 So hello and everyone and good afternoon. I am Martina Quejeta from Uruguay and I'm going to present the first chapter of my PhD thesis at Universidad Nacional de la Plata in Argentina that I finished thanks to the UNU-Wyther visiting program last year under the mentorship of Rodrigo Oliveira that's here in the audience. And the paper is entitled Impact of Female Peer Composition and Gender Norms, Perceptions and Skill Formation in Secondary School. So just to start with a brief motivation of the paper, the analysis of the importance of gender norms on the labor market outcomes has received a great deal of attention in the last decades. Gender norms comprehend behavioral expectation according to gender, what we expect by women and men, and also the relationship between genders in each society. And by that and shaping preferences and also skill formations, gender norms may become an obstacle to girls' educational achievements and also affect labor future economic outcomes and also educational outcomes. And we have a lot of evidence showing that in fact gender norms are affecting that labor outcomes. But in why pioneer theoretical frameworks based on human capital factors explain a decreasing portion of the gender gaps in education and in the labor markets in our countries, there is an emerging literature showing that, appointing that gender norms and the stereotypes in our societies could help to understand the persistence of the gender gaps and the gender inequalities. Yet, how gender norms perceptions are formed and what drive changes over time of the gender norms are still open questions. Gender norms formation process involves many actors along the life cycle and while families are key at the initial stages of our lives, the school and peers interactions are increasingly important with the age of the individual. Peer gender composition might affect our perception on the gender norms by at least two different channels. On the one hand it can affect social attitudes and behaviors between groups due to the diversity in the class composition. And there is evidence showing that the higher exposure to, for example, poor students, leads to more pro-social or more egalitarian attitudes from the rich to the poor students. And on the other hand, it can also affect the interaction between teachers and students. And there is also evidence showing that there is in fact a transmission of the gender views from the teacher to the students. So in this context, my research question is, how gender composition in secondary school affects students' perception on gender norms? And with this question on this paper, I'll be contributing to, on the one hand, to the peer effect literature. We have a lot of evidence showing that, sorry, there is a lot of evidence for developing countries and there is well-established stylized facts for developed countries. But the literature for developing countries is scarcer and this is mainly due to stringent data requirements required for credible causal strategies in our countries. But even for developed countries, the literature addressing the gendered impact of the peer effect literature is still scarce. And moreover, the effects on skill formation have been widely studied by little, we don't little about the transmission of gender norms within the school context due to peer effects. This paper is also related to the gender norms literature. The gender defects of the gender norms on economic outcomes has been widely studied as I mentioned in the motivation slides. But literature, not how are they formed and what drive the changes over time is still scarce. We know there is a family intergenerational and transmission of gender norms within families. We know that there is a transmission of gender views from teacher to students. But to the best of my knowledge, there is only one related work in paper from Javier García-Brazales, analyzing the school peer effects on the gender norms for the Vietnamese case. The main differences between that paper and mine is that for the Vietnamese case, we are going to have the long-term effects with Javier analyzes the effects of peer school composition on later educational outcomes and labor outcomes. And in this paper, I'm going to show you the contemporaneous effects of the peer composition. And also I analyzed this question from an interesting setting that is your way that is characterized by more progressive gender views within the developing countries context. And also the study of the gender inequalities within the school context is important from a policy perspective, not only from a justice perspective due to the equality of opportunities among genders, but also from a justice, sorry, efficiency perspective. That is, we know that there are long-term effects at the individual level that are affecting women's educational outcomes, career choices and labor outcomes in the future. And that can affect the allocation of talents in our societies and does affect also the economic growth after all. So just to give you a brief institutional context of the setting I'm going to analyze in this paper, in your way, mandatory basic education includes 14 years of former school and including two years of preschool education. Typically, at age 12, students enter secondary schools that comprehend six levels from the seventh to the 12th grade. The admission to public secondary school is based mainly on a geographical criteria and thus students and families can choose the school but within a limited range of options due to the region or geographic area of residence. But when opting for a private secondary school, families can choose any school without any external restriction. But nevertheless, in both public and private secondary schools, the assignment of students to classes are made by the head of