Gender, Education, and Attitudes toward Women’s Leadership in Three East Asian Countries: An Intersectional and Multilevel Approach

Round 1

Reviewer 1 Report

The manuscript was a real pleasure to read and offered important insights on the issue. However, there are a few limitations and shortcomings which hamper your arguments and influence the academic soundness of your article:

1. The hypotheses are unclear and are not constructed as research hypotheses. Please revise them and make them easier to understand. You can even transform them into research goals or objectives, but they do not help the manuscript being presented as hypotheses.
2. A considerable amount of your main research (consisting on tables with statistical tests) are not included in the manuscript and are presented as supplementary material, while less relevant and useful information (descriptive statistics) are included in the manuscript. I would advise rethinking what parts will be included ans what will be excluded from the manuscript.
3. Figure 1 and Figure 2 are not readable. The text and image are too small to comprehend.
4. Although the manuscript is written in a clear, concise and correct way, there are still small language mistakes (typos) or sections which could be rephrased in order to ensure better clarity and a few formatting issues. I would recommend proofreading the manuscript one more time.
5. The conclusions and policy recommendations developed based on the findings are, to some extent, general and cannot easily serve as true and convincing policy recommendations for decision makers. The manuscript fails to make good use of the results obtained and to translate them into original and specific policy recommendations that could help decision-makers develop sustainable public policies.

Author Response

We thank the reviewer for the helpful and constructive comments. We have revised the manuscript in the following ways:

1. We have thus transformed the hypotheses into research objectives and expectations to help readers better understand the research questions.
2. We appreciate the reviewer's thoughtful comment; in fact, we were on the fence about what to include in the text and what to include as supplementary materials. Based on this comment, we have now included the regression tables (Tables 4, 5, and 6, previously Tables S1-S3) in the revised manuscript and have revised the text (line 386, lines 574-577) to reflect this change. However, as we clarify in Section 6, it is difficult--sometimes even misleading--to directly interpret interaction effects in non-linear models, so we focus on the figures when reporting the results in text.
3. We apologize for the inconvenience the format issue must have caused. We did our best to increase the size of Figures 1 and 2 within the space allowed by the manuscript formatting template and included the same figures of bigger size and higher resolution in our supplementary file. We're hopeful that the copy-editing process will resolve any remaining issue with formatting/typesetting should this piece be accepted for publication.
4. We have proofread the manuscripts multiple times and fixed the typos and wording to our best ability.
5. To discuss potential policy implications and recommendations in the conclusion section, we have added a paragraph to emphasize the importance local and grassroots knowledge (Lines 547-558). At the same time, we caution that based on the empirical results pointing to the complexity of the interactive relationship among gender and demographic groups, more and larger-scale research is needed to identify contextual factors that shape the effect of various social institutions and individual characters and how.

Reviewer 2 Report

The article is interesting but I have a couple of comments:

• the methodology is not valid for measuring properly casual relations and for giving a response to difficult answers. I would suggest to revise if you can apply other type of methodology.
• With the multilevel analysis you could get rid of heterogeneity, but you have chosen to analysis el very similar countries. Why? Wouldn’t be more interesting to have a broaden view and analyze countries from different cuktupral backgrounds?

Author Response

We thank the reviewer for their insightful comments and we address them below:

Re: cross-sectional methodology and causality

We agree with the reviewer that our chosen method does not provide direct evidence for causality, which we acknowledge as a limitation of our research in the discussion section (lines 560-566). However, our goal is not to assess causal effects; rather, we aim to describe and document the intersectional/interactive effects of contextual and individual characteristics on attitudes toward women leadership. In the absence of large scale cross-national longitudinal surveys suited for our research questions, we chose the WVS data for its high quality and standardized measures, which enable cross-national comparison. While future research certainly can and should explore the causal mechanisms of relationships identified in this paper more thoroughly through collecting longitudinal data, the cross-sectional nature of our data does not discount the theoretical contributions we make. Regardless of the causal directions, the interactive effects of gender and education highlight the value of an intersectional approach in gender attitude research and the significant variation across the three countries attest to the need to incorporate context in the construction of intersectional frameworks. In short, we chose our current methodology based on our research questions and data availability and interpreted our findings and contributions accordingly.

Re: sample size at country level

We respectfully disagree with the reviewer that the three countries are “similar”. As we detailed in Section 4, although these three countries are often grouped in studies as “similar”, there exists important heterogeneity among them which has been addressed by previous research and is reflected in our findings. That said, there are utilities and caveats in using either large or small samples at the aggregated level when conducting multi-level analyses. We do acknowledge in our manuscript (lines 568-570) that a larger sample size would allow for statistical tests of the effects of country-level variables and cross-level interaction. However, such a project would require that researchers have already identified a number of country-level factors that are likely to be influential in shaping both gender attitudes and the effects of individual characters on gender attitudes. Studies with small sample size such as this one that examines each case in depth provide crucial information toward identifying these key variables. In these studies, the benefit of keeping the cases within the sample culturally and geographically approximate is that we can limit the dimensions of variation and isolate the most likely contextual factors that contributes to variations we observe. We believe the current project is valuable not only in the theoretical insights it generates but also the empirical foundation it lays for future exploration.