NARRATIVES and INFORMATION POLLUTION on VULNERABLE GROUPS DURING the COVID-19 PANDEMIC: Panama and Venezuela Cases

NARRATIVES AND INFORMATION POLLUTION ON VULNERABLE GROUPS DURING THE COVID-19 PANDEMIC: panama and venezuela cases

United Nations Development Programme (UNDP)

UNDP Oslo Governance Center Malin Herwig, Director

Country Office in Panama Aleida Ferreyra, Resident Representative a.i.

Country Office in Venezuela Jan Harfst, Resident Representative Rosicler Gómez, Deputy Resident Representative

Design and Layout Analiesse Ibarra Mendoza

4 | narratives and information pollution on vulnerable groups during the covid-19 pandemic FOREWORD

This report is the result of a joint effort by the UNDP offices in Panama and Venezuela, with the support of the UNDP Oslo Governance Centre and the UNDP Regional Centre for Latin America and the Caribbean. It is a contribution to address one of the greatest challenges of our time, the information pollution during the COVID-19 pandemic.

We would like to highlight three aspects of the report. First, the study focuses on the effects of the pandemic on the most vulnerable. UNDP has as one of its priorities the attention to this population, which has been clearly expressed in the Agenda 2030 call to “leave no one behind”. This work is part of an analysis of the impact of the pandemic on two specific cases: migrants returning to Venezuela and the approach to the issue of gender equality in Panama.

Secondly, the report addresses the issue from the perspective of information pollution in new and unavoidable scenarios, the social networks. In these spaces, public issues are discussed with the participation of diverse and significant actors: governments, the media, NGOs, influencers, scientists, as well as people in general. The debates in these spaces often precede and set the tone for traditional discussion spaces. They have played a crucial role in capturing the perceptions and concerns of the population during the pandemic. Intense debates on vulnerable populations take place in them, which are not exempt from information pollution.

The approach to these new scenarios of debate with innovative methodologies is an expression of UNDP’s permanent effort to update them and to incorporate new tools to approach its work. The monitoring of the networks analyzed in this work was carried out with an emphasis on quantitative methodologies with the clear purpose of supporting the databased decision-making process.

Third, the report goes beyond data analysis. Each case study culminates with a proposal for a communication and capacity building strategy to address the contamination of information on vulnerable groups. These proposals can be used as a reference to address this phenomenon in other populations of interest.

narratives and information pollution on vulnerable groups during the covid-19 pandemic | 5 6 | narratives and information pollution on vulnerable groups during the covid-19 pandemic CONTENT

SUMMARY 11 INTRODUCTION 13 METHODOLOGY 16

PART 1: RETURNED MIGRANTS TO VENEZUELA 18 Introduction 19 Sources of information 23 Twitter Google Trends Media Communications strategy 46 Conclusions 48

PART 2: WOMEN AND THE LGBTIQ+ COMMUNITY IN PANAMA 50 Introduction 51 Sources of information 53 Twitter Google Trends Media Communications strategy 68 Conclusions 71 REFERENCES 73

VENEZUELA

PANAMÁ

8 | narratives and information pollution on vulnerable groups during the covid-19 pandemic ABSTRACT

This report identifies narratives and infor- In the case of Panama, there was an increase mation pollution on two vulnerable popula- in the volume of discussion on gender issues tion groups during the COVID-19 pandemic: in a differentiated manner. “Feminism”, women and LGBTIQ+ population in Panama “Abuse”, “Gender-based violence”, “Domestic and returned migrants in Venezuela. The violence” and “Gender” had significant sources of information used are: Twitter, increases. However, the main narratives Google Trends and media during the period detected are not directly related to COVID-19 March-November 2020. and its effects. The narratives are framed in the confrontation of conservative and pro- In the case of Venezuela, the volume of dis- gressive values in the discussion of gender cussion about returned migrants increased equality, abortion rights and the recognition substantially during the pandemic, especially of gender diversity. The most frequent form the issue of those who return through non-of- of pollution is through unfounded accusa- ficial routes. The source analysed with the tions of wrongdoing against opponents. largest increase was Twitter. The discussion and stigmatization of the migrants, who Disinformation is the dominant type of return as the main vector of contagion of the information pollution in both countries. For virus , alongside with the conspiracy theories each country, a communication strategy is about its external origin, play an important proposed to mitigate the impact of pollution. role in the narratives and information pollution ; and these , also cause a high political polarization when the debate on migrants is concerned.

narratives and information pollution on vulnerable groups during the covid-19 pandemic | 9 10 | narratives and information pollution on vulnerable groups during the covid-19 pandemic INTRODUCTION

The relationship between crises and vulnerable March-November 2020. In the case of Venezuela, populations is a widely studied topic in economic meetings were held with the Venezuelan and social terms. However, little is known about the Border Observatory and representatives of crisis-vulnerable groups relationship in terms of the Caritas Venezuela in the towns of San Cristóbal, information flows discussed in the media and social Machiques and Guasdualito. networks. In this report, we identify the narratives and information pollution about women and the The source analysed in more detail is Twitter and LGBTIQ+ community in Panama and migrant is used as a comparative reference with the other returnees to Venezuela during the COVID-19 sources. Although Twitter is not a representative pandemic. The extraordinary importance of social sample in the statistical sense, it is a privileged networks and digital media in the world of medium for government and social leaders, information and the information explosion influencers and the general public to express around the pandemic, the so-called “infodemic” opinions on matters of public interest. Information (PAHO, 2020), (Lancet, 2020), (Arroyo-Sánchez AS, and opinion trends on Twitter usually precede 2020), justify this approach to the subject from the those in the media. Tweets are public and can be angle of information pollution. collected relatively easily. Recently developed techniques of text mining (Gupta, 2009) and Three sources of information were used in community detection (Blondel, 2008) allow for the the analysis: Twitter, Google Trends and media identification of narratives and information with a digital presence, during the period pollution, as well as their propagation patterns.

narratives and information pollution on vulnerable groups during the covid-19 pandemic | 11 Among the main findings of this study for the In the case of Panama, there was also an increase case of Venezuela, we found that the volume in the volume of discussion on gender issues in of discussion in the digital information sources a differentiated manner. “Feminism”, “Abuse”, mentioned about returned migrants increased “Gender-based violence”, “Domestic violence” and substantially during the pandemic, with the “Gender” had significant increases. However, the presence of information pollution . Twitter was main narratives detected are not directly related the source analysed where the increase was to COVID-19 and its effects. The narratives are greatest. The narratives and information pollution framed in the confrontation of conservative and are dominated by the discussion of returning progressive values in the discussion of gender migrants as the main vector of contagion of the equality, abortion rights and the recognition virus and conspiracy theories about the external of gender diversity. The most frequent form of origin of the virus. High political polarisation is pollution is through unfounded accusations of closely related to the debate on migrants and is wrongdoing against opponents one of the causes of information pollution.

12 | narratives and information pollution on vulnerable groups during the covid-19 pandemic In both countries, debate and information This work is part of the initiatives on information pollution is more aggressive on Twitter than pollution promoted and funded by the UNDP Oslo in the media with a digital presence, and Governance Centre. The project was designed disinformation is the dominant type of and implemented by the UNDP country offices in information pollution. In both countries, the Panama and Venezuela. population showed interest in searching the web for information about discussions with We thank the Recovery Solutions and Human information pollution, which shows that not all Mobility Team of the UNDP Crisis Bureau - Head- the population passively accepts it. In both cases, quarters for their financial support. The support communities of users who do not propagate of Governance Team of the UNDP Regional pollution were detected, showing a careful Center for Latin America and the Caribbean was attitude. For each country, a communication substantial. Niamh Hanafin’s comments during the strategy is proposed to mitigate the impact of development of the work were valuable. We also information pollution. thank Caritas de Venezuela and the Venezuelan Border Observatory for the exchange carried out.

narratives and information pollution on vulnerable groups during the covid-19 pandemic | 13 METHODOLOGY The methodology used to process the Twitter data is described in the image. The methodology used for the media is similar, except that no user communities are detected. In the case of Google Trends, only the most relevant search terms and their associated metrics were identified.

