Measuring Social Media Inequality

 

What do millions of social media images shared in New York, Bangkok, Sao Paolo, or London tell us about each city? Which parts of a city receive most attention and which remain invisible? How can we quantify and measure these patterns?  

Inequaligram project analyzes 7,442,454 public Instagram images shared in Manhattan over five months. We use measures of inequality from economics to analyze differences in sharing between parts of a city.

The ratio between a Census tract with most images and the tract with least images is staggering: 250,000 : 1. For locals, 50% of their images are shared only in 21% of Manhattan area. For tourists, this proportion is 12%. The significant parts of the city are thus largely invisible on Instagram.

The inequality of locals' Instagram sharing turns out to be bigger than inequalities in levels of income, rent, and unemployment.  The inequality of visitors' sharing is larger than income inequality in the most unequal countries. (See Analysis).  

Do locals and visitors use and represent the city in similar ways? We compare 10 most popular places for each group discuss results in Rankings section.

For additional analysis, details of our methods and discussion of the results, see  Publications

 

INTRODUCTION

Every world city has large inequalities

in income, wealth, education, social well-being, and access to services.

Social media sharing adds new inequalities. In some parts of the city people share many images that show their experiences and places they visit. In other areas, they share much less.

In this way, some parts of a city become “social media rich” while others remain “social media poor.”

Why should we care about this?

Social media allows us to share with others our urban experiences and self-representations. These collective posts create an “image of a city” for its residents and the outside world.

In this image, some areas are represented well, while others are invisible. This directly affects city economy and social life. The areas well represented in social media attract more people who spend time and money there. The invisible areas are less likely to be visited.

How can we quantify and measure these differences ?

We introduce a new concept of “social media inequality."

This concept allows us to quantitatively compare social media activities between parts of a city, a number of cities, or any other spatial areas. 

To test our ideas, we use a dataset of 7,442,454 public geo-coded Instagram images shared in Manhattan during five months (March-July) in 2014. (The next edition of our project will add analysis other cities.)

This website presents only our core findings. For details about our methods, additional findings, and discussion see our Publications.  

DATASET

To create our dataset, we collected all publicly shared Instagram images with location information inside Manhattan during five months in 2014. We then divided collected data into images that were likely shared by people living in Manhattan and images that were likely shared by visitors.

Locals

366,539

Number of accounts

5,918,408

Shared geo-tagged images

14,119,037

All assigned tags

0.537

% of images have hashtags

Visitors

505,345

Number of accounts

1,524,046

Shared geo-tagged images

2,767,822

All assigned tags

0.455

% of images have hashtags

According to U.S. Census Bureau (2015), there are 1,636,268 people residing in Manhattan. In the same year, 56.4 million people visited New York City (source: NYC statistics, 2015).

ANALYSIS

We define social media inequality using an analogy with the concept of economic inequality.

Economic inequality indicates how some economic characteristics or material resources, such as income, wealth or consumption are distributed in a city, country or between countries. Accordingly, we can define social media inequality as the measures of distribution of characteristics of social media content shared in a particular geographic area or between areas. The examples of such characteristics is the number of photos shared by all users of a social network such as Instagram in a given city area, the content of these photos, the numbers of tags, and content of these tags.

To measure social media inequality, we use the standard measure of inequality adopted in many fields: Gini index.

For example, it is used in economics to measure income or wealth inequalities. The index can vary between 0 (complete equality) and 1 (absolute inequality). What are Gini indexes for numbers of images shared on Instagram in Manhattan by locals and visitors? (This calculation numbers of images shared in each Census tract normalized by tract size.)

0.494

Locals

0.669

Visitors

To put these measures of social media inequality in perspective, we can compare them with measures for income inequality in a number of countries and Manhattan:

0.271

Finland (one of the lowest in income inequality)

0.358

Spain

0.414

USA

0.481

Mexico

0.594

Manhattan

0.658

Seychelles (one of the highest in income inequality)

To put these measures of social media inequality in perspective, we can compare them with measures for income inequality in a number of countries and Manhattan:

0.271

Finland (one of the lowest in income inequality)

0.358

Spain

0.414

USA

0.481

Mexico

0.594

Manhattan

0.658

Seychelles (one of the highest in income inequality)

 

To visually see the differences in sharing by locals and visitors, we plot locations of 200,000 images randomly selected from our dataset.

 

Ranking

Which areas in Manhattan are most popular on Instagram (i.e., have most images) for locals and for visitors? Looking at the lists of top 10 places, we see that many of them are the same and only their ranks differ. In other words, both locals and visitors prefer to represent the same places on Instagram, “filtering” the city in the same way.

