 hard work we've done over the past three years has finally or is coming finally to fruition with this conference and with the report and hopefully it will plant the seed for for common projects where we can learn and investigate more in-depth about this topic of the why's of the gender gap in in science Just maybe one word about myself. I'm a data scientist currently But my background is in astrophysics and in gravitational physics so that's where I come from and my interest in analyzing the publications or bibliographic data came from my prior work as editor of the central black mass the bibliographic database of scientific mathematical publications So our project was named gender gap in science technology engineering and mathematics This means that from the very beginning we we wanted to go over just one particular discipline into the general stem subjects and include all the possible fields and disciplines in within this stem broad field But this means also that when we spread our focus we need to take into account that different Disciplines are different from each other when we started this project five years or the The original analysis of the mathematics data five years ago We were very focused on the practices of mathematics and when the project was formulated to expand To other disciplines it needs to be remembered and understood that every scientific community has its Particularities and therefore any analysis we do about the gender gap has to take into account that not all Stem fields are born equal and this is why this talk is going to be focused in what can we learn about the gender gap Per discipline and I'll try to give a bit of light into these different disciplines fortunately though this project was Had the big lack of having very interdisciplinary Unions as collaborators so we could from the very beginning have input from the not only the mathematics community but applied mathematics and astrophysics and physics and a history of science and Chemistry and biology through all these unions That are specific to their field So this from the very beginning set our focus into being as broad as possible in disciplines and to study the challenges of each field independently as How the project was organized as you all know by now we have organized the project in these three different blocks one would be like maybe if you want a Button top approach so you start asking people so the global survey of scientists coming from the information that people give you a Scientist to try to make general assumptions that was task one the survey Then we had the data analysis of publication patterns the task two in which the focus was maybe the other way around like You have the output in scientific publications as gathered in this particular bibliographic databases And from that general data you try to make assumptions about the individuals that are behind So if you want this to test we're a bit complementary to each other and then we had the database of good practices as a way of compiling a specific actions that you can have and you can make to promote the More participation of girls and young women in stem in my talk I will only talk about results from the task one and task two because it was never the intention of the database of good practices to say things about the Specific disciplines maybe some anecdotes or some examples come from the examples in the database but it's mostly from task one and test two that we can say something about the Gender gap in the various disciplines. So I will talk only about the survey and the data analysis of publication patterns So the methodology as I said a little just a minute ago the methodology to address The the difference in that for in the various disciplines from these two Parts of the project was different in the global survey We rely on the self disclosed information of the respondents, which is subjective And which is individual and so it's a survey Whereas in the and and we enter the different fields of the people by means of the self disclosed field of study that they name So in the survey the disciplines comes through the Identification with this field. What is your primary field of study? That's how you say you are an astronomer or this is related to mathematics by self disclosed information Whereas in the data analysis of publication patterns We have to first say which field are we gonna study? Then we have to identify which is the bibliographic database that we can and want Analyze and then we choose the field by saying, okay, let's look at biology So let's find a database for biology. So different approaches again And how we analyze the data in terms of disciplines is also different in these two tasks in the global survey You have all your data and now you want to say something about mathematics Then you hold out all other variables constant and then say, okay Then holding everything constant and then looking only at this axis, which is mathematics. What can I say? That's how you restrict your big data to only one dimension, which is the discipline In the data analysis of publication patterns because we are doing different Analyzes of different data sources. We really had to adapt our code and our infrastructure to the different disciplines So we could not say, okay now do the same for informatics No, because we needed to like create first the infrastructure for each field So we cannot replicate one by one in all in the databases that we've analyzed in task two We cannot just change and say exactly the same for other discipline because the databases are Fundamentally different and they have different data. So again a bit of difference in methodology between the global survey and the and the task two But nevertheless with all these introductory caveats. Let's go into What have we learned about the gender gap for the various disciplines from the global survey So as I said the various disciplines in the global survey come from self-disclosed information That is in the field of study and these are the fields that people could choose from so when we say various disciplines for the survey We mean this disciplines How do we analyze them will we heard yesterday in the talk by Susan that in general we can say that The survey gives compelling evidence that women and men do not have the same experiences in the stem and Moreover that the experiences of women are less positive than men these across all disciplines But can we say something