Database Narrative Pollution Community Analysis and construction identification identification detection metrics

14 | narratives and information pollution on vulnerable groups during the covid-19 pandemic Information pollution

false harmful

Created by popcornarts from the Noun Project

misinformation disinformation mal-information • False context • False Connection • Leaks • Imposter Content • Misleading Content • Harassment • Manipulated Content • Hate speech • Fabricated Content 1 2 3

Adapted from (UNDP, 2020) and (Derekhshan, 2017)

In this report we use the typology of informative 1. Disinformation: Information that is false pollution proposed in the “Guidance on Information and deliberately created to harm a person, Pollution” (UNDP, 2020) and (Derakhshan, 2017) social-group, organisation or country (UNESCO and Ministry of the Presidency of the 2. Misinformation: Information that is false, but Dominican Republic, 2020): not created with the intention of causing harm. 3. Mal-information: Information that is based on real facts, but manipulated to inflict harm on a person, organisation or country

narratives and information pollution on vulnerable groups during the covid-19 pandemic | 15 PART 1 RETURNED MIGRANTS TO VENEZUELA

16 | narratives and information pollution on vulnerable groups during the covid-19 pandemic INTRODUCTION MIGRATION IN TWO ERAS

THE ECONOMIC CRISIS AND MIGRATION first period

The migration of Venezuelans is one of the consequences of the complex crisis that the country is going through. On the economic level, the GDP has been contracting since 2014, the cumulative contraction in the period 2014-2019 is 55.3% according to the Central Bank of Venezuela. In the same period, oil production, the country’s main source of income, fell by 79.2%. High inflation, the drastic fall in the purchasing power of wages and the disruption of basic public services (in the supply of electricity, water and more recently gasoline) has considerably reduced income (minimum wage less than $2 per month at the end of 2020) and the quality of life of citizens; therefore, increasing income poverty. On the political level, hyperpolarisation and mutual ignorance between the government and the opposition has affected the functioning of the country’s institutions and deteriorated social cohesion. The cumulative contraction in the period On the social front, the gains made during the oil price 2014-2019 is 55.3% according to the boom of 2002-2008 in poverty reduction, among other Central Bank of Venezuela. aspects, have been reversed. The extensive system of In the same period, oil production, the social programmes developed since 2002 has been significantly reduced. country’s main source of income, fell by 79.2%. At the international level, the US government’s economic and financial sanctions and the disregard of the government by the Trump administration, Canada, the EU and some countries in the region, aggravate the situation.

This crisis has driven 5.4 million Venezuelans to emigrate from the country, according to UNHCR. The nearby South American countries of Colombia, Brazil, Ecuador and Peru have been the main destinations for migrants.

narratives and information pollution on vulnerable groups during the covid-19 pandemic: venezuela case | 17 18 | narratives and information pollution on vulnerable groups during the covid-19 pandemic

MIGRATION IN TWO ERAS THE PANDEMIC AND THE RETURN OF MIGRANTS second period

With the arrival of COVID-19 and the social isolation measures taken to deal with it, living conditions in the host countries of Venezuelan migrants deteriorated. Informal labor (an important source of work for low-income migrants) and the economy in general were seriously affected throughout the region. Social programs to support the vulnerable population during the pandemic hardly covered migrants. This adverse economic context encouraged many Venezuelan migrants who had left the country to return, hoping to better cope with the pandemic in the shelter of their family and social networks in Venezuela. In the period March-July 2020, Colombia’s Migration Department reported that 95,000 migrants had returned to Venezuela from that country. Reverse migration has become a topic of intense discussion on social networks and in the media.

Conditions of return Migrants attempting to return face a complicated situation. Tensions between the Colombian and Venezuelan governments have restricted the flow of trade and the mobility of people across the border. At the beginning of the pandemic, the Venezuelan government set up the so-called Integral Social Attention Posts (PASI), where returning migrants entering the country by land must undergo quarantine. According to Caritas and the Venezuelan Border Observatory, the insufficient number of accommodation centres for quarantine and the deficiencies in their staffing and functioning lead to a difficult stay in these centres.

narratives and information pollution on vulnerable groups during the covid-19 pandemic: venezuela case | 19 In order to avoid the aforementioned difficulties, Patria) (Ministry of Popular Power for Foreign some migrants choose to enter the country Affairs, 2020). The Plan consists of “an air and through non official crossings called “trochas”, land bridge for the voluntary return of all those whereby migrants often fall prey to criminals and migrants and their families who lack their own mistreatment. The absence of health controls means of return”. According to official figures at these crossings makes it impossible to detect (Ministry of People’s Power for Foreign Affairs, people infected with the virus, generating an 2020), as of February 2020, more than 17,000 attitude of fear and stigmatisation of those who Venezuelans have benefited from this plan. enter illegally. In this section we analyse the impact on social Fear of contagion and aversion to returning networks of the discussion on the return of migrants is present in the media and social migrants to Venezuela, with special emphasis on networks whom are singled out as “sources of information pollution in the context of the contagion” and seen as threats. They are stigma- pandemic. We begin by showing the results tised as “biological weapons” or “bioterrorists”. obtained on Twitter, continue with the findings in Google Trends and with the monitoring of Since August 2019, the national government has media with a digital presence, then we propose a tried to facilitate the conditions for the return of communication strategy to mitigate the impact of vulnerable Venezuelans in host countries with the information pollution on returned migrants, and “ Return to the Homeland Plan “ (Plan Vuelta a la close with the conclusions.

20 | narratives and information pollution on vulnerable groups during the covid-19 pandemic database construction

SOURCES OF INFORMATION Twitter

The data Timeline A database was constructed to store all March-November 2020 tweets posted in Venezuela about migrants Number of tweets in the period March-November 2020. The collected 796,017 period starts in March because it is the month in which the coronavirus outbreak breaks out Number of users 87,961 in the country. Relevant keywords, users and tags were used to identify tweets related to Montly average of tweets 88,446 migrants. Additionally, a second database was constructed in a similar manner for the Weekly average of tweets 17,727 period June 2019 to February 2020. Compar- isons between the two databases are used June 2019-February 2020

to measure the impact of the pandemic on Number of tweets discussions about migrants. collected 405,000

Montly average of tweets 44,889

Weekly average of tweets 9,539

Relative size of first Database with respect to number of 4% tweets posted by Covid-19 in Venezuela

narratives and information pollution on vulnerable groups during the covid-19 pandemic: venezuela case | 21 analysis and metrics

Venezuela: monthly evolution

Number of tweets 200

182 171,835

164

146 134,641 127,579 128

110 97,500 92 79,441 69,228 74 The graph shows that the highest number of tweets about returned migrants in 56 45,700 41,000 Venezuela were published in July. 38 29,093 During this month the cases reported in Venezuela, Colombia and Brazil have an 20 increasing trend. July May June April March August October November September

22 | narratives and information pollution on vulnerable groups during the covid-19 pandemic analysis and metrics

Covid-19 weekly cases evolution

from March toCovid-19 November weekly cases 2020 evolution from march to november 2020

Venezuela 4K

Número de casos 2K

300K

200K Brazil

100K Número de casos

0K 80K

60K

Colombia 40K

Número de casos 20K

600

Trinidad and 400 Tobago

Número de casos 200

0 ber 3-9/may 5-11/july 7-13/june 5-11/april 19-25/july 12-18/july 1-7/march 24-30/may 17-23/may 10-16/may 21-27/june 14-20/june 2-8/august 12-18/april 19-25/april 8-14/march 9-15/august 15-21/march 22-28/march 4-10/october 23-29/august 16-22/august 28/june-4/july 1-7/november 11-17/october 18-24/octo 25-31/october 31/may-6/june 26/april-2/may 8-14/november 6-12/september 22-28/november 15-21/november 26/july-1/august 29/march-4/april 13-19/september 20-26/september 29/november-5/de.. 30/august-5/septe.. 27/september-3/oc..