the schools aiming to seek some homogeneity within the groups in the school. So some interesting point of the Uruguayan context is that the students are reassigned to different classes every year at the beginning of the school year and they share all curricular activities throughout the year. So the peer interaction I'm going to analyze in this paper is intense because students share the same curricular activities and the same room within the year but the interaction is short because each year students are changing classes within the same school. So the data I'm going to use in the paper is the Uruguayan survey for a multi-dimensional evaluation of the educational system. They are named Arista's survey and they are representative of the nine great secondary students attending urban schools in my country. The novel sampling design is basically a three-stage sampling design where first schools are randomly selected, then classes within those schools are randomly selected and finally all students within classes are gonna be interviewed by this survey. And the field work was done in October 2019. So for the estimation sample, I'm gonna take only students from public and private regular secondary schools. I'm gonna drop out of the sample the technological secondary schools. I'm gonna use only information of non-missing information in the analyzed variables. And I'm gonna only take classes with more than 12 students that are in the sample and only schools with two or more classes and this is due to the methodology I'm gonna use that I'm gonna explain you in a minute. So I'm gonna take around 60% of the original sample and in the paper I provide some descriptive statistics showing that there is no sample selection or that the final sample, it's basically equally distributed that the original sample. So just to show you some main descriptives of the data, female students are younger than male students. They are less likely to live to be only child and to live with both parents and grandparents but more likely to pertain to the lower socio-economical households in the country and they are also less likely to have an experience previous repetition in previous weights. The main independent variable in this case, it's gonna be the share of female peers in the classroom that is around 51 or 52% in the sample and the means are different because I'm gonna compute this share of female peers as they live one out to distribution that is taking out the student's hair or himself from the sample and something important here is that I'm gonna compute this variable taking the original sample and not the estimation sample and that these two account for the actual sex composition of the classes within the sample. So regarding the main outcome variables that are the student views on gender norms in this figure here, I show you the main distribution level of agreement of the students regarding six statements on gender norms. They are asked about six gender norms such as, which is a pointer here, here they are asked about the level of agreement with some statements such as women shouldn't work, men and women should earn the same wages, men are better in politics than women, women should perform all the domestic works, women are the primary caregivers and women and men should practice the same sports and for this figure and the whole analysis, I rearrange the outcome variables so that higher level of agreements like the green bars here represent more progressive gender views for the six statements. We can see that for wages, sports, they are the two statements with the higher level of agreement that is the more progressive gender views but for the six statements, female students hold more progressive gender views compared to male students in the sample. So the empirical strategy basically follows the widely used linear in mill models where the share of female peers is the live one out proportion of female peers in the classroom. This takes a vector of students' control variables and these the same variables taken as peer variables like mean within the classroom. This lambda coefficient account for school level affects effects to control for and observe school quality characteristics and also to sort out potential selection issues from students to the schools. And the error terms, the error are clustered at the class level to allow for correlation of the outcomes within the same classes at the school. So basically the empirical strategy exploit the quasi-random variation in the percentage of female peers across classroom within the same school for the identification of causal effects. So basically the main identification assumption here is that the main independent variable, the share of female peers in the classroom is not correlated with the error term. So basically the main challenge to actually identify causal effects is the potential sorting of the students and families to schools. And to deal with that or minimize that bias, I include the school level fixed effects that are possible due to the three stage sampling design of the survey. But there could also be some potential selection issues or self-selection of the students within the school but into different classes and to test if the assignment of the students to classes is as good as random I will perform two alternative tests. First I'm gonna follow the Laby and Schlosser methodology and I will randomly simulate the assignment of the students to classes and then I'm gonna compare the observed or the actual distribution of the students to classes with this randomly simulated distribution and then I'll also test if the student characteristic by the proportion of female peer that is to check if classes with different proportion of female peers have the same predetermined characteristic in my sample and the results