Skip to main content
SAGE
Search form
PDF
  • 00:00

    [MUSIC PLAYING][Measuring Inequality in Social Groups & Computational SocialScience]

  • 00:10

    MILENA TSVETKOVA: So my name is Milena Tsvetkova,and I'm an assistant professor at the LondonSchool of Economics and Political Scienceand the Department of Methodology there.[Milena Tsvetkova, Assistant Professor, London Schoolof Economics] And I'm broadly interestedin computational social science, and particularlyonline experiments, social network analysis,computational models such as agent-based models, and more

  • 00:32

    MILENA TSVETKOVA [continued]: substantively I'm more interested in cooperation,inequalities, segregation.Just basically large social problems at the group level.So in terms of computational social science,I think I was at the right place at the right time.So I did my PhD with Michael Macywho was a big figure in this field at Cornell University,

  • 00:53

    MILENA TSVETKOVA [continued]: and we collaborated a bit.I'm not sure I knew what computational social sciencewhen I entered the school, but I wasinterested in a lot of new methods,new technologies, new kind of data that was available,and just gradually fell into the field.And in particular, my interest--

  • 01:13

    MILENA TSVETKOVA [continued]: kind of substantive topic interest--I think I was always fascinated with these kindof big, large problems that society faces.And I always describe my researchas basic social research because Iam kind of studying human behavior in a very abstractedway with the idea that these kind of universal

  • 01:35

    MILENA TSVETKOVA [continued]: processes that happen regardless of human culture,or very in particular social context.So I am interested in these kind of processesthat happen once you bring large groups of people togetherand just inevitably things, certain patterns emerge.So the current project I'm working on

  • 01:57

    MILENA TSVETKOVA [continued]: is on the emergence of inequality in social groups.So this is a collaboration with the computer scientist ClaudiaWagner from Gesis in Germany and somebody from organizationsOana Vuculescu from Aarhus Universityand we are interested in how inequality emerges

  • 02:18

    MILENA TSVETKOVA [continued]: in groups of people who interact with each other.And so the idea here is that inequality of courseis a big topic.It's huge in social science.People have looked at it from many different angles,and of course recently it kind ofhas come up again and again in public consciousness.But we have like a lot of scandals

  • 02:39

    MILENA TSVETKOVA [continued]: for example, the Panama Papers if you remember.We also had social movements such as the Occupy Wall Streetor the We are the 99% movement.There was also kind of a public, well general audiencebook, that came out that was a big hit thatwas the [INAUDIBLE] book of the Captain of the 21st Century

  • 02:59

    MILENA TSVETKOVA [continued]: that all deal with inequality, right?So in the sense that inequality has been rising,particularly the United States.And we were interested in looking at a very new angle.So we people have looked at inequalityfrom kind of more macro perspective where they lookat kind of reproduction of inequalitydue to the institutions we have.

  • 03:20

    MILENA TSVETKOVA [continued]: Also we have more sociologists havelooked at the intergenerational transmission of inequalitywhere because of your parents' socioeconomic backgroundyou are disadvantaged, you have fewer opportunitiesto rise in the socioeconomic hierarchy for example.But we kind of looked at it a bit

  • 03:40

    MILENA TSVETKOVA [continued]: again in more kind of basic fundamental level, the factthat it could emerge anytime we bring large groups of peopletogether that we can see it in different realms becauseof this kind of self-reinforcing processesthat you get into a complex system.And that's what when have we havelarge groups of individuals, independent agents,

  • 04:02

    MILENA TSVETKOVA [continued]: interacting with each other.So we are-- the idea of the project wasto use online experiments, large-scale online experimentswhere we can have large groups of people comingtogether and directing a very abstract settingand they interacted with different conditions.And we want to see which conditions-- under which

  • 04:23

    MILENA TSVETKOVA [continued]: conditions there is more inequality.So we partnered with a group of game developers and a citizenscience project at Oracle's universitythat's called Science at Home.And we wanted to-- in order to kind of scale upto have these large groups, many large groups of peopleto study, we wanted to do a gamified game in a sense.