For Venezuela, Brazil and Colombia the number of cases are expressed in thousands The last weekCountry of the period (29 / November-5 / December) includes only the last two days of the month of November Source: WHO COVID-19 Dashboard (https://covid19.who.int/) For Venezuela, Brazil and Colombia the number of cases are expressed in thousands. Country Venezuela Venezuela Brazil The last week of the period (29/November-5/December) includes only the last two days of Brazil Colombia Colombia the month of November Trinidad and Tobago Trinidad and Tobago Source: WHO COVID-19 Dashboard(htps:/covid19.who.int/)

narratives and information pollution on vulnerable groups during the covid-19 pandemic: venezuela case | 23 analysis and metrics

Frequency words

Venezuelans 2.63 Migrants 2.04 Colombia 0.76 Children 0.68 Trinidad and Tobago 0.66 Government 0.55 Deportees 0.35 Maduro 0.35 Covid 0.33 Pandemic 0.32 President 0.28 Refugees 0.25 Quarantine 0.25 World 0.23 State 0.23 Homeland 0.22 Health 0.21 Situation 0.20 Countries 0.20 Connationals 0.19

0.50 1.00 1.50 2.00 2.50 3.00

Proportion with respect of total words (%)

The table highlights the importance of the terms “Colombia” and “Trinidad and Tobago” in the discussion on migrants. It also highlights the importance of “Government”, “State” and President “Maduro”. “Chil- dren”, “Refugees” and “Deportees” is the most mentioned vulnerable population in the database.

24 | narratives and information pollution on vulnerable groups during the covid-19 pandemic analysis and metrics

Issuance of tweets by States

Curaçao Grenada

Falcón Nueva Esparta La Guaira Distrito Capital Trinidad and Tobago Zulia Yaracuy Sucre Lara Carabobo Miranda Aragua

Trujillo Cojedes Monagas Portuguesa Delta Guárico Anzoátegui Mérida Amacuro Barinas Táchira

Apure Guyana Bolívar

270

17,774 Colombia

Amazonas

Brazil

148,748

The Border States with the highest flow of human and commercial mobility, Táchira and Zulia, generate the highest number of tweets. The states of Carabobo, Aragua and Distrito Capital, highly populated states, also have a high participation in the discussion.

narratives and information pollution on vulnerable groups during the covid-19 pandemic: venezuela case | 25 analysis and metrics

Pandemic impact on the discussion

381 Total 540

282 Migrants 272

29 Trocheros 133 June 2019/February 2020 March/November 2020 42 Border 77

28 Return 58 0 100 200 300 400 500 600

topic of discussion number of tweets (expressed in thousands)

The impact of the pandemic on Twitter discussion about migrants is clear. The topic that shows the greatest variation between the two periods is “trocheros”, a variation that exceeds the sum of all the variations of the other terms.

26 | narratives and information pollution on vulnerable groups during the covid-19 pandemic analysis and metrics

Percentage variations of the themes

Return 1 107.14

Border 83.33

Trocheros 358.52

Migrants -3.55

Total 41.73

-500 0 50 100 150 200 250 300 350 400

thematic percentage variations of the themes

Some users use the term “trocheros” to refer to people who enter the country through non-official crossings (“las trochas”), while others use it to refer to those who drive (and charge for it) people through these crossings. Despite the ambiguity of the term “trocheros”, the wide variation reflects concern about migrants returning through non-official crossings.

narratives and information pollution on vulnerable groups during the covid-19 pandemic: venezuela case | 27 narrative identification

Main Narratives

1 Colombian virus 7% Conspiracy theory about 2 virus origin 10%

3 Migrants as vector of contagion 83%

Narratives allow us to identify discursive patterns and common underlying messages in groupings of tweets (Dayter, 2015), we will explore how informational pollution is embedded in them and their relationship with the main user communities.

28 | narratives and information pollution on vulnerable groups during the covid-19 pandemic narrative identification

Representative Tweet

1 Conspiracy theory Conspiracy theories about the origin about virus origin of the virus are grouped into this narrative. The common argument is Search that the virus was artificially created 3 or spread by powerful groups to harm humanity.

It combines two sub-narratives:

Conspiracy theory about virus origin 1. The virus was deliberately created and spread by the Chinese government. They Trending Topics consider the pandemic as an aggression by China against ••••••• ••••• | ••••• the world. "I have no doubt that it is a virus created in a laboratory 2. The virus is created by the by the Chinese communist regime as a biological U.S. government as part of weapon". the Trump administration’s trade war against China.

Both narratives are local expressions of similar narratives at the global level, although they are connected to the political polarization in Venezuela.

narratives and information pollution on vulnerable groups during the covid-19 pandemic: venezuela case | 29 narrative identification Colombian virus

••••••• •••••

2 Colombian virus Venezuela ·········· ····· ······ This narrative accuses the Colombian government of sending Venezuelan "The government and Venezuelans have to understand that in this migrants through illegal crossings back multifactorial war against Venezuela, Colombia is using to the country to spread the virus in returningJessica Amber migrants | @jscambr as "biological Lorem ipsum dolor sit amet, consectetuer adipiscing elit, sed diam Venezuela. This return is presented as weapons".nonummy nibh euismod What tincidunt do . we do ? Let's get ourselves10 screwed526 and139 let the main cause of the epidemic in the them infect 25,000,000 people ? ••••••• ••••• | @silve_anto country. It reinforces the idea that the Lorem ipsum dolor sit amet Enough of this rubbish , please". pandemic is an external threat and an expression of the confrontation between the governments of Colombia and Venezuela.

10 526 139

·········· ··· ····· ······· 3 ••••••• ••••• | @••••••_•••• The narrative states that migrants Migrants who “A #TrocheroInfectado is entering through illegal crossings are return as vector a bioterrorist who can take of contagion the lives of you and your loved the main vectors of the virus. These ones. Enter through authorized passes, welcome to your migrants are sometimes described as homeland, but submit to quarantine, and don't come “bioterrorists” or “biological weapons” to infect Venezuelans” who intend to spread the virus in Venezuela. They accuse returning migrants of not complying with sanitary norms and of knowingly infecting their relatives and neighbors. There are frequent calls to identify and report these migrants to the authorities.

The narrative has evolved; at the beginning it was more aggressive towards migrants, then it became more moderate.

30 | narratives and information pollution on vulnerable groups during the covid-19 pandemic pollution identification

Information pollution

No pollution With pollution Total Country 784,809 11,208

1 Conspiracy theory about virus origin 65 518 2 Colombian virus 4 398 3 Migrants who return as vector of contagion 461 4,195 Does not belong to any narrative 784,279 6,097

More than half of the information pollution is not related to any of the narratives. The pollution linked to the main narratives is concentrated on returned migrants as a vector of contagion.

Created by popcornarts from the Noun Project

Disinformation Misinformation Mal-information Total Country 8,084 941 2,183 1 Conspiracy theory about virus origin 510 - 8

2 Colombian virus 259 - 139

3 Migrants who return as vector of contagion 3,431 - 764

Does not belong to any narrative 3,884 941 1,272

Migrants who return by non-official crossings are the main target of information pollution on the subject. Fear of contagion is the feeling used against migrants. Some foreign governments are also the target of conspiracy theories. Disinformation (false information with intent to harm) as a major type of information pollution is a consequence of the use of conspiracy theories as the main strategy of information pollution in all narratives. The nature of these theories blames and harms certain groups, foreign governments or migrants who return the country through non-official crossings, as the cause of the pandemic.

narratives and information pollution on vulnerable groups during the covid-19 pandemic: venezuela case | 31 analysis and metrics

Proportion of tweets with pollution by narratives

1 Conspiracy theory about virus origin

Does not belong to 4 any narrative 2 Colombian virus 54 %

3 Migrants as vector of contagion

For every 71 Misinformation tweets, 1 is With pollution 8.4% polluted 1.4%

Created by popcornarts Disinformation from the Noun Project No Pollution Mal-information 98.6% 72.1 % 19.5%

32 | narratives and information pollution on vulnerable groups during the covid-19 pandemic community detection

COMMUNITIES What are user communities?