specific to each of the disciplines? To do that we need to resort to by variant analysis that can also be used to explore gender difference in differences in perceptions According to these three different axes that we've seen already regions Societal development levels that you talks that we've already heard this morning and the disciplines which is what I'm gonna say So restricting all the data to this axis of disciplines brings the results. I'm gonna highlight next Let's start with the beginning the beginning is when do you decide that you want to go into a stem field or not? for this there was this question a What is was the period of time in which the respondents chose their primary field and? Here as we've already seen when it's gray It means that there is no significance Significant difference between men and women and when there are colors is because there is some significant difference as we see here There is gray in astronomy and in physics Pointing to evidence that there is no gender gap in astronomy and physics in what refers to when people choose their field So in principle women and men Astronomers and physicists did not experience any gender gap when they were choosing their field and if you notice for this too the Majority let's say 50% of respondents made that decision Before going to university So it may be that a factor to promote The removal of gender gap in stem fields would be to encourage people to make decisions about their future Before going to university before society maybe puts more biases and more constraints into your choices But this is just a hypothesis So yeah regarding this this question there seems not to be a gender gap in astronomy and physics But there is a gender gap in the other fields because as we see for instance in computer science There is a difference when they chose their field of study Now coming to the quality of relationship with advisor, which may be another very important factor after you've studied your PhD So how would your relationship is with your basically the most important person to you during your doctoral studies? We see a lot of gray here There is no evidence of a gender gap in biology computer science history of science or applied mathematics Meaning that people on those disciplines they didn't experience a worst relationship Again, it's a perception with their advisor just by the fact of being a man or a woman So again positive things in those fields That's the thing that When you say there is no gap in these fields is for this question because we certainly know that there is a gender gap in STEM in number of participants and in so on and so forth so by looking at these plots and identifying what are the gray areas We can at least remove some factors that may explain the gap So we can say okay There is a gender gap in astronomy, but that gender gap is not explainable just by the fact that so when the women is when women and men Choose their field because they don't see you don't see that difference in this plot The same way you cannot explain the gender gap in computer science by means of the bad relationship with the advisor because there is no statistical significance that the relationship with the advisor are worse for women than for men question Yes, but all these measurements because the significant difference is that all these numbers are Have an error bar and the error bar of this number is larger than the difference so that you cannot say that You can repeat the experiment and the experiment will be different numbers and the difference between these two experiments will be larger than this difference Therefore is not statistically significant Which to calculate that you need to have this alpha the false There's this coefficient that Susan talked about Plus also the size of the of the sample when your size is very small Then a large difference is not statistically significant because just by one person changing their answer The whole percentage is completely different So you also need a certain number of respondents to be able to tell the significant statistical significance Gray again not significant and just green is women and men is Orange and then this is because the answer is yes. No, this is just the percentage of people that say yes I had a So the number is just eighty eight percent people said yes I have a good relationship with my advice Very small right Indeed again with the interruptions of during doctoral studies We don't see evidence of gender gap in these fields astronomy biology history of science. However, if we look at regarding the Consent versus averaging them out. Yeah, I mean suppose there was a difference between You know people who got their doctorates in the 70s and people who got their dog birds in the 2000s You wouldn't see that difference here. So it seems to me. It's not really holding it constant It's averaging out all the other variables So if you want to take the agent to account then you do this multivariate analysis in which you split this data also by this Perpendicular axis, which is the age and then you will have for each of these now for these blocks all the ages Yes, right, right But again, this means that Accounting for all ages and for all other variables that are constant. We don't see so these are the results So that's when we say this again This is Rachel and Susan are the ones to explain again what means to hold the other variables Thank you. Thank you The survey because it's about perceptions and because not everybody Answers to all the questions. I guess this is also a reason Has some answers that are really not consists not maybe they are not consistent with each other No in principle astronomers don't have don't see I don't see any different in women and men astronomers Regarding the interruptions during doctoral studies. However, they do report that becoming apparent had an effect in their career progression But becoming apparent is most likely an interruption of their doctoral studies. So you see is sometimes they've Yeah, yeah, but this is a subset of the of the reasons for which you may interrupt. So you You don't have a significant interruptions in your doctoral studies But the ones that had because certainly having becoming apparent is an interruption in your doctoral studies at least Write your career and your doctoral studies. Okay Okay, that's true, that's true Then okay, maybe then there is no contradiction here Yeah, but you you would see