages among the parents, but no such differences have been observed among the non-parents. This suggests that policies and interventions addressing gender inequalities in the scientific workforce should go beyond short-term assistance such as parental leaves and extend long-term support for mothers. Other forms of sustainable family-friendly support, such as subsidized childcare, onsite childcare, flexible working schedules, and supportive working environments, are available options (Feeney and Stritch, 2019). Additionally, given the observed gender gaps in research productivity, the current metric-based evaluation, especially those based on productivity, will undoubtedly hurt mothers more. Therefore, additional metrics and qualitative evaluations considering parenthood and individual conditions would be critical to reducing gender inequalities in academic settings.

Our findings on mothers’ higher chances of encountering more elevated levels of work-family conflict in all the forms of time-, strain-, and behavior-based conflict confirm that mothers indeed suffer more from multiple social roles: mother and researcher. Previous findings suggested that mothers are children’s primary caregivers, requiring extra time and engagement in parenting activities (Dickson, 2020). We also found that mothers are significantly more likely than fathers to experience time-based conflict and less likely to receive the time, emotional, and decision support from partners, a dangerous signal for the family to sustain marital satisfaction and gender equality (Mickelson et al., 2006; Thorstad et al., 2006). On the contrary, mothers are more likely to receive financial and technical support from their partners. Mothers’ higher likelihood of receiving financial support might be related to the entrenched gender division of household labor, where fathers are more likely to be the breadwinners and bear financial responsibilities (Hauser, 2012).

Our mediating effect analysis results provide new evidence for the claim that family is “the newest battlefront in the struggle for gender equality” (Hauser, 2012) for academia, confirming that family-related reasons are significantly associated with gender gaps in career achievements of parents. Specifically, mother who are academics receive less general support (including time, emotional and decisional support) from their partners than fathers, contributing to the gender gap in productivity and collaboration. It corroborates the finding that shortened working time and limited decision support for more work engagement restrict mothers’ opportunities for networking and academic collaboration (Finkel and Olswang, 1996; Mason et al., 2013). The limited general support from partners thus may hurt the research productivity of mothers directly or indirectly by constraining research collaborations, which usually leads to higher productivity (Abbasi et al., 2011; Sonnenwald, 2007). Moreover, the higher levels of work-family conflict experienced by mothers also worsen the gender gap in satisfaction over research and perceived recognition by parents, which are highly related to women’s turnover intention (Watanabe and Falci, 2016). Reducing mothers’ work-family conflict and increasing their general support could thus reduce the gender gap in academics’ perceived career achievement. These mediating variables indicate the directions for mitigating the negative impacts of motherhood: Partner support in the forms of time, emotional, and decision support, and other collective efforts to reduce the time-based, strain-based, and behavior-based conflict for mothers.

By combining large-scale survey and bibliometric data, we reveal the decisive role of parenthood in the current gender gap in academia and the underlying mechanisms of how parenting-related conflicts and support mediate the gap. Previous studies asserted the importance of various forms of support from governments and institutions (Morgan et al., 2021). Yet, our results show that family is another new battleground: assistance and support provided by partners in time, emotional and decisional support are also vital. Addressing gender inequality in academia is a task requiring collective intelligence. In addition to parenting-related support by governments and institutions, which could be sometime inadequate due to restrictions on resources and regulations in the U.S., the support from partners and families will also help narrow the gender gap in academia.

All data needed to evaluate the conclusions in the paper are present here and in the supplementary material. Aggregated or de-identified data on variables used in this study is available on GitHub (https://github.com/UWMadisonMetaScience/parenting, copy archived at swh:1:rev:46416c03c83eb28b77f6f17650537b2b413da663).

1. 1. Abbasi A
   2. Altmann J
   3. Hossain L

   (2011) Identifying the effects of co-authorship networks on the performance of scholars: A correlation and regression analysis of performance measures and social network analysis measures

   Journal of Informetrics 5:594–607.

   https://doi.org/10.1016/j.joi.2011.05.007

   • Google Scholar
2. 1. Adams A
   2. Golsch K

   (2021) Gender-specific patterns and determinants of spillover between work and family: The role of partner support in dual-earner couples

   Journal of Family Research 33:72–98.

   https://doi.org/10.20377/jfr-373

   • Google Scholar
3. 1. Anders S

   (2004) Why the academic pipeline leaks: Fewer men than women perceive barriers to becoming professors

   Sex Roles 51:511–521.

   https://doi.org/10.1007/s11199-004-5461-9

   • Google Scholar
4. 1. Astegiano J
   2. Sebastián-González E
   3. Castanho C

   (2019) Unravelling the gender productivity gap in science: A meta-analytical review

   Royal Society Open Science 6:181566.

   https://doi.org/10.1098/rsos.181566

   • PubMed
   • Google Scholar
5. 1. Beckett L
   2. Nettiksimmons J
   3. Howell LP
   4. Villablanca AC

   (2015) Do family responsibilities and a clinical versus research faculty position affect satisfaction with career and work-life balance for medical school faculty?

   Journal of Women’s Health 24:471–480.

   https://doi.org/10.1089/jwh.2014.4858

   • PubMed
   • Google Scholar
6. 1. Bellas ML

   (1999) Emotional labor in academia: The case of professors

   The ANNALS of the American Academy of Political and Social Science 561:96–110.

   https://doi.org/10.1177/000271629956100107

   • Google Scholar
7. 1. Bendels MHK
   2. Müller R
   3. Brueggmann D
   4. Groneberg DA
   5. Lozano S

   (2018) Gender disparities in high-quality research revealed by Nature Index journals

   PLOS ONE 13:e0189136.

   https://doi.org/10.1371/journal.pone.0189136

   • PubMed
   • Google Scholar
8. 1. Bonache H
   2. Carballo V
   3. Chas A
   4. Delgado N

   (2022) Go the extra mile: A gender approach towards parenthood in early-career academics

   Journal of Gender Studies 31:125–135.

   https://doi.org/10.1080/09589236.2020.1836482

   • Google Scholar
9. 1. Caplar N
   2. Tacchella S
   3. Birrer S

   (2017) Quantitative evaluation of gender bias in astronomical publications from citation counts

   Nature Astronomy 1:0141.

   https://doi.org/10.1038/s41550-017-0182

   • Google Scholar
10. 1. Ceci SJ
    2. Ginther DK
    3. Kahn S
    4. Williams WM

    (2014) Women in academic science: A changing landscape

    Psychological Science in the Public Interest 15:75–141.

    https://doi.org/10.1177/1529100614541236

    • PubMed
    • Google Scholar
11. 1. Chesler NC
    2. Barabino G
    3. Bhatia SN
    4. Richards-Kortum R

    (2010) The pipeline still leaks and more than you think: A status report on gender diversity in biomedical engineering

    Annals of Biomedical Engineering 38:1928–1935.

    https://doi.org/10.1007/s10439-010-9958-9

    • PubMed
    • Google Scholar
12. Website

    1. Colby G
    2. Fowler C

    (2020) Data Snapshot: IPEDS Data on Full-Time Women Faculty and Faculty of Color

    Accessed March 20, 2022.

    https://www.aaup.org/news/data-snapshot-full-time-women-faculty-and-faculty-color
13. 1. Dickson M

    (2020) “He’s not good at sensing that look that says, I’m drowning here!” Academic mothers’ perceptions of spousal support

    Marriage & Family Review 56:241–263.

    https://doi.org/10.1080/01494929.2020.1712302

    • Google Scholar
14. 1. Ductor L
    2. Goyal S
    3. Prummer A

    (2021) Gender and collaboration

    The Review of Economics and Statistics.

    https://doi.org/10.1162/rest_a_01113

    • Google Scholar
15. 1. Feeney MK
    2. Stritch JM

    (2019) Family-friendly policies, gender, and work–life balance in the public sector

    Review of Public Personnel Administration 39:422–448.

    https://doi.org/10.1177/0734371X17733789

    • Google Scholar
16. 1. Ferguson M
    2. Carlson D
    3. Kacmar KM
    4. Halbesleben JRB

    (2016) The supportive spouse at work: Does being work-linked help?