Communities are groups of users who share opin- their small size. The detected communities gather ions and support each other by spreading messag- together 8,381 members, which represents 59% es. Communities are identified (Blondel, 2008) (Zal- of the users participating in the discussion, and mout, 2013) based on retweets between users. The have a very polarised political behaviour. Their influence of communities on Twitter is often very most striking characteristic is their positioning in important. In the discussion on returned migrants the confrontation between the government and to Venezuela, 5 communities and the grouping the opposition. The communities are relevant for “Other users” were detected, which brings together identifying narratives and patterns of propagation isolated users and very small groups of users that of information pollution. do not reach the structure of a community due to

narratives and information pollution on vulnerable groups during the covid-19 pandemic: venezuela case | 33 community detection

Information pollution flows

MIGRANTS WHO RETURN BY NON-OFFICIAL CROSSINGS

AS CONTAGION VECTOR

T T

R R

R A

1 N

G G

N E

S EN N

M S E

E R

R M A I T A

T I

T O T

O E POLLUTED TWEET T R

C C

O O O C C

O C M

M O M M

M U

M N

N POLLUTED TWEET I

I M T

U Y N Y U I N

TY ITY

O N O

O I

T O

T I T G T U G L E U E L N L L N O E E E P R O R A P A D T T EA E AD E OR SPR OR SPRE

POLLUTED TWEET POLLUTED TWEET

CONSPIRACY THEORIES

ABOUT THE FOREIGN SOURCE

T Y

R T

A OF THE VIRUS

N G TY

I U S

S EN N

M M E U M

I M 2 M M T T T A M

T T T

O O

O O O

C C R

POLLUTED TWEET R G R

R O E

T R O

O U T

R I P

U I

N M

G P E

G S I N N G A TR

OTHER USERS

The image represents the information pollution flows between communities. The arcs point to the source of the pollution , for example, an arc from Community B to A means that the polluted tweet was generated in A and broadcasted B. To simplify the graph, the polluted tweets were reduced to two groups. →→ The first group, in green, represents tweets that stigmatize returned migrants. →→ The second group, represented in brown, brings together the narratives “Conspiracy theories about the origin of the virus” and “Colombian virus”. Community C is the generator and relay of contamination of the second group of tweets, while A, B and the node “Other users” are generators and re-transmitters of contamination on returning migrants. D and E do not generate or spread pollution.

34 | narratives and information pollution on vulnerable groups during the covid-19 pandemic analysis and metrics

Polarisation and information pollution

We can group the identified communities Political polarisation in Venezuela is the main according to their level of political cause of information pollution about returned polarization. The group made up of migrants. In the Venezuelan political conflict, communities A, B and C are very ”politi- the opposing parties do not recognise each cally polarized and have a hyper-partisan other as adversaries, nor as members of the behaviour in the government-opposition same society having common interests but confrontation. The second “non-partisan” different positions. The fact of the matter group is made up of D and E. For this group, is that they do recognise each other only as political confrontation is not a central enemies. A significant part of the opposition aspect of the community. It is noteworthy does not recognise the government and calls that only the group of partisan communi- on the international community to disown it, ties and “Other users” are the generators to sanction it economically and to seize the and transmitters of pollution. The fact that nation’s assets abroad. The government, for the two “non-partisan” communities do not its part, accuses this sector of the opposition generate or transmit pollution suggests of being an instrument of the US government that they are vigilant in this regard. The to damage the country. As a consequence of media and human rights organizations this fierce confrontation, disinformation and belong to the non-partisan group. mal-information messages aimed at harming the “enemy” or its international allies are frequent. The main instrument employed in this confrontation is conspiracy theories

narratives and information pollution on vulnerable groups during the covid-19 pandemic: venezuela case | 35 analysis and metrics

Very significant are the increases in Google Trends is a tool that allows you to quantify »Google searches for key terms that have marked the Google searches for a term in a specific country and in discussion and pollution of returned migrants. The a specific time period. It is of interest because it allows terms “Bioterrorist” and “Biological weapon”, which identifying the information demands of Internet users appear in the first two places, show the interest on terms related to a particular topic. in the discussion and the very positive attitude of Internet users in gathering information about the discussions in social networks. This attitude is one of the recommendations of the specialists to face information pollution.

Percentage changes in Google Trends searches between the periods June 2019-February 2020 and March-November 2020.

terms

Bioterrorist 1850

Biological Weapon 1850

Quarantine 1682

Pandemic 1403

Aid for venezuelans 203

Deported 4

0 400 800 1200 1600 2000 Percentage variation

36 | narratives and information pollution on vulnerable groups during the covid-19 pandemic analysis and metrics

Percentage changes in Google Trends and topical searches on Twitter between June 2019-February 2020 and March-November 2020.

Returnees 357

66 Trocheros 1014

3 Migrants 53

0 100 200 300 400 500 600 700 800 900 1000

Percentage variation

Although all the terms or topics indicated present positive variations in Twitter and Google Trends, it is noteworthy that the magnitudes of the variations in Twitter are substantially greater. The gap between both magnitudes also reflects the differences between the demands or requirements for information, represented by Google Trends, and the supply or exposure to information to which users are exposed, represented by Twitter.

narratives and information pollution on vulnerable groups during the covid-19 pandemic: venezuela case | 37 analysis and metrics

Returnees

Curaçao Grenada

Falcón Nueva Esparta La Guaira Distrito Capital Trinidad and Tobago Zulia Yaracuy Sucre Lara Carabobo Miranda Aragua

Trujillo Cojedes Monagas Portuguesa Delta Guárico Anzoátegui Mérida Amacuro Barinas Táchira

Apure Guyana Bolívar

Colombia

Amazonas

0 Brazil

100

Search interest

38 | narratives and information pollution on vulnerable groups during the covid-19 pandemic analysis and metrics

Migrants

Curaçao Grenada

Falcón Nueva Esparta La Guaira Distrito Capital Trinidad and Tobago Zulia Yaracuy Sucre Lara Carabobo Miranda Aragua

Trujillo Cojedes Monagas Portuguesa Delta Guárico Anzoátegui Mérida Amacuro Barinas Táchira

Apure Guyana Bolívar

Colombia

Amazonas

0 Brazil

100

Search interest

narratives and information pollution on vulnerable groups during the covid-19 pandemic: venezuela case | 39 analysis and metrics

Trocheros

Curaçao Grenada

Falcón Nueva Esparta La Guaira Distrito Capital Trinidad and Tobago Zulia Yaracuy Sucre Lara Carabobo Miranda Aragua

Trujillo Cojedes Monagas Portuguesa Delta Guárico Anzoátegui Mérida Amacuro Barinas Táchira

Apure Guyana Bolívar

Colombia

Amazonas

0 Brazil

100 Search interest

40 | narratives and information pollution on vulnerable groups during the covid-19 pandemic analysis and metrics MEDIA

A total of 37 digital media were consulted using the Meltwater tool and complemented with other Internet searches. These media include digital versions of print media and exclusively digital media. The news and opinion genres were 500 considered, in addition to radio media in their digital version.

450 457 500

400 450 457 401 378 371 368 370 350 400 401 351 378 336 371 368 370 327 350 307 351 308 300 336 289 293327 289 307 308 300 267 293 250 250 289 289 250 267 230 200 250 250 250 200 230 200 194 191 195 200 182 179 194 191 195 182 179 150 145 150 142 149 150 145 149 137150 140 130 142 126 140 130 137 123 121 124 112 123 126 112 117 121 124 100 112 112 108117 100 108 92 95 88 88 92 87 89 95 89 81 8176 80 80 80 80 87 78 76 7171 78 64 64 57 50 53 50 53 57 20 20 24 24 0 0 Migrantes venezolanos retornados - Noticias (Mar 1,2020-Aug31,2020) Migrantes venezolanos retornados - Noticias (Aug 30, 2019 - Feb 29,2020) Returned Venezuelan Migrants - News (Mar 1,2020-Aug31,2020) Returned Venezuelan Migrants - News (Aug 30, 2019 - Feb 29,2020)

The news reports detected about migrants returned to Venezuela increased in the six months between March-August 2020 in relation to September 2019-February 2020.

narratives and information pollution on vulnerable groups during the covid-19 pandemic: venezuela case | 41 narrative identification

Fact Checkers

The concern about information contamination has been present on the internet (Tardáguila, 2019). Newspaper reports , forums and new organizations in the area of “fact checkers” are addressing the issue. Among the web portals and Twitter accounts that are dedicated to fact-checking are espaja.com, cazadoresdefakenews.info ,efectococuyo.com and the Venezuelan Fakenews Observatory (fakenews.cotejo.info and https://twitter.com/ observatoriofn? lang = en). 4

Main5 narratives It is emphasized that the virus is of foreign origin with the 6 implicit message that the pandemic is not a problem that External origin of virus: arose in the country. Colombia and Brazil, the countries with Brazil7 and Colombia the greatest human border mobility and trade flows, are mainly mentioned. Unlike the equivalent narrative on Twitter, Brazil is included. China and the US: This narrative is very similar to the equiva- External origin of virus: China and USA lent Twitter narrative. Similar to the equivalent narrative on Twitter, illegal migrants who enter the country and fail to comply with health regula- tions are accused of being the main vectors of contagion of Migrants as vector of contagion the virus. Calls to denounce illegal migrants are also present. However, the label “bioterrorists” or “biological weapons” is very rare in this narrative.