here. Okay, astronomy. There is no gender gap in astronomy regarding interruptions However, there is a gender gap in astronomy Regarding the effect of becoming apparent on career progression See therefore, it's very difficult to say what is the surveys telling us about the gender gap in different disciplines Because it depends on the question that you are asking Yes Yes, yes So before the thing was that these three this is any Interruption during your doctoral studies and this is becoming apparent which that doesn't necessarily happen while you are in the doctoral studies Mark but that's why I said the worst field is astronomy and in this plot the best field is applied mathematics But we don't know if there is exactly Finally, we can all agree that there is a gender gap in all disciplines in respect to experience or having the perception of having encountered sexual harassment and in this in this case the worst field is history of science, but I mean the fact that applied mathematics is 20 percent as a 20 points less than history of science must mean that the harassment in applied mathematics is Significantly less than history of science But again without knowing knowing the effect size without knowing how many of these people actually so how many people are in this Bar and how many people are in this bar is also difficult to say whether history of science is intrinsically worse than applied mathematics in what respects to sexual harassment. I Don't want my talk completely being like we don't know. It's just I'm saying that everything has to be taken with a grain of salt and This sort of closes the general questions and and you can see more of them in the report There was however so when we designed the task to the data analysis of publications We wanted to see if we could cross check Answers from the survey with the data from the databases and really is very difficult to check Looking at your publications whether you've had interruptions in your career. This is difficult So that's why the only question possibly from the survey that we could cross check with the analysis of bibliographic sources Was the one regarding submission to top journals because in the publication Analysis we were looking at top journals. So that's why I'm focusing specifically in this one For this question, which was how many articles have you submitted to top journals in the past? Years or in your career or no per year. Sorry. What's the formulation of the question? How many papers? I know how many papers have you submitted in your career into top journals the women reported? fewer 5.7 in average in the in the last five years compared to men So this is something that we can test because we have the top journals if you want and then we can look at how How they publish? But first let's look a little a little bit more about the results So this were the mean data so mean values 5.7 and 6.3, but this is the histogram of answers Obviously, it's not true that people like Come at 10 and then they stop policing is because when you when you quote the number of papers that you've published You tend to remember 10 maybe 11 or 12, but 10 15 20 This is like people people cluster in their mind how many papers they published So if you've published one then you remember but then when you've published nine or 11 So this is why the histogram has these bumps at these numbers but then you look at the Women which is on the if you want on the background of the of the picture is the Darker color and then men is the lighter one. Well, we know that I said it wrong So men is the light and and the women are the dark and then when it's like this intermediate It's because they are on top of each other and then you see okay You have certainly more men that publish a lot of articles We know because we men are more like the so when there are like these overachievers with hundred papers There are most likely to be men Whereas women tend to like submit a few papers to three etc at higher frequency So the difference between these two distributions we've checked is a statistically significant So these numbers five point seven six point three point to the fact that men Seem to perceive that they publish more to top yarn that they said that they said submit more to top your Yes, yes later. We will go and look at the top your nose and then we'll see what comes out of that Yeah, yes Yes, those are the caveats of this plot is perceptions And it's what's a top your now? So that's why This is like a first like from the survey respondents and now we're gonna go and check it with the data Everything fine. Can I continue? Yes Again with the caveat that when people submit to top journals There are there are things that are different across disciplines for instance the the order of authors per field It's very different in mathematics where you mostly publish in alphabetical order than in other fields where you have this relative contribution mostly on all of them minus mathematics or even the Negotiated order and so on so we see we know also that the publication practices in different fields are different But taking that into account. We went and wanted to check this particular data Against the data the databases that we analyzed So that's why now I move over to the analysis of publications And I start talking what have we learned about the gender gap for various disciplines from the analysis of publications And when we say various disciplines here is not the disciplines that the people self-reported Because we don't have access to we couldn't analyze biology chemistry and Computer science so we could only analyze theoretical physics through the archive mathematics through the CB mass database astronomy and astrophysics through the AGS database and Chemistry only through us selection of six top journals. So chemistry only partial. I Don't need to say why we analyze publications because Helena said it yesterday They are important for your academic career Your tenure and promotion may depend on them and they are very relevant for science policymakers, etc So that's why I analyze them and we chose these ones because we could because we had access to them because they were Open or partially open so everything that Helena said yesterday if we wanted to expand which we certainly want We would complete chemistry with the whole cast Registry the chemical abstract service We would add biology via the bio archive, which is the preprint service for biology