Discussion Question 2

POLI 100F Lecture 6: Social Networks and Politics

Gregoire Phillips

University of California, San Diego

g1philli@ucsd.edu

November 9, 2020

Gregoire Phillips (UCSD) Social Data Analysis November 9, 2020 1 / 38

Overview

1 Announcements

2 Social Networks and Politics

3 Thinking About Final Projects

Gregoire Phillips (UCSD) Social Data Analysis November 9, 2020 2 / 38

A few brief reminders

Your annotated bibliography is due Friday!

The instructions can be found under Module 3 on the Home page

Need to be submitted by this Friday at 11:59 PM PST to be considered for full credit

Gregoire Phillips (UCSD) Social Data Analysis November 9, 2020 3 / 38

A few brief reminders

Your annotated bibliography is due Friday!

The instructions can be found under Module 3 on the Home page

Need to be submitted by this Friday at 11:59 PM PST to be considered for full credit

Gregoire Phillips (UCSD) Social Data Analysis November 9, 2020 3 / 38

A few brief reminders

Your annotated bibliography is due Friday!

The instructions can be found under Module 3 on the Home page

Need to be submitted by this Friday at 11:59 PM PST to be considered for full credit

Gregoire Phillips (UCSD) Social Data Analysis November 9, 2020 3 / 38

A few brief reminders

Your annotated bibliography is due Friday!

The instructions can be found under Module 3 on the Home page

Need to be submitted by this Friday at 11:59 PM PST to be considered for full credit

Gregoire Phillips (UCSD) Social Data Analysis November 9, 2020 3 / 38

Today’s Framing Questions

Our questions:

How do social networks affect political systems and behavior?

Our goals today:

Identify how social networks affect political behaviors like voting and sponsorship of legislation in Congress

Gregoire Phillips (UCSD) Social Data Analysis November 9, 2020 4 / 38

Today’s Framing Questions

Our questions:

How do social networks affect political systems and behavior?

Our goals today:

Identify how social networks affect political behaviors like voting and sponsorship of legislation in Congress

Gregoire Phillips (UCSD) Social Data Analysis November 9, 2020 4 / 38

Voting

Very few things more relevant this year in the United States than voting behavior

Gregoire Phillips (UCSD) Social Data Analysis November 9, 2020 5 / 38

Voting

Very few things more relevant this year in the United States than voting behavior

Voting is the most important participatory feature of democratic forms of government

Voting behavior ranges drastically across and within countries

Gregoire Phillips (UCSD) Social Data Analysis November 9, 2020 6 / 38

Voting

Very few things more relevant this year in the United States than voting behavior

Voting is the most important participatory feature of democratic forms of government

Voting behavior ranges drastically across and within countries

Gregoire Phillips (UCSD) Social Data Analysis November 9, 2020 6 / 38

Voting

Differences in voting turnout internationally

Gregoire Phillips (UCSD) Social Data Analysis November 9, 2020 7 / 38

Voting

Differences in voting turnout by age group internationally

Gregoire Phillips (UCSD) Social Data Analysis November 9, 2020 8 / 38

Voting

Differences in voting turnout iby age group in US

Gregoire Phillips (UCSD) Social Data Analysis November 9, 2020 9 / 38

Voting

Differences in voting turnout by racial identity in US

Gregoire Phillips (UCSD) Social Data Analysis November 9, 2020 10 / 38

Does your vote matter?

Most relevant question: does your vote matter?

Individually, your vote is very, very unlikely to be decisive

Does this mean you shouldn’t vote?

NO, because voting isn’t an individual phenomenon

Your decision to vote affects others’ decisions to vote

Voting is a socially contagious behavior

Your decision to vote can cascade through your social network

Gregoire Phillips (UCSD) Social Data Analysis November 9, 2020 11 / 38

Does your vote matter?

Most relevant question: does your vote matter?

Individually, your vote is very, very unlikely to be decisive

Does this mean you shouldn’t vote?

NO, because voting isn’t an individual phenomenon

Your decision to vote affects others’ decisions to vote

Voting is a socially contagious behavior

Your decision to vote can cascade through your social network

Gregoire Phillips (UCSD) Social Data Analysis November 9, 2020 11 / 38

Does your vote matter?

Most relevant question: does your vote matter?

Individually, your vote is very, very unlikely to be decisive

Does this mean you shouldn’t vote?