    Journal of Occupational Health Psychology 21:37–50.

    https://doi.org/10.1037/a0039538

    • PubMed
    • Google Scholar
17. 1. Finkel SK
    2. Olswang SG

    (1996) Child rearing as a career impediment to women assistant professors

    The Review of Higher Education 19:123–139.

    https://doi.org/10.1353/rhe.1996.0027

    • Google Scholar
18. 1. Fogarty TJ

    (2009) Show me the money: Academic research as currency

    Accounting Education 18:3–6.

    https://doi.org/10.1080/09639280802554418

    • Google Scholar
19. 1. Fox MF
    2. Fonseca C
    3. Bao J

    (2011) Work and family conflict in academic science: Patterns and predictors among women and men in research universities

    Social Studies of Science 41:715–735.

    https://doi.org/10.1177/0306312711417730

    • PubMed
    • Google Scholar
20. 1. Greenhaus JH
    2. Beutell NJ

    (1985) Sources of conflict between work and family roles

    The Academy of Management Review 10:76.

    https://doi.org/10.2307/258214

    • Google Scholar
21. 1. Greenhaus JH
    2. Powell GN

    (2006) When work and family are allies: A theory of work-family enrichment

    Academy of Management Review 31:72–92.

    https://doi.org/10.5465/amr.2006.19379625

    • Google Scholar
22. 1. Guarino CM
    2. Borden VMH

    (2017) Faculty service loads and gender: Are women taking care of the academic family?

    Research in Higher Education 58:672–694.

    https://doi.org/10.1007/s11162-017-9454-2

    • Google Scholar
23. 1. Hauser O

    (2012) Pushing daddy away? A qualitative study of maternal gatekeeping

    Qualitative Sociology Review 8:34–59.

    https://doi.org/10.18778/1733-8077.8.1.03

    • Google Scholar
24. 1. Holliday EB
    2. Jagsi R
    3. Wilson LD
    4. Choi M
    5. Thomas CR
    6. Fuller CD

    (2014) Gender differences in publication productivity, academic position, career duration, and funding among U.S. academic radiation oncology faculty

    Academic Medicine 89:767–773.

    https://doi.org/10.1097/ACM.0000000000000229

    • PubMed
    • Google Scholar
25. 1. Hunter LA
    2. Leahey E

    (2010) Parenting and research productivity: New evidence and methods

    Social Studies of Science 40:433–451.

    https://doi.org/10.1177/0306312709358472

    • Google Scholar
26. 1. Juraqulova Z
    2. Byington T
    3. Kmec J

    (2015)

    The impacts of marriage on perceived academic career success: Differences by gender and discipline

    The International Journal of Gender, Science and Technology 7:363–392.

    • Google Scholar
27. 1. Kelly K
    2. Grant L

    (2012) Penalties and premiums: The impact of gender, marriage, and parenthood on faculty salaries in science, engineering and mathematics (SEM) and non-SEM fields

    Social Studies of Science 42:869–896.

    https://doi.org/10.1177/0306312712457111

    • PubMed
    • Google Scholar
28. 1. Kinnunen U
    2. Feldt T
    3. Geurts S
    4. Pulkkinen L

    (2006) Types of work-family interface: Well-being correlates of negative and positive spillover between work and family

    Scandinavian Journal of Psychology 47:149–162.

    https://doi.org/10.1111/j.1467-9450.2006.00502.x

    • PubMed
    • Google Scholar
29. 1. Larivière V
    2. Ni C
    3. Gingras Y
    4. Cronin B
    5. Sugimoto CR

    (2013) Bibliometrics: Global gender disparities in science

    Nature 504:211–213.

    https://doi.org/10.1038/504211a

    • PubMed
    • Google Scholar
30. 1. Larivière V
    2. Gingras Y
    3. Sugimoto CR
    4. Tsou A

    (2015) Team size matters: Collaboration and scientific impact since 1900

    Journal of the Association for Information Science and Technology 66:1323–1332.

    https://doi.org/10.1002/asi.23266

    • Google Scholar
31. 1. Lawson C
    2. Geuna A
    3. Finardi U

    (2021) The funding-productivity-gender nexus in science, a multistage analysis

    Research Policy 50:104182.

    https://doi.org/10.1016/j.respol.2020.104182

    • Google Scholar
32. 1. Ley TJ
    2. Hamilton BH

    (2008) The gender gap in NIH grant applications

    Science 322:1472–1474.

    https://doi.org/10.1126/science.1165878

    • PubMed
    • Google Scholar
33. 1. Lunnemann P
    2. Jensen MH
    3. Jauffred L

    (2019) Gender bias in Nobel prizes

    Palgrave Communications 5:1–4.

    https://doi.org/10.1057/s41599-019-0256-3

    • Google Scholar
34. 1. Mairesse J
    2. Pezzoni M
    3. Visentin F

    (2019) Impact of family characteristics on the gender publication gap: Evidence for physicists in France

    Interdisciplinary Science Reviews 44:204–220.

    https://doi.org/10.1080/03080188.2019.1603884

    • Google Scholar
35. 1. Maliniak D
    2. Powers R
    3. Walter BF

    (2013) The gender citation gap in international relations

    International Organization 67:889–922.

    https://doi.org/10.1017/S0020818313000209

    • Google Scholar
36. 1. Martins LL
    2. Eddleston KA
    3. Veiga JF

    (2002) Moderators of the relationship between work-family conflict and career satisfaction

    Academy of Management Journal 45:399–409.

    https://doi.org/10.2307/3069354

    • Google Scholar
37. Book

    1. Mason MA
    2. Wolfinger NH
    3. Goulden M

    (2013)

    Do Babies Matter?: Gender and Family in the Ivory Tower (First edition)

    Rutgers University Press.

    • Google Scholar
38. 1. Meho LI

    (2021) The gender gap in highly prestigious international research awards, 2001–2020

    Quantitative Science Studies 2:976–989.

    https://doi.org/10.1162/qss_a_00148

    • Google Scholar
39. 1. Mickelson KD
    2. Claffey ST
    3. Williams SL

    (2006) The moderating role of gender and gender role attitudes on the link between spousal support and marital quality

    Sex Roles 55:73–82.

    https://doi.org/10.1007/s11199-006-9061-8

    • Google Scholar
40. 1. Misra J
    2. Lundquist JH
    3. Templer A

    (2012)

    Gender, work time, and care responsibilities among faculty

    Sociological Forum 27:300–323.