  • 04:48

    MILENA TSVETKOVA [continued]: A game that's interesting to play for people.And so we are currently actually working on this.While we were doing that, we realized thatin the last couple of decades, actually therehas been a lot of research using similar games wheregroups of individuals play a cooperation game you might have

  • 05:11

    MILENA TSVETKOVA [continued]: heard of the Prisoner's Dilemma gamewhere there is an incentive to cheat othersto your own personal benefit, although collaborating,cooperating with others is more beneficial to the group.So actually what we did was that while we are developinga game which takes a lot of time that we decided to collect data

  • 05:33

    MILENA TSVETKOVA [continued]: from, a lot of these studies have been donein the last couple of decades.And to reanalyze it because a lot of these studieshave focused on the emergence of cooperation.So when we have these kind of social dilemmatype of situations, under what conditionsdo we get people to cooperate?But nobody had looked at well, OK.

  • 05:55

    MILENA TSVETKOVA [continued]: When we have a lot of people to cooperate,that means that the average wealth, the collective wealthrises, but how is this wealth distributed?Right?Which is basically the problem of inequality.And so we re-analyzed the data.We got data from 18 different studies,and we saw different conditions and how theyaffect the level of inequality.

  • 06:17

    MILENA TSVETKOVA [continued]: So yes we collected data from 18 different experiments thatwere done in the last two decades,and they varied a lot in size.So to the largest experiment we have was 625 peopleinteracting simultaneously.And the smallest we limited to people of at least--

  • 06:37

    MILENA TSVETKOVA [continued]: to groups of at least 10 people since of courseyou want to study inequality.And if you need a certain large enough groupto talk about inequality.But they varied a lot in terms of just many different aspects.There is a different experiments.Everybody has a different setup.They use a different kind of game of a cooperation

  • 06:57

    MILENA TSVETKOVA [continued]: game for people to interact with different settings,different kind of times of periods in which they interact.And so many different aspects.So that was a big challenge, so first kindof bringing all the data together.Well, first it was the question of getting it, right?So that was kind of very excitingthat so many researchers actuallywere willing to share the data.It was actually I think very encouraging

  • 07:19

    MILENA TSVETKOVA [continued]: for kind of a new academic like me.I thought it really made me excited about the fieldof academia and I mean just sharing data, but also thenmaking it all in a comparable format, right?Kind of reconciling it all although it's so different.And then the other problem was how do we now compare?

  • 07:43

    MILENA TSVETKOVA [continued]: Because it's an experiment, the problem is is that it's a veryartificial setting, and you cannot talk about effect size.So you cannot talk about if you make a certain intervention,what does it mean in real life?Because it's just a very artificial, abstract setting.

  • 08:05

    MILENA TSVETKOVA [continued]: So what we may be interested in is is there an effect at all?Right?So if you make this intervention, what happens?Does anything happen?And so to do that, we actually hadto go over different meta-analytic techniques.So there is a lot of meta-analysishas been done in usually medical studies where actually there

  • 08:27

    MILENA TSVETKOVA [continued]: are people are interested in the effect size.So they, for example, would look at a defective medicineand they want to know, well does it cure disease?And to what extent?Are there any side effects?They're very interested in kind of exactly measuringwhat's the improvement in health outcomes.And in our case, we were just interested is there

  • 08:49

    MILENA TSVETKOVA [continued]: any effect at all?So we use the very kind of very simple methodwhere we basically--it's kind of like it's a very simple idea of flipping a coin.Right?That is that if we live in a world in which thereis no effect from what we looking at, right?

  • 09:10

    MILENA TSVETKOVA [continued]: So basically you have a cause and effect but then basicallythere's no connection between the two,then if you repeat an experiment again and again,it's equivalent to flipping a coin.Like one time there would be plus effects.Sometimes there would be negative.Sometimes it would be zero.Right?But if we repeat it again and again,and every time it's in one direction,we can actually very precisely measure this,

  • 09:31

    MILENA TSVETKOVA [continued]: and say that well if we really observe this, thenwe have kind of good evidence that this will not be possibleif the underlying truth is basically no effect whatsoever.It gives us a kind of a confidenceto say that even though these experienceswe're very different they measured the same thing

  • 09:54

    MILENA TSVETKOVA [continued]: but in very different settings.And sometimes they would find the big effects.Sometimes they wouldn't.But if we keep on seeing the same direction of the effect,then we definitely can say that there'ssomething going on here.So that's kind of the method we used.So this is very kind of exciting yet scary partbecause that's both exciting and scary about developing

  • 10:18

    MILENA TSVETKOVA [continued]: games or experiments is that you have to design everythingat the beginning.Usually people who deal with observed data,they always say, well it would have been nice to study this.But we don't have available data.So this is the best we can do.Unfortunately with experiments you cannot do that because youdesign the data collection process.