NO, because voting isn’t an individual phenomenon

Your decision to vote affects others’ decisions to vote

Voting is a socially contagious behavior

Your decision to vote can cascade through your social network

Gregoire Phillips (UCSD) Social Data Analysis November 9, 2020 11 / 38

Does your vote matter?

Most relevant question: does your vote matter?

Individually, your vote is very, very unlikely to be decisive

Does this mean you shouldn’t vote?

NO, because voting isn’t an individual phenomenon

Your decision to vote affects others’ decisions to vote

Voting is a socially contagious behavior

Your decision to vote can cascade through your social network

Gregoire Phillips (UCSD) Social Data Analysis November 9, 2020 11 / 38

Does your vote matter?

Most relevant question: does your vote matter?

Individually, your vote is very, very unlikely to be decisive

Does this mean you shouldn’t vote?

NO, because voting isn’t an individual phenomenon

Your decision to vote affects others’ decisions to vote

Voting is a socially contagious behavior

Your decision to vote can cascade through your social network

Gregoire Phillips (UCSD) Social Data Analysis November 9, 2020 11 / 38

Does your vote matter?

Most relevant question: does your vote matter?

Individually, your vote is very, very unlikely to be decisive

Does this mean you shouldn’t vote?

NO, because voting isn’t an individual phenomenon

Your decision to vote affects others’ decisions to vote

Voting is a socially contagious behavior

Your decision to vote can cascade through your social network

Gregoire Phillips (UCSD) Social Data Analysis November 9, 2020 11 / 38

Your vote matters

Voting as a socially contagious behavior

Gregoire Phillips (UCSD) Social Data Analysis November 9, 2020 12 / 38

Your vote matters

Voting as a socially contagious behavior

Gregoire Phillips (UCSD) Social Data Analysis November 9, 2020 13 / 38

Your vote matters

Voting as a socially contagious behavior

Gregoire Phillips (UCSD) Social Data Analysis November 9, 2020 14 / 38

Your vote matters

Voting as a socially contagious behavior

Gregoire Phillips (UCSD) Social Data Analysis November 9, 2020 15 / 38

Social Voting

Turnout in elections is correlated

between spouses

between friends, family and coworkers

Influence matters

people very likely to say they vote because their friends and relatives vote (Knack 1992)

Gregoire Phillips (UCSD) Social Data Analysis November 9, 2020 16 / 38

Social Voting

Mobilization increases turnout

Organizational

Individual – 34 percent try to influence peers (ISLES 1996)

This tells us that voting behavior is directly contagious – but does it cascade?

Gregoire Phillips (UCSD) Social Data Analysis November 9, 2020 17 / 38

Social Voting

Mobilization increases turnout

Organizational

Individual – 34 percent try to influence peers (ISLES 1996)

This tells us that voting behavior is directly contagious – but does it cascade?

Gregoire Phillips (UCSD) Social Data Analysis November 9, 2020 17 / 38

Social Voting

Mobilization increases turnout

Organizational

Individual – 34 percent try to influence peers (ISLES 1996)

This tells us that voting behavior is directly contagious – but does it cascade?

Gregoire Phillips (UCSD) Social Data Analysis November 9, 2020 17 / 38

Turnout Cascades

If turnout is contagious, then changing a single turnout decision may cascade to many voters’ decisions

Your decision to vote may affect aggregate turnout

AND if political preferences are hgihly correlated between you and the people in your network, this can affect electoral outcomes

E.g., if your friends vote for the same people, you boost the probability of an outcome in their favor

Gregoire Phillips (UCSD) Social Data Analysis November 9, 2020 18 / 38

Turnout Cascades

If turnout is contagious, then changing a single turnout decision may cascade to many voters’ decisions

Your decision to vote may affect aggregate turnout

AND if political preferences are hgihly correlated between you and the people in your network, this can affect electoral outcomes

E.g., if your friends vote for the same people, you boost the probability of an outcome in their favor

Gregoire Phillips (UCSD) Social Data Analysis November 9, 2020 18 / 38

Turnout Cascades

If turnout is contagious, then changing a single turnout decision may cascade to many voters’ decisions

Your decision to vote may affect aggregate turnout

AND if political preferences are hgihly correlated between you and the people in your network, this can affect electoral outcomes

E.g., if your friends vote for the same people, you boost the probability of an outcome in their favor

Gregoire Phillips (UCSD) Social Data Analysis November 9, 2020 18 / 38

Turnout Cascades

If turnout is contagious, then changing a single turnout decision may cascade to many voters’ decisions

Your decision to vote may affect aggregate turnout

AND if political preferences are hgihly correlated between you and the people in your network, this can affect electoral outcomes

E.g., if your friends vote for the same people, you boost the probability of an outcome in their favor

Gregoire Phillips (UCSD) Social Data Analysis November 9, 2020 18 / 38

Turnout Cascades

As social scientists, how might we measure something like this?