    • Google Scholar
41. Report

    1. Monte L

    (2017) Fertility Research Brief

    United States Census Bureau.

    https://www.census.gov/library/publications/2017/demo/p70br-147.html

    • Google Scholar
42. 1. Morgan AC
    2. Way SF
    3. Hoefer MJD
    4. Larremore DB
    5. Galesic M
    6. Clauset A

    (2021) The unequal impact of parenthood in academia

    Science Advances 7:eabd1996.

    https://doi.org/10.1126/sciadv.abd1996

    • PubMed
    • Google Scholar
43. Website

    1. National Center for Health Statistics

    (2021) Key statistics from the national survey of family growth

    Accessed November 8, 2021.

    https://www.cdc.gov/nchs/nsfg/keystatistics.htm
44. Report

    1. National Center for Science and Engineering Statistics

    (2021) Women, Minorities, and Persons with Disabilities in Science and Engineering: 2021 (Special Report NSF 21-321)

    National Science Foundation.

    https://www.nsf.gov/statistics/wmpd/

    • Google Scholar
45. 1. Ng TWH
    2. Eby LT
    3. Sorensen KL
    4. Feldman DC

    (2005) Predictors of objective and subjective career success: A meta-analysis

    Personnel Psychology 58:367–408.

    https://doi.org/10.1111/j.1744-6570.2005.00515.x

    • Google Scholar
46. 1. Paul-Hus A
    2. Bouvier RL
    3. Ni C
    4. Sugimoto CR
    5. Pislyakov V
    6. Larivière V

    (2015) Forty years of gender disparities in Russian science: A historical bibliometric analysis

    Scientometrics 102:1541–1553.

    https://doi.org/10.1007/s11192-014-1386-4

    • Google Scholar
47. 1. Powell K

    (2021) The parenting penalties faced by scientist mothers

    Nature 595:611–613.

    https://doi.org/10.1038/d41586-021-01993-x

    • Google Scholar
48. 1. Sonnenwald DH

    (2007) Scientific collaboration

    Annual Review of Information Science and Technology 41:643–681.

    https://doi.org/10.1002/aris.2007.1440410121

    • Google Scholar
49. 1. Thomas PA
    2. Liu H
    3. Umberson D

    (2017) Family relationships and well-being

    Innovation in Aging 1:igx025.

    https://doi.org/10.1093/geroni/igx025

    • PubMed
    • Google Scholar
50. 1. Thorstad RR
    2. Anderson TL
    3. Hall MEL
    4. Willingham M
    5. Carruthers L

    (2006)

    Breaking the mold: A qualitative exploration of mothers in christian academia and their experiences of spousal support

    Journal of Family Issues 27:229–251.

    • Google Scholar
51. 1. Valcour M
    2. Ladge JJ

    (2008) Family and career path characteristics as predictors of women’s objective and subjective career success: Integrating traditional and protean career explanations

    Journal of Vocational Behavior 73:300–309.

    https://doi.org/10.1016/j.jvb.2008.06.002

    • Google Scholar
52. 1. van Daalen G
    2. Willemsen TM
    3. Sanders K

    (2006) Reducing work–family conflict through different sources of social support

    Journal of Vocational Behavior 69:462–476.

    https://doi.org/10.1016/j.jvb.2006.07.005

    • Google Scholar
53. 1. van den Besselaar P
    2. Sandström U
    3. Dorta-González P

    (2017) Vicious circles of gender bias, lower positions, and lower performance: Gender differences in scholarly productivity and impact

    PLOS ONE 12:e0183301.

    https://doi.org/10.1371/journal.pone.0183301

    • PubMed
    • Google Scholar
54. 1. Wang HM
    2. Ma AL
    3. Guo TT

    (2020) Gender concept, work pressure, and work-family conflict

    American Journal of Men’s Health 14:1557988320957522.

    https://doi.org/10.1177/1557988320957522

    • PubMed
    • Google Scholar
55. 1. Watanabe M
    2. Falci CD

    (2016) A demands and resources approach to understanding faculty turnover intentions due to work–family balance

    Journal of Family Issues 37:393–415.

    https://doi.org/10.1177/0192513X14530972

    • Google Scholar
56. 1. Witteman HO
    2. Hendricks M
    3. Straus S
    4. Tannenbaum C

    (2019) Are gender gaps due to evaluations of the applicant or the science? A natural experiment at a national funding agency

    Lancet 393:531–540.

    https://doi.org/10.1016/S0140-6736(18)32611-4

    • PubMed
    • Google Scholar
57. 1. Xu YJ

    (2008) Gender disparity in STEM disciplines: A study of faculty attrition and turnover intentions

    Research in Higher Education 49:607–624.

    https://doi.org/10.1007/s11162-008-9097-4

    • Google Scholar

Decision letter after peer review:

Thank you for submitting your article "How parenthood contributes to gender gaps in academia" to eLife for consideration as a Feature Article. Your article has been reviewed by two peer reviewers, and the evaluation has been overseen by two members of the eLife Features Team (Helga Groll/Peter Rodgers). The following individual involved in review of your submission has agreed to reveal their identity: Diego Kozlowski.

The reviewers and editors have discussed the reviews and we have drafted this decision letter to help you prepare a revised submission.

Summary:

Zheng and colleagues present an interesting study based on analyses of a large-scale survey examining the relationship between parenthood and gender gaps in academia. This subject is of considerable relevance to the academic community and would be an appropriate thematic fit for eLife. However, the article needs significant improvement in clarity of language and the authors engage in quite a bit of conjecture that is not supported by their primary findings. Discussions of gender, parenting and career achievement all involve sensitive topics. Precision of language and staying within bounds of what is supported by the data are critical to clearly communicate the author's findings and avoid offense. This is a theme that will come up in many of the specific comments below.

The article overall shows a very thorough analysis, it is very well written, and has a careful design. It is an important contribution to the field, and clarifies very specific mechanisms of gender inequality. Below there are a number of recommendations and concerns. We think that clarifying those would help them to make the article even stronger.

Essential revisions:

1. Why do "All" parent categories have higher ARPs and ARCs (for both men and women) compared to nonparents (Table S10)? Do female parents produce more publications and receive more citations than female non-parents? Do male parents produce more publications and receive more citations than male non-parents? This seems like an important point to discuss and one which is at odds with the author's overall thesis. Also, as mentioned below, Morgan et al. 2021 report that faculty of both genders who have children are slightly more productive than those who do not. This point should be discussed and put into context with the author's data/conclusions.

2. I did not extensively math check, but in areas where I did, the numbers were confusing. Unless I am misunderstanding where the numbers in the text are coming from.

For example, on page 10,

– "women's ARP is 0.28 units lower than men": On the graph the numbers are 2.17-1.79 = 0.38?

– "women's ARC is 0.21 units lower than men": On the graph the numbers are 2.27-2.08 = 0.19?

