^for their main effect of interest? I only took a look at Table 2 yesterday.
The Gender Problem in Economics is SOLVED!
-
^for their main effect of interest? I only took a look at Table 2 yesterday.
The main result of 8pp that is mentioned in the abstract has p=0.089. If they had just said that on twitter everyone would have switched from believing it to saying "that's not statistically significant".
-
^for their main effect of interest? I only took a look at Table 2 yesterday.
Yes for the main effect quoted in the abstract. This is the kind of paper that says p>0.10 is "marginally significant", just read the parts describing the results once controls are included.
-
What is the identification strategy here?
It seems like many things vary between the role models, not just gender. In particular, I imagine their personalities differ. This seems pretty important in their setting.
Is it a response to gender, or to charisma differences?
If they had 100 role models instead of 2 it might be more believable.Why the no good?
When I think of the male professors in my department - old, white, grey hair, boring AF, theorists - and compare them to the female professors - youngish, white, long hair, applied economists - there's a vast difference... and not just their gender.
Is it surprising that young people listen to a dynamic young person, and not to boring old fart?
I do not see why gender is relevant here. It seems it is only relevant because it supports a specific world view.
-
What is the identification strategy here?
It seems like many things vary between the role models, not just gender. In particular, I imagine their personalities differ. This seems pretty important in their setting.
Is it a response to gender, or to charisma differences?
If they had 100 role models instead of 2 it might be more believable.Why the no good?
When I think of the male professors in my department - old, white, grey hair, boring AF, theorists - and compare them to the female professors - youngish, white, long hair, applied economists - there's a vast difference... and not just their gender.
Is it surprising that young people listen to a dynamic young person, and not to boring old fart?
I do not see why gender is relevant here. It seems it is only relevant because it supports a specific world view.Please read the paper. The treatment group had presentations by two female almuni. The control group had no exposure to the role models. There is no difference between role models to measure, because everyone in the treated group was given the same treatment. The authors address the lack of male role models and discuss what this means for the study and their identification in the paper. They also control for the gender of the class instructor.
-
What is the identification strategy here?
It seems like many things vary between the role models, not just gender. In particular, I imagine their personalities differ. This seems pretty important in their setting.
Is it a response to gender, or to charisma differences?
If they had 100 role models instead of 2 it might be more believable.Why the no good?
When I think of the male professors in my department - old, white, grey hair, boring AF, theorists - and compare them to the female professors - youngish, white, long hair, applied economists - there's a vast difference... and not just their gender.
Is it surprising that young people listen to a dynamic young person, and not to boring old fart?
I do not see why gender is relevant here. It seems it is only relevant because it supports a specific world view.Please read the paper. The treatment group had presentations by two female almuni. The control group had no exposure to the role models. There is no difference between role models to measure, because everyone in the treated group was given the same treatment. The authors address the lack of male role models and discuss what this means for the study and their identification in the paper. They also control for the gender of the class instructor.
The point of my post is that the true counterfactual - exposing students to a male role model - is unknown, and the Control here is not suitable as a control. Multiple things are varied between the control and treatment - the exposure to a role model is the sum of potentially hundreds of effects, a consequence of attractiveness, personality, charisma, gender, social class, income effects, information effects.. all of which the literature highlights as important, and none of which you can identify. Especially not the gender element.
Why does the paper focus on gender? Why not personality? Why not charisma? If the focus is gender, it seems strange not to include a comparison of the female effect to the male effect. The male effect might be double the size and in the same direct. It might be half, and in the opposite direction. This will depend PROBABLY NOT ON THE GENDER, but some other important characteristic - I imagine charisma. What can you say about female role models' impact womens' enrollment in economics if you don't have the counterfactual male impact?! (hint: NOTHING)
The next issue I have with this paper, is that it is in effect not an audit study, but is trying to be an audit study and is sold as an audit study.
There are two approaches in the audit study literature: this paper fails to correctly implement either. In the first, auditors (in this paper, charismatic professors) are usually 'matched' on observable characteristics - looks, height, weight, skin tone, personality, etc - with one aspect varying, e.g. gender. For the second approach, you have many, many auditors with high variablity in observables and then control for their observables in the regression to try and net out the gender effect.