similar to the archive And we would add certainly computer science via the DB LP database, which is a database in the University in Germany that is now run by a Leibniz Institute So certainly our analysis can and should be completed But in what I'm saying now I have to restrict to the data that we could analyze in these three years And here are the top journals. I think this this plot was shown yesterday by Helena These are ten journals in math that you can call top because they are quite prominent They in on the left. We have mostly journals from societies like the American mathematical society in France London and so on whereas in the right. We have like more Topic specific prestigious journals and I think my Helena already mentioned yesterday. We see Like wait in some of them quite a sign of of so let me explain first what we are seeing by publication year We are seeing the fractional The fraction of authorship by women in these journals. Okay, so it means in 1990 in this nice journal, we had nine percent of all authorship that were by women Okay, and this tendency. Okay, you can say in the 1970s. There were fewer mathematicians. So then it's just normal Maybe they were only five percent of the all the mathematicians and it's normal that in top journals They are so little fine, but 40 years later. We have three times as many mathematicians that are women So you should be seeing a positive trend and rather the opposite in the AMS Journal we see an stagnation or even a decline in Benzio's Mathematica is also quite flat I mean we I wouldn't call this progress Progress if we consider that the number of women mathematician has has tripled over the past 40 years So I would expect that in absence of any bias the number of women that publish in these top journals follows the general trend of the increased number of women mathematician overall if it doesn't happen is because To two explanations women are not good enough mathematicians to submit and publish to top journals Which we should not accept or that the submission and acceptance processing top journals is biased But we lack so much data and so much insight about these practices that is very difficult to say what the reason is for this data Let me I know I don't have I mean lunch is waiting for us. So I Don't have to say everything again. These are the six top journals for astronomy and astrophysics in which we know that So if you take only the six you are covering a big big chunk of the Astronomy and astrophysics community here. We do see a positive tendency So we see the percentages of women authors going from something between five and ten percent To in the 70s to something close to 20 percent in the in this century Which sort of reflects and mimics the growth of the field the growth of the percentage of women in the field So here I would say that the journal the top journals Yes, they are representative of the poor population So I'm more inclined to think that there is no bias or no implicit bias or Subjective bias when women submit to astronomy In theoretical physics, this is a disaster pretty much having journals that have 0% of the articles of one year by women and I cannot believe that there are 0% physicists Submitting to them in one year. So it's either women think this is a very high journal and my I will never get the yes again, what what what she said is if in our in our In an ideal world all the journals have all the representation of women that has that mirrors exactly the reality Maybe some some journals have more woman Maybe less woman but on average they would mirror that but then you start picking like the top from the This rankings if you want or from just asking Experts tell you what are the good journals and then on those There was the fraction of women is significantly less than on and on the reality The number of people that actually publish in physics, then you see a hint of for a bias Yeah, we don't know that because publishers will never give you this data So it's my conclusion of all this is that we need transparency into submission rates If we cannot get them from the publishers then we should by all means analyze grant proposals because those are public and those Go to public agencies and we should ask for the data from grant proposals And when this has been done the results as it is it that women publish submit already less Maybe because of discouragement and when they submit late there are also granted grants less So it's in both of these stages at least in grants I cannot say anything about your nose because I don't have the data But I suspect those two aspects are Confounding here and is that you lack the motivation or the support or whatever to first submit And then when your paper or grant is there you also face some bias from the editorial board because Studies have been shown that both female and women if the female and male reviewers Are harsher with women's papers both so I'm not saying that is because the editorial board is just meant I'm saying because in general this is there is the subconscious bias than a paper by a woman is intrinsically less or she has to prove more Yes, blind blind review and blind submission, of course. Yes Chemistry is a good surprise. Mark tells us is not surprising. It's not worthy because chemistry also is for those more or less The trends of astronomy and astrophysics So if you were to tell me what are my results what have I learned about the gender gap through the analysis of top journals Is that astrophysics and chemistry don't seem to show the large a gender gap whereas mathematics and physics do show our gender gap and The applied fields seems to have larger female representation than the pure fields This is seen here in in the journals that are applied here have a better So for instance the CM journal of mathematics have better trends than the pure fields But mostly what I learned from this is that we need More transparent data on this process because biases that exist are unknown and immeasurable Let me continue now we went from this top journals to the whole database and the first thing we did was to Go to every cohort so every year there are new authors that start publishing in the database, you know the first year the first paper by Marie-François was in this year So she in some year and then she is in the cohort of that year And then we compute per year how many of these authors that start publishing this year Which is a proxy for your first paper