Or on page 11,

– "ARP, ARC and ARCo are 0.35, 0.2 and 0.10 units…" On the graph the numbers are 2.27-1.82 = 0.45, 2.28-2.13 = 0.15 and 1.09-0.98 = 0.11?

3. The author's discuss the "motherhood penalty". There is also a well-documented "fatherhood premium." How does this fit in with the author's results? Do the author's see evidence of a "fatherhood premium" in their data?

4. Morgan et al. (https://www.science.org/doi/10.1126/sciadv.abd1996) surveyed 3064 tenure-track faculty in 450 departments in the US and Canada. What advances are made in the current datasets/analyses? How do the results compare? In addition to the point mentioned above about faculty of both genders who have children being reported as more productive than those without, Morgan et al. also reports that results are inconclusive as to whether long-term publication rates are affected by parenthood. Can the authors place their study and results into context here?

5. Were there any differences for men/women parents/non-parents observed between different fields within academia? Are the conclusions the same regardless of discipline? Are the disparities the same regardless of discipline?

6. The authors do not distinguish between respondents who currently have childcare responsibilities (i.e., children < 18 currently residing with them) versus those who have ever had children (i.e., children > 18, non-custodial parents, etc.). The authors also do not distinguish between same-sex and opposite sex parents, where gender roles and expectations may differ. These should be raised in the Limitations section of the manuscript, as the current analyses focus on a model of a traditional, opposite-sex, nuclear family.

7. Data needs to be fully referenced within the text and shown if referenced. For example, on page 7 of the Results section, the authors refer to women in academia being more likely to have fewer children (presumably as compared with men?). There is not a reference for a data table here and the data does not appear to be in the supplemental tables? The authors then state "This trend is different from the US general population, where more adult women have children and have more children than men." (REF). Do the authors mean primary caretaking responsibilities of children or children that reside with them?

8. On page 11, last paragraph, the authors discuss results around career stage. This needs a reference to the data table.

9. On page 7, the authors state: "Among those without children, women (59.9%) are more likely than men (43.2%) to report that they chose to be childfree due to career considerations…". If I understand correctly, the survey question was: "Is the number of children (include 0) you currently have related to your career considerations (more or less)?" The interpretation is cause and effect (women chose to be childfree due to career) whereas the survey question asks if these were related considerations (which could include additional factors?).

10. In the Introduction, the authors use the term "non-binaries" to refer to individuals whose self-reported gender identity on the survey was "Other". Please consult the APA guidelines for bias free language: https://apastyle.apa.org/style-grammar-guidelines/bias-free-language/gender

11. In the Introduction, the authors state, "The multigenerational gender role beliefs, which characterize men and women with a simplistic set of features, may direct women academics toward women's normative roles in the family (REFs). Some women academics may become 'maternal gatekeepers" who are reluctant to relinquish family responsibility" (REFs). This seems to stray from the facts and into speculative territory not supported by the author's data.

12. In the Introduction, the authors state, "Although partner support is believed to be more critical to women's career success than to men, especially for those in parenthood, women's demand for partner support is less satisfied" (REFs). What does this mean? Women are more dependent on partner support for career success regardless of parenthood? Is this supported by data? Men need less partner support even in parenthood? Doesn't the author's data show male academic parents receive more time support, etc.?

13. This sentence between page 8-9 is confusing: "While the satisfaction over research is significantly different between mothers and fathers, our results show no significant gender differences in the career satisfaction of parents…therefore, it is likely that women consider more on other gains (e.g., emotional gains) from teaching and service while assessing their overall career satisfaction, while men emphasize research." What do the authors mean by "women consider more on other gains (e.g., emotional gains)" and is this data-driven or speculation? And what is meant by "men emphasize research"? It seems a stretch to attribute motivation / feelings to the author's data.

14. What does this sentence on page 13 mean, "women are usually disproportionately associated with technical work in research" (REF)?

15. In the Discussion section, the authors state, "the perceived negative impact of parenting is one of the major systematic barriers for women's self-selecting attrition from academia…" Can the authors support the claims of self-attrition and self-attrition due to a perceived negative impact of parenting? This seems to imply that women are deciding it's too hard to have children and an academic career, and then deciding to leave academia. Is there data to support this? This also entirely ignores bias/discrimination against women and mothers, which is well documented in academia.

16. In the Discussion section, while I don't believe the authors intended this statement to be offensive, some portions may come across that way: "Women in academia…tend to slough off the burden of children… Given the motherhood penalty, it is unsurprising that many women stay in academia by paying the price of being childfree…" In particular, this places value judgements around having children or not having children. It also seems unsupported by the data.

17. In the Discussion section, the authors discuss the finding that "mothers are more likely to receive financial and technical support from their partners." By stating "It reflects the entrenched gender division of household labor that expands with the new childcare task: Mothers are recognized, and even morally self-recognized, as the babysitters and fathers are the breadwinners" (REF). "As a social norm…fathers bear the financial responsibility. In contrast, fathers compensate mothers' loss due to extra childcare responsibility by offering financial and technical support." Again, can the authors draw this conclusion from their data?

18. In the Discussion section, the authors again make reference to "high attrition rates of mothers in the scientific workforce due to lower satisfaction rates." Can the authors backup the idea of higher attrition rates of mothers compared with non-mothers in the scientific workforce, and also that the reasons for attrition are due to lower satisfaction rates (i.e., self-selection)?

19. P.6. I am happy to see that the authors included non-binary people in their survey. I also understand that the small sample makes it impossible to consider NB into the regression models. Nevertheless, it would be very important to include their responses somehow. A qualitative analysis of the 28 responses from NB people after the quantitative analysis of men and women, to see where they locate within the more general results would be a very interesting complementary analysis.

20. It is not clear to me why the authors decided to send the survey to 25% of the population. If there was a technical limitation, it would be good to state it. Besides this, there is a 9% response rate. It would be helpful to see some robustness checks o whether that 9% is representative of the population or not. Some comparative statistics between respondents and non-respondents would help to disclose some potential self-selection biases.

21. The sample size of parents and non-parents is different, with a larger number of parents to non-parents (roughly 80% to 20%). Statistical measures are usually dependent on sample size, with larger sample sizes having more rejection power. Is it possible that the differences seen between groups, no significant bias on non-parents and significant bias on parents is just and statistical artifact of the sample sizes? The authors could try to check if this is the case using bootstrapped samples of equal size, or also consider directly the absolute values of the metrics to see if there is an important difference between the two groups -besides the statistical difference-.

Further on the previous point. On Figure 1A the authors show the different subjective satisfaction measures by gender and group. What I can infer from that figure is that the biggest difference in gaps between groups -i.e where parent's gap is larger than non-parent gaps- is on career satisfaction and community recognition. On research, although parents do present a larger gap, the non-parents' group also has a bias that is larger than on the other metrics. So this reinforces my doubt about the sample sizes of the two groups and how it affects the statistical results.

To be clear, I do not think this invalidates the results at all. The analysis has been thoroughly done. I would like to see some more robustness checks and details on this issue, and also on the interpretation of SUEST values.

22. P. 12 and table S11. The authors explain that work-family conflict can take the form of time, strain, and behavior-based conflict. They show the difference between groups for the work-family conflict, but they do not mention that for the three specific forms the non-parent group also showed significant differences between men and women. How can we interpret these apparently contradictory results?