Option 1: Lab Experiment

Experimental design that creates a network and manipulates discussion of some turnout behavior

Hold less costly or inconsequential “election” with incentives for voters

Measure the actual transmission of behavior

Option 2: Field Experiment

Experimental design that leverages existing social network and manipulates discussion of some turnout behavior

Randomly assign treatment to a group of individuals, then measure “cascade” of treatment outcomes through network

Analyze results against a control group that didn’t receive seeded treatment

Gregoire Phillips (UCSD) Social Data Analysis November 9, 2020 19 / 38

Turnout Cascades

As social scientists, how might we measure something like this? Option 1: Lab Experiment

Experimental design that creates a network and manipulates discussion of some turnout behavior

Hold less costly or inconsequential “election” with incentives for voters

Measure the actual transmission of behavior

Option 2: Field Experiment

Experimental design that leverages existing social network and manipulates discussion of some turnout behavior

Randomly assign treatment to a group of individuals, then measure “cascade” of treatment outcomes through network

Analyze results against a control group that didn’t receive seeded treatment

Gregoire Phillips (UCSD) Social Data Analysis November 9, 2020 19 / 38

Turnout Cascades

As social scientists, how might we measure something like this? Option 1: Lab Experiment

Experimental design that creates a network and manipulates discussion of some turnout behavior

Hold less costly or inconsequential “election” with incentives for voters

Measure the actual transmission of behavior

Option 2: Field Experiment

Experimental design that leverages existing social network and manipulates discussion of some turnout behavior

Randomly assign treatment to a group of individuals, then measure “cascade” of treatment outcomes through network

Analyze results against a control group that didn’t receive seeded treatment

Gregoire Phillips (UCSD) Social Data Analysis November 9, 2020 19 / 38

Turnout Cascades

As social scientists, how might we measure something like this? Option 1: Lab Experiment

Experimental design that creates a network and manipulates discussion of some turnout behavior

Hold less costly or inconsequential “election” with incentives for voters

Measure the actual transmission of behavior

Option 2: Field Experiment

Experimental design that leverages existing social network and manipulates discussion of some turnout behavior

Randomly assign treatment to a group of individuals, then measure “cascade” of treatment outcomes through network

Analyze results against a control group that didn’t receive seeded treatment

Gregoire Phillips (UCSD) Social Data Analysis November 9, 2020 19 / 38

Turnout Cascades

As social scientists, how might we measure something like this? Option 1: Lab Experiment

Experimental design that creates a network and manipulates discussion of some turnout behavior

Hold less costly or inconsequential “election” with incentives for voters

Measure the actual transmission of behavior

Option 2: Field Experiment

Experimental design that leverages existing social network and manipulates discussion of some turnout behavior

Randomly assign treatment to a group of individuals, then measure “cascade” of treatment outcomes through network

Analyze results against a control group that didn’t receive seeded treatment

Gregoire Phillips (UCSD) Social Data Analysis November 9, 2020 19 / 38

Turnout Cascades

Survey Experiments

1986 South Bend Election Study (SBES)

1996 St. Louis Election Study (ISLES)

Gregoire Phillips (UCSD) Social Data Analysis November 9, 2020 20 / 38

Turnout Cascades

Survey Experiments

1986 South Bend Election Study (SBES)

1996 St. Louis Election Study (ISLES)

Gregoire Phillips (UCSD) Social Data Analysis November 9, 2020 20 / 38

Turnout Cascades

These show turnout does cascade

Gregoire Phillips (UCSD) Social Data Analysis November 9, 2020 21 / 38

Turnout Cascades

These show turnout does cascade

On average, 1 decision to vote will motivate 3 others to also go to the polls, and 2 of those people are likely to vote the same way as you do

Clustering of individuals with similar political orientation creates an incentive for you to generate additional turnout if you want your favored candidate to win

Your decision to vote is even more powerful than your vote itself – particularly if other people know about it!

Gregoire Phillips (UCSD) Social Data Analysis November 9, 2020 22 / 38

Turnout Cascades

Does this work on the internet? Yes!

