arXiv subjects

Ivan Smirnov

Publications and source records attributed to Ivan Smirnov.

26 records · Page 2Linked to original sources

arXiv2020

Quota-based debiasing can decrease representation of already underrepresented groups

Many important decisions in societies such as school admissions, hiring, or elections are based on the selection of top-ranking individuals from a larger pool of candidates. This process is often subject to biases, which typically manifest as an under-representation of certain groups among the selected or accepted individuals. The most common approach to this issue is debiasing, for example via the introduction of quotas that ensure proportional representation of groups with respect to a certain, often binary attribute. Cases include quotas for women on corporate boards or ethnic quotas in elections. This, however, has the potential to induce changes in representation with respect to other attributes. For the case of two correlated binary attributes we show that quota-based debiasing based on a single attribute can worsen the representation of already underrepresented groups and decrease overall fairness of selection. We use several data sets from a broad range of domains from recidivism risk assessments to scientific citations to assess this effect in real-world settings. Our results demonstrate the importance of including all relevant attributes in debiasing procedures and that more efforts need to be put into eliminating the root causes of inequalities as purely numerical solutions such as quota-based debiasing might lead to unintended consequences.

cs.CY↗

arXiv2020

High academic performance is associated with shorter sleep and later bedtimes for young adults

Shorter sleep is known to be negatively associated with academic performance. However, this result has mostly been found in homogenous samples (e.g., students from one university) or when using relative measures of academic performance, such as grade point average. Consequently, the relationship between academic performance and sleep patterns at the population level is not well understood. In this paper, we examined the relationship between academic performance as measured by a standardized test and sleep patterns using data from a Russian panel study (N = 4,400) that was nationally representative for one age cohort (20-21 years old). In addition to self-reported sleep patterns, the data set contained information about participants' online activities over a period of up to 10 years, which allowed us to track the evolution of this relationship over time. We found that high academic performance was associated with shorter sleep, later bedtime, and increased online activity at night. The relationship between high academic performance and online activity at night was stable over a period of 5 years. Our findings suggest that the relationship between academic performance and sleep patterns can be more complex than previously believed and that high performance may be achieved at the expense of individual well-being.

cs.SI↗

arXiv2019

Generalizable prediction of academic performance from short texts on social media

It has already been established that digital traces can be used to predict various human attributes. In most cases, however, predictive models rely on features that are specific to a particular source of digital trace data. In contrast, short texts written by users $-$ tweets, posts, or comments $-$ are ubiquitous across multiple platforms. In this paper, we explore the predictive power of short texts with respect to the academic performance of their authors. We use data from a representative panel of Russian students that includes information about their educational outcomes and activity on a popular networking site, VK. We build a model to predict academic performance from users' posts on VK and then apply it to a different context. In particular, we show that the model could reproduce rankings of schools and universities from the posts of their students on social media. We also find that the same model could predict academic performance from tweets as well as from VK posts. The generalizability of a model trained on a relatively small data set could be explained by the use of continuous word representations trained on a much larger corpus of social media posts. This also allows for greater interpretability of model predictions.

cs.SI↗

arXiv2018

Schools are segregated by educational outcomes in the digital space

The Internet provides students with a unique opportunity to connect and maintain social ties with peers from other schools, irrespective of how far they are from each other. However, little is known about the real structure of such online relationships. In this paper, we investigate the structure of interschool friendship on a popular social networking site. We use data from 36,951 students from 590 schools of a large European city. We find that the probability of a friendship tie between students from neighboring schools is high and that it decreases with the distance between schools following the power law. We also find that students are more likely to be connected if the educational outcomes of their schools are similar. We show that this fact is not a consequence of residential segregation. While high- and low-performing schools are evenly distributed across the city, this is not the case for the digital space, where schools turn out to be segregated by educational outcomes. There is no significant correlation between the educational outcomes of a school and its geographical neighbors; however, there is a strong correlation between the educational outcomes of a school and its digital neighbors. These results challenge the common assumption that the Internet is a borderless space, and may have important implications for the understanding of educational inequality in the digital age.

cs.SI↗

arXiv2018

Gender Bias in Sharenting: Both Men and Women Mention Sons More Often Than Daughters on Social Media

Gender inequality starts before birth. Parents tend to prefer boys over girls, which is manifested in reproductive behavior, marital life, and parents' pastimes and investments in their children. While social media and sharing information about children (so-called "sharenting") have become an integral part of parenthood, it is not well-known if and how gender preference shapes online behavior of users. In this paper, we investigate public mentions of daughters and sons on social media. We use data from a popular social networking site on public posts from 635,665 users. We find that both men and women mention sons more often than daughters in their posts. We also find that posts featuring sons get more "likes" on average. Our results indicate that girls are underrepresented in parents' digital narratives about their children. This gender imbalance may send a message that girls are less important than boys, or that they deserve less attention, thus reinforcing gender inequality.

cs.SI↗

arXiv2017

The Digital Flynn Effect: Complexity of Posts on Social Media Increases over Time

Parents and teachers often express concern about the extensive use of social media by youngsters. Some of them see emoticons, undecipherable initialisms and loose grammar typical for social media as evidence of language degradation. In this paper, we use a simple measure of text complexity to investigate how the complexity of public posts on a popular social networking site changes over time. We analyze a unique dataset that contains texts posted by 942, 336 users from a large European city across nine years. We show that the chosen complexity measure is correlated with the academic performance of users: users from high-performing schools produce more complex texts than users from low-performing schools. We also find that complexity of posts increases with age. Finally, we demonstrate that overall language complexity of posts on the social networking site is constantly increasing. We call this phenomenon the digital Flynn effect. Our results may suggest that the worries about language degradation are not warranted.

physics.soc-ph↗

arXiv2016

Formation of homophily in academic performance: students prefer to change their friends rather than performance

Homophily, the tendency of individuals to associate with others who share similar traits, has been identified as a major driving force in the formation and evolution of social ties. In many cases, it is not clear if homophily is the result of a socialization process, where individuals change their traits according to the dominance of that trait in their local social networks, or if it results from a selection process, in which individuals reshape their social networks so that their traits match those in the new environment. Here we demonstrate the detailed temporal formation of strong homophily in academic achievements of high school and university students. We analyze a unique dataset that contains information about the detailed time evolution of a friendship network of 6,000 students across 42 months. Combining the evolving social network data with the time series of the academic performance (GPA) of individual students, we show that academic homophily is a result of selection: students prefer to gradually reorganize their social networks according to their performance levels, rather than adapting their performance to the level of their local group. We find no signs for a pull effect, where a social environment of good performers motivates bad students to improve their performance. We are able to understand the underlying dynamics of grades and networks with a simple model. The lack of a social pull effect in classical educational settings could have important implications for the understanding of the observed persistence of segregation, inequality and social immobility in societies.

physics.soc-ph↗

arXiv2014

Robust Source Localization in a Random Shallow Water Channel

This paper addresses source localization problem in a random shallow water channel. We present an extension of the generalized MUSIC method to the case, %in which when the signal correlation matrix is imprecisely known. The algorithm is validated by %simulations and its application to the experimental data observed in the Barents Sea. It has been found that the approach proposed demonstrates its excellent performance.

physics.ao-ph↗