23. P. 19 Regarding the policy recommendations, I would add that the metric-based evaluation, and especially the productivity-based career evaluations would harm mothers, as the paper has shown. So, qualitative evaluations that take into consideration parenthood and the individual conditions would also be very important to reduce gender biases.

Essential revisions:

1. Why do "All" parent categories have higher ARPs and ARCs (for both men and women) compared to nonparents (Table S10)? Do female parents produce more publications and receive more citations than female non-parents? Do male parents produce more publications and receive more citations than male non-parents? This seems like an important point to discuss and one which is at odds with the author's overall thesis. Also, as mentioned below, Morgan et al. 2021 report that faculty of both genders who have children are slightly more productive than those who do not. This point should be discussed and put into context with the author's data/conclusions.

Thank you for the close attention to the results. We agree that the mean values of ARP and ARC look higher for parents than for non-parents. However, we did not compare the mean values between parents and non-parents directly due to the considerations of (1) the skewness in distributions and (2) the absence of control variables. Mean values are sensitive to extreme values and depend on the distributions of data. We plotted the distributions of ARPs and ARCs by gender, parenthood status, and career stage (see Author response image 1). We found that the values of ARPs and ARCs overall fall into normal distributions with close means and “long tails” across different genders and parent groups, but the tails are longer for the parent groups overall. Additionally, the effect of control variables is not considered in the direct comparison of means, which could be problematic.

Therefore, we used regression analysis to compare between parents and non-parents, controlling for a set of covariates (similar to other analysis in the project). As shown in SI Appendix Table S11, there is no significant difference between mothers and non-mothers in ARP, ARC, or ARCo. However, fathers do have higher ARP and ARCo values than non-fathers, but not ARC. The finding of higher ARP for fathers than non-fathers is consistent with that of Morgan et al. (2021). Yet, we did not find such a pattern between mothers and non-mothers, which is different from Morgan et al. 2021. Additionally, given the potential positive relationship between collaboration and productivity (Larivière et al., 2015), the relationship between the ARP values of fathers and non-fathers could also be related to their collaboration, in addition to fatherhood. We added the following text into the manuscript:

“Additionally, the mean values of the of ARP and ARC are generally higher for the parent group than the non-parent group for both genders (see Table Supplement Table S10). Yet, it should be noted that the direct comparison of means ignores the potential skewness in data distribution and the effect of control variables. Therefore, we further performed regression analysis to compare the parent and non-parent group (see Table Supplement Table S11). We found no significant difference between mothers and non-mothers in ARP, ARC, or ARCo. However, fathers do have higher ARP and ARCo values than non-fathers, but not ARP. The finding of higher ARP for fathers than non-fathers is consistent with that in Morgan et al. (2021). Yet, we did not find such a pattern between mothers and non-mothers, which is different from Morgan et al. (2021). Given the potential positive relationship between collaboration and productivity (Larivière et al., 2015), the relationship between the ARP values of fathers and non-fathers could also be related to their collaboration, in addition to fatherhood.”

Author response image 1

Download asset Open asset

2. I did not extensively math check, but in areas where I did, the numbers were confusing. Unless I am misunderstanding where the numbers in the text are coming from.

For example, on page 10,

– "women's ARP is 0.28 units lower than men": On the graph the numbers are 2.17-1.79 = 0.38?

– "women's ARC is 0.21 units lower than men": On the graph the numbers are 2.27-2.08 = 0.19?

Or on page 11,

– "ARP, ARC and ARCo are 0.35, 0.2 and 0.10 units…" On the graph the numbers are 2.27-1.82 = 0.45, 2.28-2.13 = 0.15 and 1.09-0.98 = 0.11?

We appreciate the comment. The number in statements like "women's ARP is 0.28 units lower than men" refers to the linear regression coefficient for the independent variable gender. We did not try to compare the mean values in these statements, knowing that would likely be skewed due to situations like extreme values, distributions, and control variables. To clarify the confusion, we have moved the 95% confidence intervals and the p-values closer to the coefficients. The revised text reads as follows:

“Our results further confirmed gender disparities in objective career achievements (See Figure 1 and Table Supplement Table S10). Regardless of parenthood status, women are outperformed by men in all three measures of objective career achievement. Specifically, women’s ARP is 0.28 (95% CI [-0.39, -0.17], p=0.000) units lower than men, women’s ARC is 0.21 (95% CI [-0.38, -0.05], p=0.013) units lower than men, and women’s ARCo is 0.08 (95% CI [-0.14, -0.01], p=0.023) unit lower than men. This echoes previous findings that women produce fewer publications (Astegiano et al., 2019; Besselaar and Sandström, 2017), receive lower citations (Larivière et al., 2013; Maliniak et al., 2013), and have fewer coauthors (Ductor et al., 2021) than their male counterparts.”

3. The author's discuss the "motherhood penalty". There is also a well-documented "fatherhood premium." How does this fit in with the author's results? Do the author's see evidence of a "fatherhood premium" in their data?

We are grateful for these questions. To answer these questions, we further compared the parent and non-parent group in terms of their objective and subjective career achievements (see SI Appendix Table S11). We did not find “fatherhood premium” in any of the three measures of subjective career achievements. However, we did find that fathers are more productive than non-fathers, showing some levels of fatherhood premium in research productivity. In the meantime, we also observed that fathers are more collaborative than non-fathers, which could be related to the higher productivity of fathers besides fatherhood, given possible causal effects between collaboration and productivity (Larivière et al., 2015).

4. Morgan et al. (https://www.science.org/doi/10.1126/sciadv.abd1996) surveyed 3064 tenure-track faculty in 450 departments in the US and Canada. What advances are made in the current datasets/analyses? How do the results compare? In addition to the point mentioned above about faculty of both genders who have children being reported as more productive than those without, Morgan et al. also reports that results are inconclusive as to whether long-term publication rates are affected by parenthood. Can the authors place their study and results into context here?

We appreciate that the reviewers brought up the important contribution by Morgan et al. (2021). Compared with their study, our study is different in many ways, including (but not limited to):

(1) Morgan et al.’s study focused on the gender gap in productivity due to parenting, while our study examines gender gaps in a broader spectrum by covering both objective (productivity, citation, and collaboration) and subjective (research satisfaction, career satisfaction and community recognition) career achievements.

(2) Morgan et al.’s study focused on three disciplines (computer science, business, and history), while our study examines all disciplines (as covered in Web of Science).

(3) Morgan et al.’s study analyzed productivity, position satisfaction and their relationships with parental leave policy, while our study focused on the role of family (especially partners) in shaping the gender disparities we observe.

The difference between the Morgan et al.’s study and this project goes beyond that. But we think our research aligns with Morgan et al.’s study in some way and complements it with additional findings by focusing on a wider spectrum of disciplines, assessing both subjective and objective career achievements, and concentrating on the role of partners and families.

Regarding productivity, the Morgan et al.’s study found that mothers’ productivity drops at a higher rate than fathers’ right after the birth of their children. They also found that parents are overall slightly more productive than non-parents regardless of gender. We did also find fathers are more productivity than non-fathers overall, but we did not find mothers are more productive than non-mothers. As mentioned in our responses to comments #1 and #3, the higher number of collaborators fathers have than non-fathers might also be related to fathers’ higher productivity. Future research might consider looking into the causality between these two factors.