42 | narrativasnarratives ycontaminación and information informativa pollution on sobre vulnerable grupos vulnerables groups during durante the covid-19 la pandemia pandemic de covid-19 narrative identification

External origin of virus: Brazil and Colombia 4,932

Migrants as vector of contagion 638

External origin of virus: China and USA 307

Others 1,123

The narratives identified in the digital media are very similar to those on Twitter. As in the social network, the most frequent types of information pollution are disinformation and mal-information. Unlike Twitter, the narrative with the highest volume is “External origin of the virus: Brazil and Colombia”. It is also noteworthy that the narratives on social networks are much more aggressive than the corresponding ones in the media.

On the other hand, although the critical area for the return of Venezuelan migrants is the borderline, the information pollution detected in digital media with a national scope is greater than in those with a regional scope.

narratives and information pollution on vulnerable groups during the covid-19 pandemic: venezuela case | 43 COMMUNICATION STRATEGY to mitigate information pollution: Venezuela case

→→General objective Provide information to the public that allows the identification of information pollution through communication actions.

→→Specific objectives • Sensitize public opinion to mitigate the effects of stigmatization on the returned migrant population. • Report on how to combat the effects of information pollution. • Contribute to building capacities in community teams and communication professionals to counteract the effects of information pollution, in regions especially affected by stigmatization.

→→Target audience • Community leaders and volunteers in service activities, in social organizations, sports and NGOs. • Social communicators, especially in border areas. • National, state and municipal government institutions involved in responding to the pandemic. • State and municipal health institutions, involved in the care of the pandemic.

→→Action plan Exposure Objective Approach Activity Pieces Channels time Indicators

• Media of national Develop a Testimonial / regional scope Campaign in which Sensitize public • Boardcast • Numbers of neighbors, family and opinion to mitigate External television messages issued migrants promote Between 1 the effects of communication • Radio 20” • Social media • Number of likes positive messages to 2 1 stigmatization for behavior • Videos 40” • Radio and amount of about preventing months towards the returned change • Regional news content shared discrimination caused migrant population portals / border by mal-information sections and disinformation.

44 | narratives and information pollution on vulnerable groups during the covid-19 pandemic Exposure Objective Approach Activity Pieces Channels time Indicators

Number of visits to • Web UNDP the microsite. • National / regional Number of views • Microsite media portals and clics on banners. Develop an informa- • Banners • Social media Number of likes and tion campaign on: • Post on social • Community amount of content media centers External shared. • What is informative • Posters • Bus Stations Communica- Amount of printed Report the effects of pollution? • Leaflets • Municipal markets tion, for material information pollution • How does social • Signs • Parcel sites behavior distributed. on social and coexistence affect? • Loudspeakers • Supermarkets, 6 months 2 change / Number of community life and • Forms of stigmati- • Talks for bakeries Communication photographs of how to combat it zation community • Public offices for advocacy exposed material. • How to stop teams • Health centers Number of talks pollution and • Dialogues with • Traffic lights made stigma? community • Talks Number of officials and • PodCast workshops carried servants • WhatsApp out workshops Interviews with participants

Contribute to building capacities in Create alliances with a communication knowledge center and • Webex professionals and Communication Informative regional media to • WhatsApps community teams for for Guidance 6 months Consecutive cycles 3 design a course for • Class by Radio the identification and appropriation Laboratory social promoters and containment of the journalists spread of informational disorders.

narratives and information pollution on vulnerable groups during the covid-19 pandemic: venezuela case | 45 CONCLUSIONS

• During the pandemic, there has been a • Returning migrants by non-official crossings substantial increase in the volume of are victims of stigmatisation. On Twitter discussion on social networks and in the and in the media, messages are frequent media about the return of migrants to in which migrants who do not comply with Venezuela. The number of tweets, Google sanitary controls are perceived as vectors of Trends searches for terms and press contagion, which, combined with conspiracy releases about the return of migrants theories, give rise to the qualification of increased considerably. “biological weapons” or “bioterrorists”.

• The narratives and information pollution • Information pollution is more aggressive on identified on Twitter and in the media can Twitter than in the media. The lower legal be classified into two main groups. The first, liability for publishing content on networks conspiracy theories about the external compared to the media seems to be the origin of the virus. The second, which cause of this difference. points to returned migrants by non-official crossings as the main vector of contagion. • Fear of contagion is the feeling magnified by Conspiracy theories are the privileged information pollution. contamination strategy in both groups.

46 | narratives and information pollution on vulnerable groups during the covid-19 pandemic • The discussion about the pandemic and • Google Trends confirms the trends of Twitter, returned migrants is closely linked to the although the demand for information in country’s highly polarised political debate. Google Trends is substantially lower than the Political polarisation in Venezuela is the supply of information on Twitter. The gap main cause of information pollution about between information demand and supply returned migrants. needs to be analysed in more detail.

• The user communities detected on Twitter • Google Trends evidenced the interest on the discussion of migrants are also heavily generated by the discussion on migrants and influenced by political polarisation. Highly the very positive attitude of internet users in polarised communities spread information gathering information about the discussions pollution, while non-polarised communities on social networks. This attitude is one of do not generate pollution. the specialists’ recommendations for dealing with information pollution. • Strategies to address information pollution must take polarisation into account. • The communication strategy to mitigate Reducing political polarisation is a condition the effects of information pollution on for reducing pollution. Venezuelan migrants returning to the country during the pandemic should provide information to the public to identify information pollution and be implemented at the national and subnational levels with public, private, community and NGO institutions.

narratives and information pollution on vulnerable groups during the covid-19 pandemic: venezuela case | 47 PART 2 WOMEN AND THE LGBTIQ+ COMMUNITY IN PANAMA

48 | narratives and information pollution on vulnerable groups during the covid-19 pandemic INTRODUCTION

Panama declared a state of emergency for COVID-19 on In the first months of the pandemic, public debate on March 13, 2020 and the full quarantine went into effect on social media focused on approval of health management, March 24* . Schools were closed on March 10. The gradual followed by questions about the allocation of resources to opening of the country began in September, starting address the crisis and the lack of transparency. with the lifting of gender-based movement, ordering free movement on Saturdays and maintaining total closure on During the months of confinement, a forceful discussion Sundays. In December the measures were resumed due to seemed to jeopardise the authorities’ response. There the increase in cases. were calls for disobedience with public demonstrations, neglect of traffic times and the holding of public events Before the pandemic, the government, in office since 1 where hundreds of people gathered. July 2019, was still adjusting its work plan. Panama was among the fastest growing economies in Latin America; Social media reports of domestic violence increased, but conversely, faced at the same time its reality as a country formal complaints decreased. Part of this situation was struggling with poverty, corruption and inequality. due to the fact that formal reporting channels remained A dialogue to address the construction of a new closed. Femicides increased by 64% compared to 2019. constitution had begun. A first assessment indicates that, confined to their Panama has a population of 4.28 million (68% live in urban aggressors, women continued to suffer from gender-based areas); 2.63 million have internet service (62%); 2.40 violence without being able to report their cases. Social million are active users of social networks. Facebook has networks began to be populated with messages against 1.8 million registered users, Instagram 1.6 and Twitter or in favour of gender equality, centered on the demands 469,000. There are currently 4.86 million active mobile of gender equality rights groups and feminist groups and phone lines, representing 114% of the population, 33% users who questioned the lack of information regarding of Panamanians are connected to the internet via mobile cases of violence. phones (Hootsuite - We are Social, 2020). At the same time, with the establishment of circulation Although the Twitter audience is not as representative in schedules for the purchase of food and medicines based number of users as Facebook or Instagram, Twitter is the on biological sex, complaints multiplied in social networks privileged platform in the country to discuss public, social and traditional media of transgender people being and political issues. discriminated against and not being allowed access to supermarkets and pharmacies on women’s shopping days