However, we also see important differences. In the tourists list, 9 areas out of 10 are located in a single part of Manhattan:  Midtown. For locals, 6 areas are in Midtown and 4 are in Downtown. The part of the city above 59th Street does not appear in either visitors or local rankings.

Locals

Rank

Location

Tract

1.

Times Square

119

2.

Rockefeller Center

104

3.

Times Square

125

4.

Chelsea Market

83

5.

Central Park South and 5th Ave.

112.01

6.

Bleecker and Lafayette

55.02

7.

Times Square

113

8.

Union Square

52

9.

59th Street and Madison Ave.

112.02

10.

Cooper Square

4

Visitors

Rank

Location

Tract

1.

Times Square

119

2.

Times Square

125

3.

Rockefeller Center

104

4.

Empire State Building

76

5.

59th Street and Madison Ave.

112.02

6.

Times Square

131

7.

Central Park South and 5th Ave.

112.01

8.

Madison Square Garden

101

9.

Bryant Park

84

10.

Chelsea Market

83

Day

Rank

Location

Tract

1.

Times Square

113

2.

Central Park South and 5th Ave.

112.01

3.

Rockefeller Center

104

4.

Times Square

119

5.

59th Street snd Madison Ave.

112.02

6.

Union Square

52

7.

Bleecker and Lafayette

55.02

8.

Times Square

125

9.

Chelsea Market

83

10.

Bryant Park

84

Night

Rank

Location

Tract

1.

Times Square

119

2.

Times Square

125

3.

Chelsea Market

83

4.

Rockefeller Center

104

5.

Cooper Square

42

6.

Bleecker and Lafayette

55.02

7.

Ludlow and Stanton

30.01

8.

Madison Square Garden

101

9.

Central Park South and 5th Ave.

112.01

10.

Union Square

52

Photos by locals in selected NYC areas below 59th Street

For locals, we see a clear separation between work areas (Financial District), shopping areas (SoHo) and nightlife areas: East Village, Lower East Side, Meatpacking District. 

Photos by locals in selected NYC areas above 59th Street

The patterns in the neighborhoods located above 59th Street are different. Overall, the number of shared photos keep increasing and peaks latter in the evening. And in the areas with lower average income (East Harlem, Morningside Heights, and Washington Heights), this number keeps increasing until midnight. (Read our article linked in Publications for an explanation we offer for this pattern.)

Photos by visitors in selected NYC areas below 59th Street

In late evening, visitors continue to be active in three downtown areas where locals' activity already decreased: East Village Lower East Side, Meatpacking District. But they are also active in another area - Midtown. This area has the biggest concentration of hotels in the city and also key attractions such as Times Square. (Note: Because the number of images shared by visitors in the neighborhoods above 110th Street is relatively low, these patterns are not as reliable, so we don't include this graph.) 

PUBLICATIONS

Agustin Indaco and Lev Manovich. "Urban Social Media Inequality: Definition, Measurements, and application." 2016.
Full paper (23,000 words): 

http://arxiv.org/abs/1607.01845

Agustin Indaco and Lev Manovich. "Urban Social Media Inequality: Definition, Measurements, and application." Urban Studies and Practices Journal (usp.hse.ru), forthcoming. Paper PDF (5000 words):

http://manovich.net/index.php/projects/social-media-inequality

CONTACT AND CREDITS

Lev Manovich

Computer Science, The Graduate Center, City University of New York (CUNY). 

Agustin Indaco

Economics, The Graduate Center, City University of New York (CUNY). 

Davis Muiznieks

Website design

Project history:

This study uses 7,442,454 public geo-coded Instagram images shared in Manhattan during five months (March-July) in 2014. This dataset was originally created for On Broadway (http://on-broadway.net) project commissioned by New York Public Library. This work took place in Software Studies Lab. Jay Chow downloaded all Instagram data and images. Mehrdad Yazdani prepared the data, separated locals and visitors, analyzed hashtags, and also aligned image locations with Census boundaries working together with Ran Goldblatt.

On Broadway allowed us to develop the conceptual framework for Inequaligram project: analysis of social media patterns across small city areas, and comparison with Census indicators for these areas. Manovich proposed the term "social media inequality" to refer to differences in social media between areas. Indaco proposed use of inequality measures from economics for this purpose. 

inequaligram is one of many projects created by Software Studies Lab. We analyze millions of images shared on Instagram and Twitter in many cities around the world. These projects include selfiecity.net, on-broadway.nyc and the-everyday.net.

 

On Broadway:
http://on-broadway.net

 

Support:

The Graduate Center, City University of New York

California Institute for Telecommunications and Information at University of California, San Diego

The Andrew W. Mellon Foundation