As for the long-term impact of parenthood on productivity, because we did not collect the time of each childbirth for parents, our analysis on the long-term impact of parenthood on productivity was limited. As an approximation, we analyzed the productivity gender gap in both parent and non-parent group based on career stage (see Table Supplement Table S10), based on the assumption the career stage and length is reflective of the time of childbirth. We are aware that this might not be the case for many academics. Overall, our result shows that mothers are less productive than fathers across early, middle, and late career. But women and men in the non-parent group do not show this pattern. Therefore, the relationship between parenthood and gender gaps in productivity seems to go beyond early or mid-career and exists across all career stages.

5. Were there any differences for men/women parents/non-parents observed between different fields within academia? Are the conclusions the same regardless of discipline? Are the disparities the same regardless of discipline?

We thank the reviewers for bringing up these important questions. We compared the subjective and objective career achievements between genders by disciplinary area. Regarding objective career achievements, we found that fathers are more productive than mothers in medical sciences, and natural sciences and engineering, but not in arts and humanities, and social sciences. Because we rely on data from Web of Science, it is likely that academics productivity is not fully captured for arts and humanities, and social sciences due to the known coverage gap in these two areas by Web of Science. Regarding subjective career achievements, women are more likely to be less satisfied with their research than men in arts and humanities, medical sciences, and natural sciences and engineering. Mothers are more likely to have lower research or career satisfaction in all four disciplinary areas, and less likely to feel recognized by their scholarly communities in natural science and engineering, and medical sciences.

While the conclusions for some disciplinary areas are slightly different from others, this does not necessarily indicate that our previous conclusions are invalid. These differences are expected due to variances in scholarship practices among those disciplinary areas.

Author response image 2

Download asset Open asset

Author response image 3

Download asset Open asset

Author response image 4

Download asset Open asset

Author response image 5

Download asset Open asset

6. The authors do not distinguish between respondents who currently have childcare responsibilities (i.e., children < 18 currently residing with them) versus those who have ever had children (i.e., children > 18, non-custodial parents, etc.). The authors also do not distinguish between same-sex and opposite sex parents, where gender roles and expectations may differ. These should be raised in the Limitations section of the manuscript, as the current analyses focus on a model of a traditional, opposite-sex, nuclear family.

We completely agree with the issues brought up by the reviewers. We have added the following text to our Limitation section per reviewers’ suggestions.

“The current study is also limited by the fact that the survey did not collect data to distinguishing same-sex and opposite-sex partners, where gender roles and expectations may differ. Additionally, the current study lacks data on the types of childcare responsibilities by parents, such as whether they are custodial parents or non-custodial parents, and whether children live with them. To further understand how childcare responsibilities contributes to gender gaps in academia, data and analyses based on the factors mentioned above will be critical for understanding the true mechanisms and possible policy implications for address gender inequalities in the scientific workforce.”

7. Data needs to be fully referenced within the text and shown if referenced. For example, on page 7 of the Results section, the authors refer to women in academia being more likely to have fewer children (presumably as compared with men?). There is not a reference for a data table here and the data does not appear to be in the supplemental tables? The authors then state "This trend is different from the US general population, where more adult women have children and have more children than men." (REF). Do the authors mean primary caretaking responsibilities of children or children that reside with them?

We thank the reviewers for bringing this up. In the statement mentioned by the reviewer, we did not reference a table because it only involved one single regression analysis. We have appended the details of this regression to Table S6 and added the reference to the text.

As for the statement "This trend is different from the US general population, where more adult women have children and have more children than men", it was based on the 2017 Survey of Income and Program Participation by the U.S. Census Bureau. Based on the survey description, we don’t believe they differ in terms of childcare responsibilities. Our survey did not distinguish parents in this manner either, which we acknowledge as a limitation of our study. Thus, this statement simply commented on the number of children they have. We have revised the sentence as follows.

This trend is different from the general population in the United States, where the percentage of adult women who have children is higher than that of adult men, and the average number of children mothers have is higher than that of fathers (Monte, 2017; National Center for Health Statistics, 2021).

8. On page 11, last paragraph, the authors discuss results around career stage. This needs a reference to the data table.

We have added Table S10 to this paragraph as the reference table.

9. On page 7, the authors state: "Among those without children, women (59.9%) are more likely than men (43.2%) to report that they chose to be childfree due to career considerations…". If I understand correctly, the survey question was: "Is the number of children (include 0) you currently have related to your career considerations (more or less)?" The interpretation is cause and effect (women chose to be childfree due to career) whereas the survey question asks if these were related considerations (which could include additional factors?).

We highly appreciate the comment, which we completely agree with. We have changed the mentioned texts to the following to avoid causal implications:

“Among those without children, women (59.9%) are more likely than men (43.2%) to report that their choices to be childfree were related to career considerations (OR=2.10, 95%CI [1.69, 2.62], p=0.000; see Table Supplement Table S7), while controlling for discipline, career stage, and race.”

10. In the Introduction, the authors use the term "non-binaries" to refer to individuals whose self-reported gender identity on the survey was "Other". Please consult the APA guidelines for bias free language: https://apastyle.apa.org/style-grammar-guidelines/bias-free-language/gender

We highly appreciate the reviewers for the comment and suggestions. We’d like to clarify that we classified those who chose the “Other” category and added text comments that imply them to be in the non-binary category as non-binaries. Such comments may include (but are not limited to) keywords such as genderqueer, agender, bigender, gender-fluid, non-binary, gender-nonconforming, gender-neutral, and gender-creative.

11. In the Introduction, the authors state, "The multigenerational gender role beliefs, which characterize men and women with a simplistic set of features, may direct women academics toward women's normative roles in the family (REFs). Some women academics may become 'maternal gatekeepers" who are reluctant to relinquish family responsibility" (REFs). This seems to stray from the facts and into speculative territory not supported by the author's data.

Thank you for your comments on the Introduction part. We agree that these statements are a bit stray from the facts and not supported by our data. We thus removed the statements from our manuscript.

12. In the Introduction, the authors state, "Although partner support is believed to be more critical to women's career success than to men, especially for those in parenthood, women's demand for partner support is less satisfied" (REFs). What does this mean? Women are more dependent on partner support for career success regardless of parenthood? Is this supported by data? Men need less partner support even in parenthood? Doesn't the author's data show male academic parents receive more time support, etc.?

We thank the reviewers for the comment. We have modified our text to the following to avoid confusions:

“Although partner support has been found to help reduce mothers’ work-family conflict to an extent greater than fathers’, women’s demand for partner support is often less satisfied (Adams and Golsch, 2021; Dickson, 2020).”

13. This sentence between page 8-9 is confusing: "While the satisfaction over research is significantly different between mothers and fathers, our results show no significant gender differences in the career satisfaction of parents…therefore, it is likely that women consider more on other gains (e.g., emotional gains) from teaching and service while assessing their overall career satisfaction, while men emphasize research." What do the authors mean by "women consider more on other gains (e.g., emotional gains)" and is this data-driven or speculation? And what is meant by "men emphasize research"? It seems a stretch to attribute motivation / feelings to the author's data.