*Note : By 14 October, most of the restrictive measures had been lifted, leaving a total quarantine on Sundays, beach use and mass events.

narratives and information pollution on vulnerable groups during the covid-19 pandemic: panama case | 49 because their personal identity card showed their male Partnerships are especially important to get a better sex, and they were not allowed to shop on days assigned picture of the situation, so we interviewed journalists, for men because their physical appearance did not social researchers, women’s associations and the match that of the public assigned for that day. Several LGBTIQ+ community to get a scope of the problem and pronouncements by human rights organisations, how it is addressed by each specific group. and even the Inter-American Court of Human Rights, increased the dialogue on social media about Mainly we did not find, as in the case of Venezuela, this situation. a common narrative between COVID-19 and informative contamination towards the group of In both positions, for or against, we found relevant, women and LGBTIQ+ population directly. What the timely and valuable information in its source, content COVID-19 scenario raised, and mainly the quarantine, and approach; but we also found disinformation, was a situation that managed to make visible manipulation of information or pollution and, above all, conditions that were already underlying in society in hate messages. terms of violence and discrimination. The study led us to identify a marked power struggle between To try to better understand what was happening, this pro-human rights groups and conservative groups study allowed us to delve deeper into the situation of that use social networks as a stage for confrontation information pollution, sources, content, scope and in defence of their positions. consequences. To identify audiences, communities and user groups that were generating pollution to finally An important thing that the study allowed us to identify allow us to propose actions to support the cleaning was that the messages, polluted or not, are signed by of the information environment in the response to real users, not by robots or call centres. The generators COVID-19 and to raise the voice of vulnerable groups. of the messages identify themselves and are proud to do so. It is necessary to delve deeper into communities Another relevant starting point for focusing the study of dispersed users who do not seem to align themselves on women was that during the two previous electoral with identified communities or clusters of information processes (2004 and 2009) there was an increase in and who generate pollution. hate speech, the dissemination of fake news and attacks on candidates, in notorious cases against women An important group identified are social communicators, candidates. who contribute a large part of the polluted message, either by originating it or amplifying it with retweets. This Civil society groups working on content analysis or community deserves further study, but also provides an generating content for digital conversations were key to opportunity to work with them in the future to adjust the better understanding this information ecosystem. dialogue for human rights and respect for people.

50 | narratives and information pollution on vulnerable groups during the covid-19 pandemic database construction

INFORMATION SOURCES Twitter

The data Timeline Similar to the Venezuelan case, two databases March-November 2020 were built with tweets published in Panama. The first and second databases contain tweets Number of tweets 294,330 published in the period March-November 2020 collected and from June 2019 to February 2020, respec- Number of users tively. Both bases were built by selecting the 27,649 tweets that contained keywords related to Montly average of Tweets women or the LGBTIQ + community in Panama. 32,703 Weekly average of Tweets 5,856

June 2019- February 2020

Number of tweets collected 208,000

Montly average of Tweets 23,778

Weekly average of Tweets 5,231 Relative size of first Database with respect to number of tweets posted by 13% Covid-19 in Panama

narratives and information pollution on vulnerable groups during the covid-19 pandemic: panama case | 51 analysis and metrics

Panama: monthly evolution

70,000

64,000 58,371 58,000

52,000

46,000

40,000 34,812 32,165 34,000 30,463 29,001 33,797 28,000 29,339

22,000 23,271 23,111 16,000

10,000 July May June April March August October November September

52 | narratives and information pollution on vulnerable groups during the covid-19 pandemic analysis and metrics

Frequency words

Woman 2.30 Violence 0.67 Men 0.66 Girl 0.53 Years 0.42 Panama 0.42 Feminists 0.42 Pregnant 0.39 Health 0.32 Rights 0.28 Life 0.27 Help 0.25 Country 0.25 People 0.24 Gender 0.24 Respect 0.23 Feminism 0.21 Sexual 0.21 0.18 Male chauvinist 0.16 Family 0 0.5 1.0 1.50 2.00 2.50

Proportion with respect of total words (%)

The terms “Woman”, “Violence”, “Girl”, “Feminists”, “Pregnant”, “Gender”, “Feminism”, “Respect” and “Family” stand out, all of them relevant in the discussion on gender. Based on these words, checked against with the COVID-19 approach, the contents were grouped for both Twitter and traditional media.

narratives and information pollution on vulnerable groups during the covid-19 pandemic: panama case | 53 analysis and metrics

Issuance of tweets by Provinces and Indigenous Regions

Bocas Guna Yala del toro Panamá Madugandí Colón

West Wargandí Panamá Chiriquí Ngöbe-Buglé Coclé Emberá Veraguas

Herrera Darién

Los Santos

3,817 4 27,400

Only the provincial-level Indigenous Regions are included

The population of Panama is centered in the provinces of Panama and Colon. In connectivity, both provinces also lead. Coclé and Veraguas together have around 600 thousand inhabitants. Other studies managed by UNDP have found that participation in digital dialogue is high in both provinces.

54 | narratives and information pollution on vulnerable groups during the covid-19 pandemic analysis and metrics

Pandemic impact on the discussion

271 Total 325 58 Feminism 67 22 Feminicide/Femicide 23 52 Sexual abuse 58

85 Gender-based violence 101 3 Domestic violence 7 9 LGBTIQ+ 3 42 Gender 66

0 50 100 150 200 250 300 350

Number of tweets (expressed in thousands)

June 2019/February 2020 March/November 2020

narratives and information pollution on vulnerable groups during the covid-19 pandemic: panama case | 55 analysis and metrics

Percentage variations of the themes Between the periods June 2019-February 2020 and March-November 2020

Total 19.93

15.52 Feminism

Feminicide/Femicide 4.55

Sexual abuse 11.54

Gender-based violence 18.82

Domestic violence 133.33

LGBTIQ+ -66.67

Gender 57.14

-90 -60 -30 0 30 60 90 120 150 percentage variations of the themes

General discussion of women and the LGBTIQ+ community increased during the pandemic, although unevenly. “Feminism”, “Abuse”, “Gender-based violence”, “Domestic violence” and “Gender” had clear increases. “Femicide “ increased little, while LGBTIQ+ contracted.

56 | narratives and information pollution on vulnerable groups during the covid-19 pandemic narrative identification

Main Narratives

The main narratives are framed by the confrontation of conservative and progressive values in the discussion of gender equality. Although the discussion of gender is not directly related to COVID-19 and its effects, the volume and intensity of discussion increases during the pandemic. No “bots” were detected in the discussion.

Do not 1 represent me 55%

2 Feminazi 2%

3 Gender=Sex 11%

4 Marriage = Woman + man 32%

narratives and information pollution on vulnerable groups during the covid-19 pandemic: panama case | 57 Search narrative identification

MAIN NARRATIVES

Teorías conspirativas sobre el origen del virus

Representative tweet Trending Topics 1 Do not This narrative groups together positions represent me contrary to Panamanian feminist organiza- ••••••• ••••• | @•••••• tions or spokespersons. It includes two "Why all the feminists rushed sub-narratives: 3 to express "GENDER VIOLENCE" about the brutal homicide of 1. Alternative definitions of feminism Campo Limbergh and kept or criticisms for inadequate quiet when it was known that representation of feminist the murderer was a lesbian and principles. LGTB militant? #karmaisabitch"El virus colombiano 2. Anti-abortion and anti-gender diversity positions by conservative groups.

Both sub-narratives express criticisms of the ••••••• ••••• values espoused by feminism. @••••• 2 Feminazi Panamá •••••• •• ••• •• It is a narrative that has its origins in “To Panamanian women, this is conservative movements in the U.S. It 3 what these #Feminazis seek attempts to disqualify feminism as a radical LEGAL ABORTION, but not and inflexible ideology. It is linked to similar the one for rape NOT they seek narratives in other countries in the region. legal abortion when they feel like It seeks to undermine feminism and the it, because they are irresponsible LGBTIQ+ community and is linked to radical q can't put on a 0.50 condom right-wing movement parties. and avoid pregnancy. #FeminazisDoNotRepresentMe” . Tweet representativo

58 | narratives and information pollution on vulnerable groups during the covid-19 pandemic Jessica Amber | @jscambr

Lorem ipsum dolor sit amet, consectetuer adipiscing elit, sed diam nonummy nibh euismod tincidunt .