Bond et al. (2012): 61 million person Facebook experiment on the influence of posting your voting behavior on validated voting outcomes

Authors randomized whether or not messages alerting people of election day included information “social message” of friends who voted

They find it has a significant effect!

Gregoire Phillips (UCSD) Social Data Analysis November 9, 2020 23 / 38

Turnout Cascades

Does this work on the internet? Yes!

Bond et al. (2012): 61 million person Facebook experiment on the influence of posting your voting behavior on validated voting outcomes

Authors randomized whether or not messages alerting people of election day included information “social message” of friends who voted

They find it has a significant effect!

Gregoire Phillips (UCSD) Social Data Analysis November 9, 2020 23 / 38

Turnout Cascades

Does this work on the internet? Yes!

Bond et al. (2012): 61 million person Facebook experiment on the influence of posting your voting behavior on validated voting outcomes

Authors randomized whether or not messages alerting people of election day included information “social message” of friends who voted

They find it has a significant effect!

Gregoire Phillips (UCSD) Social Data Analysis November 9, 2020 23 / 38

Turnout Cascades

Gregoire Phillips (UCSD) Social Data Analysis November 9, 2020 24 / 38

Turnout Cascades

Gregoire Phillips (UCSD) Social Data Analysis November 9, 2020 25 / 38

Turnout Cascades

Gregoire Phillips (UCSD) Social Data Analysis November 9, 2020 26 / 38

Turnout Cascades

Gregoire Phillips (UCSD) Social Data Analysis November 9, 2020 27 / 38

Turnout Cascades

Gregoire Phillips (UCSD) Social Data Analysis November 9, 2020 28 / 38

Politician Social Networks

How do we know who works with whom in US Congress? What can we learn about how Congress works?

Congress – both the House of Representatives and the Senate – are social networks

We can think of who works on what as being affected and facilitated by who knows and talks to who

It is common to use votes, but richer data comes from who cosponsors legislation

Fowler (2006) collected this data to explore cosponsorhip networks and learn about the structure of lawmaking bodies in practice

Gregoire Phillips (UCSD) Social Data Analysis November 9, 2020 29 / 38

Politician Social Networks

How do we know who works with whom in US Congress? What can we learn about how Congress works?

Congress – both the House of Representatives and the Senate – are social networks

We can think of who works on what as being affected and facilitated by who knows and talks to who

It is common to use votes, but richer data comes from who cosponsors legislation

Fowler (2006) collected this data to explore cosponsorhip networks and learn about the structure of lawmaking bodies in practice

Gregoire Phillips (UCSD) Social Data Analysis November 9, 2020 29 / 38

Politician Social Networks

How do we know who works with whom in US Congress? What can we learn about how Congress works?

Congress – both the House of Representatives and the Senate – are social networks

We can think of who works on what as being affected and facilitated by who knows and talks to who

It is common to use votes, but richer data comes from who cosponsors legislation

Fowler (2006) collected this data to explore cosponsorhip networks and learn about the structure of lawmaking bodies in practice

Gregoire Phillips (UCSD) Social Data Analysis November 9, 2020 29 / 38

Politician Social Networks

How do we know who works with whom in US Congress? What can we learn about how Congress works?

Congress – both the House of Representatives and the Senate – are social networks

We can think of who works on what as being affected and facilitated by who knows and talks to who

It is common to use votes, but richer data comes from who cosponsors legislation

Fowler (2006) collected this data to explore cosponsorhip networks and learn about the structure of lawmaking bodies in practice

Gregoire Phillips (UCSD) Social Data Analysis November 9, 2020 29 / 38

Politician Social Networks

How do we know who works with whom in US Congress? What can we learn about how Congress works?

Congress – both the House of Representatives and the Senate – are social networks

We can think of who works on what as being affected and facilitated by who knows and talks to who

It is common to use votes, but richer data comes from who cosponsors legislation

Fowler (2006) collected this data to explore cosponsorhip networks and learn about the structure of lawmaking bodies in practice

Gregoire Phillips (UCSD) Social Data Analysis November 9, 2020 29 / 38

Politician Social Networks

Gregoire Phillips (UCSD) Social Data Analysis November 9, 2020 30 / 38

Politician Social Networks

Gregoire Phillips (UCSD) Social Data Analysis November 9, 2020 31 / 38

Politician Social Networks

Gregoire Phillips (UCSD) Social Data Analysis November 9, 2020 32 / 38

Politician Social Networks

What do we learn from this kind of network? We learn:

