Hey Shanling,

Thanks for an interesting project. I agree with Shalaka and Chin Man's helpful comments, namely as it relates to potential selection bias due to using OECD countries. In addition, I wanted to highlight a few points as potential ways of strengthening future iterations: (1) mentioning the hypotheses earlier would help readers know what to be on the lookout for as they progress through the next; (2) renaming the variables to something more intuitive would also improve the comprehension and discussion of your work; (3) and lastly, it is probably wise to investigate your rather large coefficient for the gii_gii variable. This could be due to perfect separation or missing data in certain cells perhaps. One way around this could be to check and present the distribution of this variable and then consider collapsing or recategorizing the measures accordingly. I nevertheless liked your use of residual plots and the confusion matrix as visuals, and I found your discussion on limitations, specifically with regards to the small observations, to feed well into your suggestions for future research.

Cheers,
Julia

Hi Shanling,

Interesting research question and you went about your analysis in a methodical way. A couple of small suggestions- 1) would it make sense to have three categories for the variable for the legal status of abortion (available on request, available in certain circumstances, not available) to incorporate the in between cases in your main model? 2) your finding on gender inequality is indeed (as you have pointed out) counterintuitive. It would be interesting to have some understanding of how this index - gii was constructed to better understand this. 3) While you have explained why dr_ig would be a good explanatory variable, you have not really justified why you have chosen most of the other explanatory variables. 4) OECD countries, I would think are not very representative of these dynamics at a global level. I agree with Chin Man's comment on how all of them even being a part of the OECD speaks to the globalisation element, and creates a specific, skewed sample. 5) Maybe you could include religious influence or something that could act as a proxy for it, as a control, as a lot of this legislation and narratives around abortion that determine its legal status are underpinned by religious leanings and how strong they are.

Hi Shanling,

This is a thought-provoking analysis on one of the most controversial issues in society nowadays. Here is my opinion about this research:

1. Could I know more about how the dependent variable classifies the “grey cases” (e.g. countries which allow legal abortions only in certain but not all circumstances)? Related to this how does
this variable define cases in which abortion legalisation is decided (mainly) below the national level (e.g. the US)?

2. What is the reason for specifically focusing on the OECD countries instead of on a broader geographical scale, given one of your independent variables is the degree of globalisation (which
by definition is a phenomenon concerning the whole world) a country experiences? Since all the countries in your sample are predisposed to being members of a multilateral organisation (i.e. the OECD), the level of globalisation of these countries may also be predisposed to a certain degree and thus this variable may have relatively small variation compared to a sample size with larger geographical coverage.

3. The marginal effects of the main independent variables could also have been presented for a more substantive interpretation.

4. Do you think it will be more suitable to operationalise the legal status of abortion in countries as a nominal instead of binary variable? Why or why not?

Best,
Chin Man