This paper does NONE of those things. Where are the other observable characteristic controls for the professors (attractiveness, height, weight etc)? They're obviously excluded because you have N=2 auditors. No variation whatsoever. There are so many audit studies that are far better than this one, conducted with far better methodology, with far better results that are published no where near as highly because the results are not sexy.
At the moment this paper has an EXPOSURE TO PERSON EFFECT, and even that is not exactly clear. There is no causality in this paper, not a single effect has been identified. At best, it has identified the sum of 100 effects associated with being exposed to a person. However, it is in a top journal of a discipline obsessed with causality.
-
Writing from my phone. I've been here since a time when barbasol was still in your hotel room.
I was 100% here for the rupture scandal and I know ejmr can deliver on men like schiraldi too.
If you don't see that since a few months some people get over excited buy gender stuff however you don't look on the right direction.
That paper to me is ulrm managing to get an aej with an imperfect paper on a timely topic. They are fully transparent. I celebrate. Well done to them.
On EJMR we have always fought for:
1- papers with real implication on economics
2- papers not controlled by the HRM cabal
3- papers that share their data
1- is very clear, for 2- one of the authors is based in Lancaster UK and for 3- the data is available
Moreover, many people here have long argued against quotas and in favour or more incentive-based way of tackling gender disparities.
This paper suggests one. It needs exploring, I am not 100% convinced by everything, but it clearly goes in the right direction.
And I can see the argument for role models when you are already enrolled in a principles class. It makes much more sense than teh LaCour story which was about meeting random people at the door.
So I don't get what makes some people so upset about it. I hope it has nothing to do with the gender of the authors.You represent yourself as someone who has been on EJMR for a while. If that were true, you’d be able to think of plenty of times when we criticized the research of men, and you’d know this wasn’t about gender. The sexism accusation is how the Ruptures authors tried to dodge our criticism as well. I suspect you just arrived here from EconTwitter.
This paper was published in a major journal and it is being heavily promoted by EconTwitter, yet it appears to have serious problems. This case kind of reminds me of the Wu paper. Do the authors of this paper have HRM connections? -
The point of my post is that the true counterfactual - exposing students to a male role model - is unknown, and the Control here is not suitable as a control. Multiple things are varied between the control and treatment - the exposure to a role model is the sum of potentially hundreds of effects, a consequence of attractiveness, personality, charisma, gender, social class, income effects, information effects.. all of which the literature highlights as important, and none of which you can identify. Especially not the gender element.
....
At the moment this paper has an EXPOSURE TO PERSON EFFECT, and even that is not exactly clear. There is no causality in this paper, not a single effect has been identified. At best, it has identified the sum of 100 effects associated with being exposed to a person. However, it is in a top journal of a discipline obsessed with causality.
I agree with you on some dimensions, but not all. There are two issues you raise:
1) What is the treatment effect being measured?
2) What is the appropriate counterfactual?On point 1) I generally agree (except with the assertion that "there is no causality in this paper"). The authors (causally) identify an "exposure to person" effect. They separate the empirical analysis by gender of students, and claim to show that there is an effect for females but not for males. Their interpretation is that this has to do with gender, but they cannot rule out other factors - maybe the speaker had non-gender characteristics that resonated more with females than males. I think this alternative explanation is implausible, but you are correct that to definitively rule it out more work would need to be done.
On point 2) I disagree that a male role model is the appropriate counterfactual. If we think of a regular, run-of-the-mill introductory economics class there are zero external speakers. So that seems like a good control. Ideally, the authors would have had a control group of zero role models and treatments of both genders that control for other characteristics, as they describe in the paper. However, this is infeasible unless they hired actors or had an extremely large set of classes to work with.
-
On EJMR we have always fought for:
1- papers with real implication on economics
2- papers not controlled by the HRM cabal
3- papers that share their data
1- is very clear, for 2- one of the authors is based in Lancaster UK and for 3- the data is available
Moreover, many people here have long argued against quotas and in favour or more incentive-based way of tackling gender disparities.
This paper suggests one. It needs exploring, I am not 100% convinced by everything, but it clearly goes in the right direction.
And I can see the argument for role models when you are already enrolled in a principles class. It makes much more sense than teh LaCour story which was about meeting random people at the door.