Institutional ties matter: committee chairs, majority and minority leaders all attract in-committee and in-party sponsorship

Regional ties play into who co-sponsors legislation: being from the same state or contiguous district

Issue ties are sticky: if you cosponsor on a particular issue for one bill, you are very likely to do so again

Personal ties matter: personal friendships can transcend party lines and lead to higher co-sponsorship patterns

Gregoire Phillips (UCSD) Social Data Analysis November 9, 2020 33 / 38

Politician Social Networks

What do we learn from this kind of network? We learn:

Institutional ties matter: committee chairs, majority and minority leaders all attract in-committee and in-party sponsorship

Regional ties play into who co-sponsors legislation: being from the same state or contiguous district

Issue ties are sticky: if you cosponsor on a particular issue for one bill, you are very likely to do so again

Personal ties matter: personal friendships can transcend party lines and lead to higher co-sponsorship patterns

Gregoire Phillips (UCSD) Social Data Analysis November 9, 2020 33 / 38

Politician Social Networks

What do we learn from this kind of network? We learn:

Institutional ties matter: committee chairs, majority and minority leaders all attract in-committee and in-party sponsorship

Regional ties play into who co-sponsors legislation: being from the same state or contiguous district

Issue ties are sticky: if you cosponsor on a particular issue for one bill, you are very likely to do so again

Personal ties matter: personal friendships can transcend party lines and lead to higher co-sponsorship patterns

Gregoire Phillips (UCSD) Social Data Analysis November 9, 2020 33 / 38

Politician Social Networks

What do we learn from this kind of network? We learn:

Institutional ties matter: committee chairs, majority and minority leaders all attract in-committee and in-party sponsorship

Regional ties play into who co-sponsors legislation: being from the same state or contiguous district

Issue ties are sticky: if you cosponsor on a particular issue for one bill, you are very likely to do so again

Personal ties matter: personal friendships can transcend party lines and lead to higher co-sponsorship patterns

Gregoire Phillips (UCSD) Social Data Analysis November 9, 2020 33 / 38

Politician Social Networks

What do we learn from this kind of network? We learn:

Institutional ties matter: committee chairs, majority and minority leaders all attract in-committee and in-party sponsorship

Regional ties play into who co-sponsors legislation: being from the same state or contiguous district

Issue ties are sticky: if you cosponsor on a particular issue for one bill, you are very likely to do so again

Personal ties matter: personal friendships can transcend party lines and lead to higher co-sponsorship patterns

Gregoire Phillips (UCSD) Social Data Analysis November 9, 2020 33 / 38

Politician Social Networks

What do we learn from this kind of network? We learn:

Institutional ties matter: committee chairs, majority and minority leaders all attract in-committee and in-party sponsorship

Regional ties play into who co-sponsors legislation: being from the same state or contiguous district

Issue ties are sticky: if you cosponsor on a particular issue for one bill, you are very likely to do so again

Personal ties matter: personal friendships can transcend party lines and lead to higher co-sponsorship patterns

Gregoire Phillips (UCSD) Social Data Analysis November 9, 2020 33 / 38

Approaching the Course’s Final Descent

For your final project, we are putting together a proposal that includes

A research question

A literature review

A theory that helps answer your question

Some hypotheses on how we would prove that your theory does answer the question

A proposed research design on how you would test those hypotheses

At this stage, you should have a rough idea of the topic you want to write on, and perhaps the research question you would like to pose in your proposal

Gregoire Phillips (UCSD) Social Data Analysis November 9, 2020 34 / 38

Approaching the Course’s Final Descent

For your final project, we are putting together a proposal that includes

A research question

A literature review

A theory that helps answer your question

Some hypotheses on how we would prove that your theory does answer the question

A proposed research design on how you would test those hypotheses

At this stage, you should have a rough idea of the topic you want to write on, and perhaps the research question you would like to pose in your proposal

Gregoire Phillips (UCSD) Social Data Analysis November 9, 2020 34 / 38

Research Question

A research question should address some unexplained variation in the world that you are going to try to explain to your reader. Some examples of variation in the world:

Mask wearing compliance in San Diego

Support for the Black Lives Matter movement in the United States

Hong Kong citizen support for the new security law in Hong Kong

Turnout among young people in US elections

Gregoire Phillips (UCSD) Social Data Analysis November 9, 2020 35 / 38

Research Question

A research question should address some unexplained variation in the world that you are going to try to explain to your reader. Some examples of variation in the world:

Mask wearing compliance in San Diego

Support for the Black Lives Matter movement in the United States

Hong Kong citizen support for the new security law in Hong Kong

Turnout among young people in US elections

Gregoire Phillips (UCSD) Social Data Analysis November 9, 2020 35 / 38

Research Question

We transform these sources of variation into research questions by asking why there is variation in these things, or posing a more direct question about whether or not something explains this variation:

What explains variation in mask-wearing compliance among adults in San Diego?