  • 10:38

    MILENA TSVETKOVA [continued]: Right?So there is no excuse.So that's kind of the scary parts whereyou have to foresee all kinds of things that may go wrong,all kinds of possibilities that maybe youcan of you know reconsider the research question.The other challenge we are facing nowis the gamification aspect.Right?So how do you--

  • 10:59

    MILENA TSVETKOVA [continued]: so you want to do an experiment that's very simple,controls a lot of things because we only interested in the oneparticular aspect that we want to study,to isolate this cause and effect,but how do we make this a very interesting game for peoplethat want to play?And that's a bit of a back and forth for game developersnow because of course they want to kind of add

  • 11:19

    MILENA TSVETKOVA [continued]: all these exciting little features thatmake people like come back in and play more and more,but that's the part that would make the experiment not valid.So we're working on this.Hopefully soon we'll have a pilot, maybe in a month or two.So we'll have to test this, and it'sgoing to be a iterative process where we will have to adjust,

  • 11:41

    MILENA TSVETKOVA [continued]: test again, and hopefully in a few monthswe'll be able to collect the data.I mean I think a lot of people are excited about collectinglots of data.Oh there is very new cool web siteand a new, cool online community.It would be cool to gather data, but oftenas a social scientist, I am much more interested in the researchquestion.And I think that often there is already existing data that

  • 12:04

    MILENA TSVETKOVA [continued]: could answer a question.Social scientists, or actually I mean scientists in general,cannot talk about industry, of course.But they're often willing to share the data.So I think that was very, very interesting for me that we--while we were thinking about designing our own experimentswhich of course are necessary, but we alsorealized that we can just obtain data from others

  • 12:28

    MILENA TSVETKOVA [continued]: and so we don't have to reinvent the wheel.We don't have to replicate these studies.And of course they are limits to this approachbecause these studies were not designed exactlywith our research question.So there is kind of this well our approach wasa bit more data-driven.So there are some limitations, but at the same time,we learned some new things that now when we actuallygo to the board and start designing all of our project,

  • 12:50

    MILENA TSVETKOVA [continued]: we can do it.So I think often with just access,so much easy access to lots of data online,people get a bit too excited about it,not realizing that the research question in a sense issometimes more important and it's--

  • 13:10

    MILENA TSVETKOVA [continued]: there is no-- there is little value of kindof reinventing the wheel.Right?Like rediscovering something thatis a well-known fact in a new setting,but sometimes it's more interesting to come upwith a new research question and maybe reusing data toolto answer it.[MUSIC PLAYING]

Video Info

Publisher: SAGE Publications Ltd

Publication Year: 2019

Video Type:Video Case

Methods: Agent-based simulation, Computational modeling

Keywords: agent-based modeling and simulation in the social sciences; cooperation/cooperative studies; data analysis; games; group behavior; inequality; intergenerational mobility; internet; internet data collection; Social inequality; Social science research; technology ... Show More

Segment Info

Segment Num.: 1

Persons Discussed:

Events Discussed:

Keywords:

Abstract

Associate Professor, Milena Tsvetkova, PhD, discusses her research on the emergence of inequality in online social groups, including the use of existing cooperation-study data, the challenge of analyzing differing data sets, asking relevant research questions, and working with online game designers.

Video Info

Publication Info

Publisher:
SAGE Publications Ltd
Publication Year:
2019
Product:
SAGE Research Methods Video: Data Science, Big Data Analytics, and Digital Methods
Publication Place:
London, United Kingdom
SAGE Original Production Type:
SAGE Experts
ISBN:
9781526491534
DOI
http://dx.doi.org/10.4135/9781526491534
Copyright Statement:
(c) SAGE Publications Ltd., 2019

People

Academic:
Milena Tsvetkova

Segment Info

Title:

Segment Num: 1

Keywords:

Segment Start Time:

Segment End Time:

People

Things Discussed

Organizations Discussed:

Events Discussed:

Places Discussed:

Persons Discussed:

Methods Map

Agent-based simulation

An application of computer simulation to the social sciences useful for studying the complexity of social systems. Agents are software objects programmed to have autonomy and goals, and to act and react towards their environment and other agents within it. Running the simulation allows emerging patterns of activity to be observed and compared against those found in the real-world.
Agent-based simulation
Measuring Inequality in Social Groups

Associate Professor, Milena Tsvetkova, PhD, discusses her research on the emergence of inequality in online social groups, including the use of existing cooperation-study data, the challenge of analyzing differing data sets, asking relevant research questions, and working with online game designers.