So I don't get what makes some people so upset about it. I hope it has nothing to do with the gender of the authors.How about
4- No overselling of results
-
The point of my post is that the true counterfactual - exposing students to a male role model - is unknown, and the Control here is not suitable as a control. Multiple things are varied between the control and treatment - the exposure to a role model is the sum of potentially hundreds of effects, a consequence of attractiveness, personality, charisma, gender, social class, income effects, information effects.. all of which the literature highlights as important, and none of which you can identify. Especially not the gender element.
....
At the moment this paper has an EXPOSURE TO PERSON EFFECT, and even that is not exactly clear. There is no causality in this paper, not a single effect has been identified. At best, it has identified the sum of 100 effects associated with being exposed to a person. However, it is in a top journal of a discipline obsessed with causality.I agree with you on some dimensions, but not all. There are two issues you raise:
1) What is the treatment effect being measured?
2) What is the appropriate counterfactual?
On point 1) I generally agree (except with the assertion that "there is no causality in this paper"). The authors (causally) identify an "exposure to person" effect. They separate the empirical analysis by gender of students, and claim to show that there is an effect for females but not for males. Their interpretation is that this has to do with gender, but they cannot rule out other factors - maybe the speaker had non-gender characteristics that resonated more with females than males. I think this alternative explanation is implausible, but you are correct that to definitively rule it out more work would need to be done.
On point 2) I disagree that a male role model is the appropriate counterfactual. If we think of a regular, run-of-the-mill introductory economics class there are zero external speakers. So that seems like a good control. Ideally, the authors would have had a control group of zero role models and treatments of both genders that control for other characteristics, as they describe in the paper. However, this is infeasible unless they hired actors or had an extremely large set of classes to work with.1. You might think it is implausible, but you have no evidence whatsoever to suggest otherwise. So many things could be driving the result (as suggested) and are just as plausible - there are large literatures studying the things I have outlined. They are not picked out of a hat.
2. The male role model is of course the correct counter factual in this case - the paper is titled, ''The importance of female role models''. This implies two things i) gender matters and ii) role models matter. The only way you can identify the importance of gender is by a direct comparison of female and male, controlling for all other important observables (as outlined). The only way you can identify the role model effect is by exposing the treated to many different role models. If the effects are identical by gender - gender doesn't matter. If the effect is on average zero or specific to certain auditors, then it is an auditor effect - a personality effect.
This paper is massively oversold and identifies exactly nothing to do with gender, yet, the entire paper is about gender. This paper could easily have been written as 'the importance of white role models' with the EXACT same results but that 'boys are not impacted by white role models'. It could even be framed as 'the importance of role models who have hair'.
NB: I assume both the role models were white/had hair, please substitute color/hair having as required.
-
The reason women get attacked here is because they are the ones publishing all this garbage on gender gap and fertility stuff which is not economics. They’re the ones that opened the door to turning economics into sociology and now the “intersectionalism” and race BS has come with it. I used to be very positive on having more women in economics, but who else is to blame? White American women/liberals are 90% in “labor” (fertility/gender) Econ or development, where only feel good results publish.
I checked out of academia 2-3 years ago and it is clear to me I made the right choice.
-
That paper to me is ulrm managing to get an aej with an imperfect paper on a timely topic. They are fully transparent. I celebrate. Well done to them.
I agree entirely. Although I do think the earlier poster that looked at their data and finds discrepancies in the "econmajor" variable should be addressed. I believe there is a fully reasonable explanation, as everything else looks transparent. But given it is an aej, there should at least be an explanation.
-
The problem is that we use mathematics, every STEM discipline that uses mathematics has the same problem. the solution is simple either we stop using mathematics or something needs to be done in K-12 about female mathematics education. It is clear that the humanities people behind gender equity want us to stop using mathematics and we therefore have an impasse.
You've stumbled on the right problem but the wrong solution. The reason why women don't pick math heavy fields is because they're more work. You can't bulls**t math. We don't have to work as hard to have a nice life and our career success is not so heavily weighted in our social status as it is for moids.
-
If the college major in question was astronomy, the AEJ would not have published the article. IOW, it's not economics - it's a low-grade psych article that only got published because of the current political climate, and that's a really sad reflection on the AEA.
This is absolutely correct.
If it was well done - the methods state of the art, and the conclusions drawn from good evidence - I would have no issue with the paper... but it isn't. This paper is only published because it's about gender and it is written by two women.