Does viewership of CNN influence support for the BLM movement in the US?

Are Hong Kong citizens that have family overseas more likely to oppose new security laws in Hong Kong?

What explains the substantial variation in turnout among young people of different races in US elections?

Gregoire Phillips (UCSD) Social Data Analysis November 9, 2020 36 / 38

Research Question

We transform these sources of variation into research questions by asking why there is variation in these things, or posing a more direct question about whether or not something explains this variation:

What explains variation in mask-wearing compliance among adults in San Diego?

Does viewership of CNN influence support for the BLM movement in the US?

Are Hong Kong citizens that have family overseas more likely to oppose new security laws in Hong Kong?

What explains the substantial variation in turnout among young people of different races in US elections?

Gregoire Phillips (UCSD) Social Data Analysis November 9, 2020 36 / 38

Literature Review

Once you have a research question, you can dive into the academic and policy literature on topics related to your research question:

Academic articles or policy studies on the determinants of mask-wearing compliance

Academic articles, policy briefs, or expert analysis on how media consumption affects support for social movements

I had you do the literature review before some of you nailed down your research question so that you could dive into some literature to make sure your topics were feasible, so you probably already have a start to your literature review!

Gregoire Phillips (UCSD) Social Data Analysis November 9, 2020 37 / 38

Literature Review

Once you have a research question, you can dive into the academic and policy literature on topics related to your research question:

Academic articles or policy studies on the determinants of mask-wearing compliance

Academic articles, policy briefs, or expert analysis on how media consumption affects support for social movements

I had you do the literature review before some of you nailed down your research question so that you could dive into some literature to make sure your topics were feasible, so you probably already have a start to your literature review!

Gregoire Phillips (UCSD) Social Data Analysis November 9, 2020 37 / 38

Literature Review

Once you have a research question, you can dive into the academic and policy literature on topics related to your research question:

Academic articles or policy studies on the determinants of mask-wearing compliance

Academic articles, policy briefs, or expert analysis on how media consumption affects support for social movements

I had you do the literature review before some of you nailed down your research question so that you could dive into some literature to make sure your topics were feasible, so you probably already have a start to your literature review!

Gregoire Phillips (UCSD) Social Data Analysis November 9, 2020 37 / 38

Theory

What do you think explains the variation, and how? The theory section of this proposal is your opportunity to tell us. In this course, we ask that you focus on explanations that relate to social networks:

Individuals posting pictures wearing masks are likely to increase mask-wearing compliance within their social networks

Individuals sharing support for the BLM movement on social media are likely to influence others to form opinions on the movement, but only those who consume media from similar sources are likely to adopt their position.

This is where we want you to be at in the next couple of weeks. We will move to longer discussions of research design toward the end of the course.

Gregoire Phillips (UCSD) Social Data Analysis November 9, 2020 38 / 38

Theory

What do you think explains the variation, and how? The theory section of this proposal is your opportunity to tell us. In this course, we ask that you focus on explanations that relate to social networks:

Individuals posting pictures wearing masks are likely to increase mask-wearing compliance within their social networks

Individuals sharing support for the BLM movement on social media are likely to influence others to form opinions on the movement, but only those who consume media from similar sources are likely to adopt their position.

This is where we want you to be at in the next couple of weeks. We will move to longer discussions of research design toward the end of the course.

Gregoire Phillips (UCSD) Social Data Analysis November 9, 2020 38 / 38

Theory

What do you think explains the variation, and how? The theory section of this proposal is your opportunity to tell us. In this course, we ask that you focus on explanations that relate to social networks:

Individuals posting pictures wearing masks are likely to increase mask-wearing compliance within their social networks

Individuals sharing support for the BLM movement on social media are likely to influence others to form opinions on the movement, but only those who consume media from similar sources are likely to adopt their position.

This is where we want you to be at in the next couple of weeks. We will move to longer discussions of research design toward the end of the course.

Gregoire Phillips (UCSD) Social Data Analysis November 9, 2020 38 / 38