We thank the reviewers for these comments. We agree that the original statement is a little stretchy from our data. We have changed the related text to the following to avoid possible ambiguities and conjecture.

While the satisfaction over research is significantly different between mothers and fathers, our results show no significant gender difference in the career satisfaction of parents (OR=0.89, 95%CI [0.77,1.03], p=0.129). Furthermore, of those women who expressed satisfaction over their career achievements, 13.6% were not satisfied with their research achievements. The number is 8.9% for men (see Table Supplement Table S9). Research, teaching, and service being the three primary duties for many tenure-track faculty, it has been found that teaching and service are more heavily loaded on women than on men (Bellas, 1999; Guarino and Borden, 2017). If teaching and service are indeed loaded heavier on women, then mothers’ lower satisfaction rates on research could be related to the extra amount of time and effort they put on these aspects, on top of mothers’ already

14. What does this sentence on page 13 mean, "women are usually disproportionately associated with technical work in research" (REF)?

We thank the reviewers for the question. We meant to cite the Macaluso et al. (2016) study’s finding that women were significantly more likely to be associated with tasks such as performing experiments, especially in laboratories. Given that experiments might not necessary be the technical support in our study, we removed the statement to avoid any possible stretch of our results.

15. In the Discussion section, the authors state, "the perceived negative impact of parenting is one of the major systematic barriers for women's self-selecting attrition from academia…" Can the authors support the claims of self-attrition and self-attrition due to a perceived negative impact of parenting? This seems to imply that women are deciding it's too hard to have children and an academic career, and then deciding to leave academia. Is there data to support this? This also entirely ignores bias/discrimination against women and mothers, which is well documented in academia.

We thank the reviewers for the questions and comments. By the statement, "the perceived negative impact of parenting is one of the major systematic barriers for women's self-selecting attrition from academia, contributing to the "leaky pipeline" in academia (Anders, 2004)”, we meant to express our agreement with the findings in the Anders’s (2004) study, where they found systemic barriers associated with parenting discourage women from pursuing academic careers. We agree with the reviewers that parenting related issue is only one reason for the “leaky pipeline”, and many issues such as bias and discriminations against women and mothers also contribute significantly to gender inequalities in science. This was also why we emphasized that “the perceived negative impact of parenting is one of major systemic barriers…”. To avoid possible confusions, we rephrased the statement as follows:

“The perceived negative impact of parenting could intensify gender inequalities, as previous research (Anders, 2004) found it was one of the major systemic barriers for women's self-selecting away from academia, contributing to the academic "leaky pipeline".”

Reference:

Anders, S. M. van. (2004). Why the Academic Pipeline Leaks: Fewer Men than Women Perceive Barriers to Becoming Professors. Sex Roles, 51(9), 511–521. https://doi.org/10.1007/s11199-004-5461-9

16. In the Discussion section, while I don't believe the authors intended this statement to be offensive, some portions may come across that way: "Women in academia…tend to slough off the burden of children… Given the motherhood penalty, it is unsurprising that many women stay in academia by paying the price of being childfree…" In particular, this places value judgements around having children or not having children. It also seems unsupported by the data.

We highly appreciate the comment. We have corrected our language in the text to be:

“On the other hand, we found that women are less likely than men to have children and have fewer children. The insignificant differences between non-fathers and non-mothers in their productivity, citation, and collaboration imply that parenthood contributes to the gendered differences in these three aspects for parents. Given the motherhood penalty, it is unsurprising that academic women have fewer children and are less likely to have children than men, which our results suggested as being related to career considerations: Our results show non-mothers are more likely than non-fathers to report their decision to be childfree was related to career considerations and mother academics are more likely than father academics to report that the lower number of children they have was also related to career considerations.”

17. In the Discussion section, the authors discuss the finding that "mothers are more likely to receive financial and technical support from their partners." By stating "It reflects the entrenched gender division of household labor that expands with the new childcare task: Mothers are recognized, and even morally self-recognized, as the babysitters and fathers are the breadwinners" (REF). "As a social norm…fathers bear the financial responsibility. In contrast, fathers compensate mothers' loss due to extra childcare responsibility by offering financial and technical support." Again, can the authors draw this conclusion from their data?

We highly appreciate the comments. In the Discussion section, by stating "It reflects the entrenched gender division of household labor that expands with the new childcare task: Mothers are recognized, and even morally self-recognized, as the babysitters and fathers are the breadwinners (Hauser, 2012). As a social norm, it is more common that mothers sacrifice their working time, emotionally support fathers' work, and understand their decisions because fathers bear the financial responsibility.", we are trying to connect our findings to potential reasons behind that. We have changed our statements to the following:

“On the contrary, mothers are more likely to receive financial and technical support from their partners. Mothers’ higher likelihood of receiving financial support might be related to the entrenched gender division of household labor, where fathers are more likely to be the breadwinners and bear financial responsibilities (Hauser, 2012).”

18. In the Discussion section, the authors again make reference to "high attrition rates of mothers in the scientific workforce due to lower satisfaction rates." Can the authors backup the idea of higher attrition rates of mothers compared with non-mothers in the scientific workforce, and also that the reasons for attrition are due to lower satisfaction rates (i.e., self-selection)?

We highly appreciate the comments. Our data do not support the claims directly. We intended to cite previous research and connect with our findings. We have changed the corresponding text as follows:

“Moreover, the higher levels of work-family conflict experienced by mothers also worsen the gender gap in satisfaction over research and perceived recognition by parents, which are highly related to women's turnover intention (Watanabe and Falci, 2016). Reducing mothers' work-family conflict and increasing their general support could thus reduce the gender gap in academics' perceived career achievement. These mediating variables indicate the directions for mitigating the negative impacts of motherhood: Partner support in the forms of time, emotional, and decision support, and other collective efforts to reduce the time-based, strain-based, and behavior-based conflict for mothers.”

19. P.6. I am happy to see that the authors included non-binary people in their survey. I also understand that the small sample makes it impossible to consider NB into the regression models. Nevertheless, it would be very important to include their responses somehow. A qualitative analysis of the 28 responses from NB people after the quantitative analysis of men and women, to see where they locate within the more general results would be a very interesting complementary analysis.

We agree with the reviewers that the inclusion of non-binary respondents is critical, but any regression analysis will not be feasible due to the limited sample size. Here is the descriptive analysis regarding the 28 no-binary respondents.

In total, 28 non-binary participants responded to our survey, of which 18 ever committed partnership. Seven of them have children, among which six (85.7%) indicated that having children has a negative impact on their career, and all seven (100%) respondents reported that the number of children they have is related to career considerations.

Regarding the three measures of subjective career achievement, 5 out of 7 (71.4%) parents are not satisfied with their research achievements, as compared to 4 out of 11 (36.4%) non-parents. For career satisfaction, 4 out of 7 (57.1%) parents expressed disagreement, as compared to 2 out of 11 (18.2%) non-parents. For scholarly recognition, 1 out of 6 (16.7%) parents expressed disagreement, as compared to 2 out of 11 (18.2%) for non-parents.