for both test supports the identification strategy I'm gonna use in this paper. And finally I check if after controlling for school fixed effects and individual control variables and peer control variables, I still have enough variation in the share of female peers among classes within the school that is gonna be the main independent variable and the test also supports this identification strategy. So going quickly into the main results, this table here reports the main results of the paper that is the effects of the proportion of female peers among on gender norms. We can see that almost for four out of six of the statements the coefficients are positive and significant and there is only two exception that it's wages and sport that if you remember where the two statements with the higher level of agreement or more progressive gender views in the whole sample that is they have less marching for improvements for this under the circumstances. And so the gender norm index that is an average of the other six statements. It's positive and significant. And so their main result here is that the higher exposure to female peers in the classroom lets to more progressive views among secondary students in the sample. And the magnitude of quite a magnitude is important and because if you remember we are talking about contemporaneous effect like immediate effects of interacting with female peers and we are talking of only one year interaction. So it's quite short compared to other previous papers. Yeah. Share of female. So is this an increase of 1% or an increase of 1%? 10 percentage points, yeah, sorry. So an increase in 10 percentage point in the share of female peers in the classroom lent to a 3.84% of a standard deviation increase in, yeah. So then I split the sample between by student sex and we can see here that while for both male students are and female students, the coefficient is positive. It's only significant for male students. And so the baseline results are driven mainly by male students that it's important because they were the ones holding more traditional gender views among the whole sample. And this is true for almost all of the separate statements. I then analyze other heterogeneous effects by other predator mind characteristic of the sample and we can see that while for repeaters and by socioeconomic level, both coefficients both are positive. It's only significant among a student with previous repetition and pertaining to the 40% of lower socio-economical context. And it's also mainly explained by students living outside the capital city in my country. So just moving to other outcomes that I analyze in the paper to analyze the consequences of female peer effects. I exploit one dimension that is also asked in the survey that is a frequency of which students declare helping with domestic chores within the household. So students are asked about the frequency of which they help cooking, clothing, washing clothes, cleaning or caring for siblings or other family members. And then I compute a total index of domestic or time devoted to domestic chores. And what we can see here is that while for male students there is no significant effect of interacting with more female peers in the classroom, there is a significant effect for female students that are reducing the time devoted to caring activities and total domestic work activities. I also analyze the peer effects on some cognitive skills. I take advantage of the survey that ask also that within the survey, they also perform two standardized cognitive tests on mathematics and language. And one important aspect here is that the tests were designed by the survey and it's also qualified by the survey itself. So it's unlikely to be biased by teachers by the teachers. And so what we found is that there is no significant effect among male students and again there is a significant effect of female peers composition on female students on mathematics which is important because it's an area of cognitive skills where female students under perform male students and this is an stylized fact. So I perform some robustness checks of the baseline results. For a mother's time I'm not going to go through all of this but the results are on the paper. But I basically include teacher and class controls. I test if the missing observations on outcome variables and control variables affect the main results. I also perform the, I re-estimate the, oh sorry, I re-estimate the model for a sub-sample of schools with the higher probability of random assignment of students to classes. And I also re-estimate the results dropping one school but at a time. And the results on the gender norms index may basically hold under these tests. So just to conclude, in this paper I investigate how gender composition in secondary school affects student perception on gender norms in the Uruguayan context using a nationally representative survey in the country. And I found robust, significant and positive effects that a higher exposure to female peers in the classroom affect or led to more progressive gender views for old students. But the effects are mostly driven by male students and from students from lower socioeconomical background that are the ones holding more traditional gender views. But the effects of female peer composition goes beyond that result. And I also show some evidence that they are changing actual behavior and female students are reducing the time devoted to domestic work and also improving academic performance in mathematics that are improving academic performance in mathematics. So even short interaction in secondary school so it's just only one year may have substantial effects in reducing gender stereotypes and changing actual behaviors of secondary school students. And that's it, thank you very much. Thank you.