10 526 139 narrative identification

••••••• ••••• | @silve_anto

Lorem ipsum dolor sit amet

10 526 139 3 Gender is The main argument of this narrative is that ••••••• ••• ••••• •••••• equal to Sex ••••••• ••••• | @•••••_••••• there is no distinction between gender and . sex in the biological sense, thus denying “It is not that magically a gender diversity and the right to choose 3 person can change from sexual identity. male to female or vice versa. Medical science proves us It is a narrative that is widespread through- right, sex cannot change and out Latin America. if you are not happy with your gender it is called “Gender Dysphoria" #BiologyNotIdeology #NoToGenderIdeology."

4 Marriage is It is the most recently formed narrative. It between man ••••••• ••••• | @•••••••• rejects same-sex marriage. Conservative and woman "The union of a man and a Sectors are drivers of this narrative. woman produces life because 3 it comes from God, instead the homosexual union will never produce life, what it produces is death". #nogaymarriage

narratives and information pollution on vulnerable groups during the covid-19 pandemic: panama case | 59

10 526 139 pollution identification

Information pollution

No pollution With Pollution Total Country 293,628 772 1 Do not represent me 814 162 2 Feminazi 33 - 3 Gender=Sex 129 70 4 Marriage= man + woman 541 36 Does not belong to any narrative 292,111 504 A polluted tweet can belong to more than one narrative

The critical narrative about the values of feminism, “Do not represent me”, is the most polluted . The narrative against gender diversity, “Marriage=Male + Female”, also ranks high.

Created by popcornarts from the Noun Project Disinformation Misinformation Mal-information Total Country 584 47 141 1 Do not represent me 133 - 29 2 3 Feminazi - - - 4 Gender=Sex 55 - 15 Marriage= Man + Woman - - 36 Does not belong to any narrative 396 47 61

The most frequent form of pollution in the confrontation between conservative and progressive values in Panama is through false accusations of crimes against opponents. Usually these are accusations against feminists or members of the LGBTIQ+ community; however, there are also accusations against conservative religious sectors that are very active against LGBTIQ+ people. It is noteworthy that feminist advocates do not generate or spread information pollution. Therefore, disinformation (false information with intent to harm) is the dominant type of pollution, as seen in the table.

60 | narratives and information pollution on vulnerable groups during the covid-19 pandemic analysis and metrics

Proportion of tweets with pollution by narratives

marriage = woman + man

does not belong to any narrative 65% gender=sex

For every do not represent me 381 tweets, 1 is polluted

With pollution Misinformation

Created by popcornarts 0.3% from the Noun Project 6.1%

No Pollution Disinformation 99.7% 75.6% Mal-information 18.3%

narratives and information pollution on vulnerable groups during the covid-19 pandemic: panama case | 61 community identification

COMMUNITIES

In the discussion on gender during the pandemic in and relayed by B. It is interesting to note that only Panama, six communities and the grouping “Other Community E is a generator of polluted tweets, the users” were identified, which brings together isolated rest of the polluted tweets come from the “Other users and very small groups of users that do not reach users” grouping. This is a significant difference from the structure of a community due to their small size. the Venezuelan case, where most of the communities The arc from B to “Other users” indicates that the generated pollution. However, 4 of the 6 communities informative pollution was generated in “Other users” spread contamination generated by ”Other users”. and was relayed by Community B. In other words, the polluted tweet is generated by “Other users”

OR CO AT MM ER U N N E IT

G Y

T O T T I

Y R R R Y T

Y T

A A T N A I I U

IT N N L N N N O N S S S U M U L M U M M I M T I T O IT M T M T M T M TE CO ER CO G ER CO R P E N D A E E RA R TE OR SP EE T

POLLUTED TWEET D

N O

N T

O U

T L

L G E O POLLUTED T W N P E D R A

POLLUTED TWEET AT E

E OR SPR

N GE TY S I

N N M E

R U

I A M T T M

T O E O

CLASH VALUE R

G POLLUTED TWEET

G R

R T

T U

O I

I U N S

P N A

I R N T G

OTHER USERS

62 | narratives and information pollution on vulnerable groups during the covid-19 pandemic analysis and metrics

Percentage changes in Google Trends searches between the periods June 2019-February 2020 and March-November 2020.

290 Sexual identity

Women’s Rights 79

Gender ideology 15

Gender equality -68

-100 - 50 0 50 100 200 250 300 350

thematic percentage variations of the themes

The variations of key terms of the gender discussion are unequal. The term “gender ideology”, in third place, evidences the very positive attitude of Internet users in gathering information on discussions with information pollution in social networks. Taking advantage of this attitude is one of the recommendations of the specialists to face information pollution.

narratives and information pollution on vulnerable groups during the covid-19 pandemic: panama case | 63 analysis and metrics

Percentage changes in Google Trends and topical searches on Twitter between June 2019-February 2020 and March-November 2020

terms

68 Gender 57

-6 LGBTIQ+ -67

139 Domestic violence 133

Gender -based violence 1558 19 -36 Sexual abuse 12

Feminicide/Femicide 6 5 Feminism 49 16

percentage variations

The sign (positive or negative) of the variations in Twitter and Google Trends coincide except for the term “sexual abuse”. The negative sign in Google Trends is probably due to the intense search for the term in the first period, motivated by the discussion of a law on sexual abuse of minors in parliament. The large increase of the term “gender-based violence” is striking, which shows the interest of users to be informed about the subject. Unlike the Venezuelan case, Panama’s percentage variations are higher in Google Trends than in Twitter.

64 | narratives and information pollution on vulnerable groups during the covid-19 pandemic analysis and metrics

MEDIA

More than 6,000 news items were consulted Much of the information in networks originates following the research narratives, using the from the digital accounts of traditional media, who Meltwater tool complemented by other internet amplify the information in their digital channels, searches. These media include digital versions contributing elements to the conversation in so- of print media and exclusively digital media. cial networks. The media and the journalists who Informative and opinion genres were considered. represent them are the main communities where pollution was found.

Media exposure 800Media exposure 754

800 700 667 754 597 700 667 600 637 534 597 603 600 500 534 637 544 450 603 500 450 544 400

400 300 243 300 216 200 202 196 188 243 172 180 200 216 202 196 140 107 180 188 100 100 92172 88 90 88 54 60 70 54 56 58 140 68 100 100 31 10727 27 31 0 1792 88 90 88 54 60 70 54 56 58 68 ma31 rzoMarch abrilApril27 mayoMay 27 junioJune Jjuliouly Aagugust os to septiembreSeptember octubreOctober noviembre November 31 0 17 ma rch april Violencia basadamay en género june Feminicidiosjuly Discriminaciónaugust Acososeptember y abuso octobe r nov ember

Gender-based violence Femicide Discrimination Harassment and abuse

narratives and information pollution on vulnerable groups during the covid-19 pandemic: panama case | 65 COMMUNICATION STRATEGY to mitigate information pollution: Panama case

→→General objective Establish a communication plan to address the pollution of information generated in social networks, making a communication proposal that makes visible the factors that promote the lack of protection and therefore the violation of the human rights of women and the LGBTIQ+ community.

→→Specific objectives • Improve and strengthen the messages of prevention and defence of the human rights of the LGBTIQ+ population and women in Panama. • Identify the key actors and target audiences that will participate in the process to establish a communication strategy for each actor within the plan (women and LGBTIQ+ population, especially trans people). • Build and reinforce key messages from each community on vulnerability/discrimination (there is a problem - how to tackle it). • Develop a common campaign (education), with a focus on human rights and an emphasis on sexual diversity. • Develop crisis/counter-propaganda plan to address information contamination of these issues (thematic mapping - stakeholder mapping and media tier). • Mitigate the effects of information pollution by addressing stakeholders (stakeholder mapping) and establishing alliances (e.g. civil society organisations - media). • Support the formation of networks or action groups between women’s groups and LGBTIQ+ associations.