is a proxy for your PhD stage sort of so how many people enter the field per year? And those that percentage of women has been steadily increasing from 10 percent to 30 percent That's what I meant that we have tripled the number of women Entering mathematics over the past 30 years So that's why that this is the plot that doesn't correlate with this right because if you have 30% women today in mathematics, you cannot expect that only 8% of them publish in top journals. Okay, you can say they are too young today fine But then why are the numbers so stagnant since 30 years ago? See so this is what this is what we get from this analysis compared with the top journals I see a gender gap in the publications in top journals So I see a bias when it comes to entering really the high ranks of your profession. Maybe the glass Sealing effects all those things in astronomy and astrophysics though We see again the same trend from 10% in the 70s to 25 30 percent today By the way here is the number absolute numbers So you can see that mathematics is a field in which approximately 10,000 people enter new every day Whereas astronomy is a field that is about half the size of mathematics today But the trend of astrophysics is mimicked in the top journals of astrophysics So this is why I say in astronomy and astrophysics I don't see the gender gap in what comes when it comes to the publications in top journals and then theoretical physics 8% to 20% again I see the bias or I see a gender gap because it doesn't correlate with these flat lines here Moving on we can try to think of an explanation What is different in mathematics from astronomy and from theoretical physics? Certainly the size of the teams Mathematics has been so what I'm showing here is per year What's the percentage of papers that have one author, two authors, three or four plus And then I look at the years So this is mathematics on the top You can see that in the 70s 80% of the mathematics papers had one author 19% had two and then the few percent had three or more So it's a field in which you write your paper And now in these years now it has become a field in which you write your paper you and your friend Maybe you're your friend and a third friend But this is it, it's a small team community So ties are very important ties to your doctor advisor Ties to your very small university team And it's you and your friends so to speak Now we look at astronomy and astrophysics In the 70 only 30% of all the astronomy and astrophysics papers were written by one person We know that in astronomy you is you and your I don't know your other 500 friends So it's very different And what does this foster? Maybe this fosters less fear to be in collaborations Or to be out there and publish less More inclusivity because the teams are big I don't know it could be an explanation But certainly when things are closed And gate wait and there are few people in power This is advantages for women So opening the field, opening the collaborations Maybe positive to close the gender gap But of course I cannot say here Let's all mathematicians publish papers with 100 people But you know what the trend is The trend is maybe more collaboration, more diversity, more inclusivity Helps close the gender gap Doesn't go that well with theoretical physics Because theoretical physics is sort of in between maths and astrophysics So there are certainly big collaborations in theoretical physics But not as large as the astronomy ones So what can we say? This is as bad as mathematics But this is still a bit more collaborative field than mathematics So yeah, we need to see I cannot give completely for sure answers But because the time is ticking Let's go to my summary This was a big project This was a very complicated project And this is of course a very serious and hard problem That is very difficult to solve with our just little insights And also when it was designed It was designed in this way That the survey answers some questions The task to answer some other questions And it has been difficult to find consistent answers From the project Because what people perceive And what people self-report And also the way the survey has been designed With this snowball sampling May not be the full reality It's not, it cannot be But also the analysis of the publication databases It's not the full reality That's just the manifestation of people That publish a lot But we know that there are many scientists Physicists, mathematicians, biologists, etc Whose main focus is not on publishing Maybe it's on teaching So they are also not reflected In our task too Therefore it's just normal And to be expected that these results Are difficult to reconcile And none of the two pictures that we get From the survey and from the task too None of them are complete All of them are incomplete And this is it because reality is very difficult To model However there are some things that we can say Because if not that would be very bad So from the survey We can say for instance There is no perception of inequity Inequity across fields Regarding submission to top journals But from the analysis of publications We see a mismatch between the percentage Of total female authors And female authors in top journals So we know from the data Of the analysis of publications That there is some bias in publication To top journals, bias that is not seen In the self-reported perceptions So maybe a bit Of continuation of the project Could be to try and reconcile These two views of the Of the problem Again, I say it again That the difference is in perceptions That were extracted from the survey Do not necessarily align with the results From the publication analysis But anyways from the point of view Of the scientific outputs So from the point of view of the publications Astronomy and astrophysics Present the smallest gender gap Of all three analysis sources So that we can say That we've learned from the gender gap Across disciplines That there is a larger gender gap In mathematics and theoretical physics Than in astronomy and astrophysics Regarding the publications And maybe a potential explanation For it is that astronomy and astrophysics Is a more collaborative And inclusive field If you want like with larger teams That plays an important role Into making the Coming of new women Easier I'm going to leave some 10 minutes For questions, comment some feedback And I thank you very much for your time and attention Thank you Lucia