Regarding the three measures of objective career achievement, the ARP is 1.34 for parents and 1.36 for non-parents. The ARC is 1.58 for parents and 1.79 for the non-parents. The ARCo is 0.92 for the parents and 0.98 for the non-parents.

The limited descriptive analysis shows that parenthood is also related to non-binary respondents’ career achievements. However, given the small sample size, the above comparisons may lack statistical power and therefor be interpreted with limited generalizability.

20. It is not clear to me why the authors decided to send the survey to 25% of the population. If there was a technical limitation, it would be good to state it. Besides this, there is a 9% response rate. It would be helpful to see some robustness checks o whether that 9% is representative of the population or not. Some comparative statistics between respondents and non-respondents would help to disclose some potential self-selection biases.

We thank the reviewers for the suggestions. We took a random sample consisting of 25% (99,168) of the population mainly because the version of Qualtrics (the tool we used to distribute the survey) had a weekly limit (10,000) for the number of email addresses that we could send the survey to. We thus decided to take a sample so that we could distribute the survey in a reasonably short period of time. Our response rate is lower than the Morgan et al.’s study, but a bit higher than some other studies (Ni et al., 2021).

To evaluate the representatives of our analytical sample, we compared the gender and discipline area compositions in the population, surveyed group, respondents, and final analytical sample. The distributions are shown in the updated Table S1. Please note that the gender categorization in the following table was decided using the method by Larivière et al. (2013), which is different from the gender classification in our analyses (based on self-reported data from the survey).

References:

Larivière, V., Ni, C., Gingras, Y., Cronin, B., and Sugimoto, C. R. (2013). Bibliometrics: Global gender disparities in science. Nature, 504(7479), 211–213. https://doi.org/10.1038/504211a

Ni, C., Smith, E., Yuan, H., Larivière, V., and Sugimoto, C. R. (2021). The gendered nature of authorship. Science Advances, 7(36), eabe4639. https://doi.org/10.1126/sciadv.abe4639

21. The sample size of parents and non-parents is different, with a larger number of parents to non-parents (roughly 80% to 20%). Statistical measures are usually dependent on sample size, with larger sample sizes having more rejection power. Is it possible that the differences seen between groups, no significant bias on non-parents and significant bias on parents is just and statistical artifact of the sample sizes? The authors could try to check if this is the case using bootstrapped samples of equal size, or also consider directly the absolute values of the metrics to see if there is an important difference between the two groups -besides the statistical difference.

Further on the previous point. On Figure 1A the authors show the different subjective satisfaction measures by gender and group. What I can infer from that figure is that the biggest difference in gaps between groups -i.e where parent's gap is larger than non-parent gaps- is on career satisfaction and community recognition. On research, although parents do present a larger gap, the non-parents' group also has a bias that is larger than on the other metrics. So this reinforces my doubt about the sample sizes of the two groups and how it affects the statistical results.

To be clear, I do not think this invalidates the results at all. The analysis has been thoroughly done. I would like to see some more robustness checks and details on this issue, and also on the interpretation of SUEST values.

We appreciate the comments. We agree that the relatively small sample size of the non-parent group and the difference of sample sizes could potentially raise concerns. Therefore, as a robustness test of our results, we used non-parametric bootstrapping to estimate the 95% bias-corrected confidence intervals of our models across the non-parent and parent groups. The test is performed 5,000 times by resampling with replacement from each group respectively to obtain a series of new samples of the same sample sizes. If the 95% bias-corrected confidence interval does not contain zero, we will consider the result as significant. Our bootstrap results (see Author response table 1) show that overall, the results are consistent with the results in our original manuscript.

The p values of SUEST are used to show whether the women/men odds ratio value in the parent group is significantly different from that in the non-parent group. If the p-value is below the predetermined significance level, we will accept the null hypothesis that the effect of gender (women/men) on the dependent variable is significantly different in the regression models for the parent and non-parent groups. We also performed bootstrapping to check the robustness of the results, in which we obtained 5,000 cross-group coefficient differences for gender (women/men) between the non-parent and parent groups to draw empirical p values. We found that the bootstrapping results are consistent with the results using the SUEST methods.

22. P. 12 and table S11. The authors explain that work-family conflict can take the form of time, strain, and behavior-based conflict. They show the difference between groups for the work-family conflict, but they do not mention that for the three specific forms the non-parent group also showed significant differences between men and women. How can we interpret these apparently contradictory results?

We thank the reviewers for their comments. We agree that there are noticeable differences between women and men in the non-parent group concerning time, strain, and behavior-based conflict. However, the levels of conflicts at all three forms are different between the non-parent and parent group (see Table S12):

• The percentage of non-parents experiencing work-family conflict: time-based conflict (women=39.1% vs. men=27.1%), strain-based conflict (women=16.1% vs. men=14.1%), and behavior-based conflict (women=8.8% vs Men=5.1%).

• The percentage of parents experiencing work-family conflict: time-based conflict (women=54.7 % vs. men=39.0 %), strain-based conflict (women=30.8% vs. men=20.6%), and behavior-based conflict (women=11.5% vs men=7.9%).

Overall, the level of conflicts is much lower in the non-parent group than in the parent group, although it still shows difference between non-parent women and men. This may further imply that women are more likely than men to experience the three types of conflict, but having children may exacerbate the gap. Therefore, we do not think these results are contradictory. Instead, this further implies that children matter a lot in the difference in work-family conflict experienced by women and men.

23. P. 19 Regarding the policy recommendations, I would add that the metric-based evaluation, and especially the productivity-based career evaluations would harm mothers, as the paper has shown. So, qualitative evaluations that take into consideration parenthood and the individual conditions would also be very important to reduce gender biases.

We highly appreciate the suggestions. We have added the following text to the Discussion section on Page 19.

“Additionally, given the observed gender gaps in research productivity, the current metric-based evaluation, especially those based on productivity, will undoubtedly hurt mothers more. Therefore, additional metrics and qualitative evaluations that take into account parenthood and individual conditions would be critical to reduce gender biases in academic settings.”

Author details

1. Xiang Zheng

   Xiang Zheng is in the Information School, University of Wisconsin-Madison, Madison, United States

   Contribution

   Conceptualization, Data curation, Software, Formal analysis, Visualization, Methodology, Writing – original draft, Writing – review and editing

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-6619-5504

2. Haimiao Yuan

   Haimiao Yuan is in the College of Education, The University of Iowa, Iowa City, United States

   Contribution

   Conceptualization, Data curation, Software, Formal analysis, Investigation, Visualization, Methodology, Writing – original draft, Writing – review and editing

   Competing interests

   No competing interests declared

3. Chaoqun Ni

   Chaoqun Ni is in the Information School, University of Wisconsin-Madison, Madison, United States

   Contribution

   Conceptualization, Resources, Data curation, Software, Formal analysis, Supervision, Funding acquisition, Validation, Investigation, Visualization, Methodology, Writing – original draft, Project administration, Writing – review and editing

   For correspondence

   chaoqun.ni@wisc.edu

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-4130-7602

• 3,032

  views

• 362

  downloads

• 14

  citations

Views, downloads and citations are aggregated across all versions of this paper published by eLife.