→→Target audience • First level: Women’s groups, unions, associations and similar objects of study. • Second Level: Social communicators (Media Tier). • Third Level: Strategic allies. • General public.

66 | narratives and information pollution on vulnerable groups during the covid-19 pandemic →→Action plan First phase: Self-awareness raising, training and recognition (choose two associations as a pilot scheme). Exposure Objective Approach Activity Pieces Channels time Indicators Sensitize the Internal Develop workshops Prepare pieces Internal media: 1 month Workshops held vulnerable popula- communication. for the identification for internal web, newsletter, Talks, trainings tion (object and and elaboration of communication: forum, carried out subject of study) key messages and Key messages discussion Number of about the problem. prepare (frequently asked participants. spokespersons. questions manual), Recognize and spokesperson address it. manual.

Provide internal Prepare responses communication for each prospective tools. LGBTQ + scenario or crisis Train pilot organizations on the effects Develop of information communication pollution and how to manual on Violence, combat it. Gender, Human Rights, LGBTIQ+

narratives and information pollution on vulnerable groups during the covid-19 pandemic: panama case | 67 Second phase: advocacy and relationship with stakeholders, including strategic allies Exposure Objective Approach Activity Pieces Channels time Indicators External Cycle of 1 to 1 Develop Microsite Web 2 months Number of visits Communication interviews, forums, to the microsite. conferences / Activate social National and workshops to networks and local media (TV) Number of views position key promotional and clics on concepts and channels (WhatsApp Social media banners. messages related to and Telegram) for information the dissemination of Radio Number of likes contamination in the the campaign and amount of context: Women and Panama Metro content shared. violence / Human Produce Graphic rights and sexual Arts: (Banners, Muppi Amount of printed diversity post on social material media, posters, Billboard distributed. Training workshops leaflets, signs, for social communi- animations) Number of cators in traditional photographs of and digital media. Multimedia exposed material. Development - Video (campaign) Number of talks -Jingle, podcast, made testimonials. . Number of workshops carried out

Interviews with participants

68 | narratives and information pollution on vulnerable groups during the covid-19 pandemic CONCLUSIONS

• During the pandemic, the volume of discussion • Feminist advocates do not generate or spread on gender issues in Panama has increased in a information pollution, despite being victims differentiated manner. “Feminism”, “Abuse”, of pollution. “Gender-based violence”, “Domestic violence” and “Gender” had significant increases. The • Only one community was found to generate number of tweets, searches for terms in Google pollution, the rest of the information pollution is Trends and press releases related to gender generated outside the main communities, increased. by isolated users. However, most of the communities retweet polluted tweets. • The main narratives detected are not directly related to COVID-19 and its effects. The • Google Trends confirms Twitter trends for narratives are framed in the confrontation of almost all the terms considered. Google Trends conservative and progressive values in the evidenced the interest of those who have discussion of gender equality, abortion rights, generated this discussion on gender and the very values promoted by feminist organisations positive attitude of internet users in gathering and spokespersons, and recognition of gender information about discussions on social diversity and respect for the LGBTIQ+ networks. Taking advantage of this attitude is community. one of the specialists’ recommendations for dealing with information pollution. • Pollution is present in the main narratives detected. The most frequent form of pollution • All the groups and individuals interviewed are in the confrontation between conservative clear about the situation, but have few or no and progressive values in Panama is through communication tools to address it. This is one unfounded accusations of murder against of the main reasons why defining an action plan opponents. Usually these are accusations based on a communication strategy can educate against feminists or members of the LGBTIQ+ the population through digital tools. community; however, there are also accusations against conservative religious sectors that are very active against LGBTIQ+ people. Disinformation (false information with intent to harm) is the dominant type of pollution.

narratives and information pollution on vulnerable groups during the covid-19 pandemic | 69 Some comparative comments between Panama and Venezuela

• In both countries, discussion and information pollution about the respective vulnerable populations increased with the pandemic. However, only in the Venezuelan case are the main narratives about vulnerable groups directly related to COVID-19.

• In both countries, the debate and information pollution is more aggressive on Twitter than in the media.

• Disinformation is the dominant type of information pollution in both countries.

• The findings in the different sources of information considered do not contradict each other.

• The main cause of information pollution in Venezuela is political polarisation, while in Panama it is the confrontation between conservative and progressive values.

• Both countries have communities of users who do not spread pollution, showing a careful and responsible attitude.

• Levels of political polarisation seem to impact the patterns of generation and propagation of information pollution. The higher the polarisation, the more pervasive the pollution among communities.

• In both countries, the population showed interest in searching the web for information about discussions with information pollution, which is evidence that not all people passively accept information pollution.

70 | narratives and information pollution on vulnerable groups during the covid-19 pandemic REFERENCES

Arroyo-Sánchez AS, C. J. (2020). Infodemia, la otra pandemia durante la enfermedad por coronavirus 2019. An Fac med. , 81(2):230-3. .

Blondel, V. D. (2008). Fast unfolding of communities in large networks. Journal of statistical mechanics: theory and experiments.

Dayter, D. (2015). Small stories and extended narratives on Twitter. Discourse, Context & Media, 19-26.

Derakhshan, C. W. (2017). Information Disorder: Toward an interdisciplinary framework for research and policymaking. Council of Europe .

Gupta, V. &. (2009). A survey of text mining techniques and applications. J, 1(1), . Journal of emerging technologies in web intelligence, 60-76.

Lancet. (17 de July de 2020). Editorial. Obtained from The Lancet Infectious Deseases: https://www.thelancet.com/journals/laninf/article/PIIS1473-3099(20)30565-X/fulltext

Ministry of Popular Power for Foreign Relations. (February 2020). Obtained from Ministry of Popular Power for Foreign Relations: http://mppre.gob.ve/wp-content/uploads/2020/02/2020-02-10_Bolet%C3%ADn_Vuelta_a_la_Patria.pdf

Ministry of Popular Power for Foreign Relations. (2020). Plan vuelta a la Patria. Obtained from Ministry of Popular Power for Foreign Relations: http://www.mppre.gob.ve/temas/vuelta-a-la-patria/

OPS. (2020). Entender la infodemia y la desinformación en la lucha contra la COVID-19. Obtained from https://www.paho.org/es/: https://www.paho.org/es/file/64245/download?token=vqDvd7jC

Tardáguila, C. (7 de November de 2019). Obtained from Poynter: https://www.poynter.org/fact-checking/2019/against-all-odds-fact-checking-is-flourishing-in-venezuela/

United Nations Development Programme. (2020). Responding to COVID-19 Information Pollution. New York: UNDP.

Zalmout, N. &. (2013). Multidimensional community detection in Twitter. 8th International Conference for Internet Technology and Secured Transactions (ICITST-2013) (págs. 83-88). IEEE.

narratives and information pollution on vulnerable groups during the covid-19 pandemic | 71 UNDP Country Team

UNDP in Panama

Manuel Rodríguez. Communications Communications Specialist of the UNDP Panama Country Office. Specialist in communication strategies. Co-writer of this report. Focal point for Panama.

UNDP in Venezuela

Daniel Barráez. Senior Economist, UNDP Venezuela Country Office. Leader of the project’s team of data scientists and drafter of this report. Focal point for Venezuela.

Ana María Chirinos-Leañez. Economist, UNDP Venezuela Country Office. Identification of narratives in Panama and Venezuela

Alejandro Labarca. Associate Statistician, UNDP Venezuela Country Office. Specialist in computational statistics.

Wilmer González. Specialist in machine learning and community identification.

Manuel Solorzano. Specialist in natural language processing and Python text mining.

Silvia Orozco Pabón. Communications Assistant, UNDP Venezuela Country Office. Agenda development with key actors, media monitoring and communications strategy coordinator in Venezuela.

Andreína Itriago. Media and opinion matrix analyst.

Nirma Hernández. Communications strategist with emphasis on working with communities.

UNDP Regional Centre for Latin America and the Caribbean

Emanuele Sapienza. Policy Advisor in the area of Governance at the UNDP Regional Centre for Latin America and the Caribbean based in Panama. Specialist in inclusive political processes.