La reciente jornada de protestas en Colombia por parte del sector agrario, ha revestido rasgos destacables: Adhesión de múltiples sectores, respaldo de la opinión pública, acuerdos que implican revisar términos de un tratado de libre comercio con Estados Unidos y el peor momento de la presidencia de Juan Manuel Santos.
Estos hechos no pasan desapercibidos para un país donde históricamente la protesta social ha sido marginal y por el contrario ha sido víctima de la criminalización por parte del Estado. Los resultados pueden inducir -a los optimistas- a pensar que algo está cambiando y este fenómeno será sin duda, objeto de investigaciones que puedan ofrecer algunas respuestas.
Aunque son múltiples los actores involucrados, he querido realizar un ejercicio analítico sobre el papel desempeñado por los medios, en este caso prensa escrita, sobre la cobertura realizada durante los 19 días de duración del “paro agrario”. He seleccionado el material de portada de dos diarios de circulación nacional: “El Tiempo” y “El Espectador” y uno de circulación local “El Universal” de Cartagena. Para comprender el nivel de llegada, les incluyo un ranking nacional de acuerdo al Estudio General de Medios 2010 de los diarios mencionados:
La idea de seleccionar las portadas como objeto de estudio está radicada en gran parte en la teoría de Agenda- setting de Maxwell MacCombs y Donald Shaw, que sostiene que los medios de comunicación al darle relevancia o no a un tema, jerarquizan su percepción en la audiencia. Esta teoría ha otorgado por años un marco analítico para los medios y mejor aún, ha puesto de relieve la no neutralidad del acto de selección del editor.
Las portadas son la hojeada rápida, la scrolleada de la mañana y el tema de discusión para el momento incomodo en el ascensor. Conscientes o no, nos marcan una agenda y de allí su importancia. Con la ayuda de PageoneX, una herramienta open source para el análisis, codificación y visualización de portadas, decidí examinar detenidamente el comportamiento de estos diarios durante el periodo de protesta, diferenciando el área dedicada a la misma, de las otras temáticas y del contenido publicitario.
El resultado pueden verlo en el siguiente gráfico:
La conclusión que salta a simple vista es que el único diario de los tres que mantuvo un espacio constante en frecuencia y área para el tema fue “El Espectador”, mientras que el comportamiento de “El Tiempo” fue variable, “El Universal” por su parte se destaca por la poca cobertura realizada. Sin embargo y teniendo en cuenta que en el caso de “El Tiempo” fue imposible conseguir los datos para dos fechas como se observa en el área en blanco etiquetada s/d (sin datos), decidí añadir otro nivel al análisis, incorporando la lista de titulares de las noticias relacionadas al paro y resaltando en color, los términos con alta carga conceptual, que a mi juicio incurren en calificación del hecho.
Titulares de “El Tiempo”
17/8 16.000 policías, contra bloqueos
19/8 División en dos sectores claves del Paro
20/8 Incidentes en vías en la primera jornada de paro
21/8 Gobierno hará diálogos por separado con no violentos.
22/8 Boyacá, el más afectado por las protestas violentas
23/8 La protesta es con bloqueos, piedras y hasta desperdicio de leche
24/8 Incidentes en vía a la Calera por el paro
25/8 Paro golpea a tres regiones
26/8 No me van a enfrentar con los campesinos
27/8 Santos acepta negociar con campesinos de tres regiones
28/8 Paro: con oferta del gobierno empezó el dialogo
30/8 El vandalismo opaca la protesta en Bogotá
Marcha a favor de campesinos en la capital degenera en violencia y graves danos
31/8 Queda libre de bloqueos el corazón de la protesta
Hay expectativa en Boyacá por la reanudación de diálogos con campesinos
4/9 Protesta: más de 18000 indígenas marchan en Popayán
5/9 Protestas le pasan la cuenta de cobro a Santos.
Aunque el resaltado pueda señalarse como un ejercicio altamente subjetivo, la diferencia del tono es destacable si se compara con los titulares del Espectador, que destaca el carácter pacífico y la importancia de la protesta.
Titulares de “El Espectador”
19/8 Al paro nacional del agro se sumaron los camioneros y los mineros. Cafeteros del Huila no saldrán a protestar.
20/8 Diálogo, pero sin bloqueos
21/8 Protesta pacifica
23/8 En el asfalto
24/8 Paro llego a Bogotá
25/8 ¿Y si bajamos la guardia?
26/8 El costo diario del paro
27/8 Crisis agraria
28/8 Se crece el paro
29/8 A la espera
30/8 Así no
31/8 Después de la tormenta
1/9 ¿Por qué pasó?
2/9 Sembrando acuerdos
3/9 Renuncia protocolaria
Como ultimo nivel de análisis, propongo las imágenes de las portadas que acompañan las notas y que pueden verse en la versión interactiva. “El Tiempo” muestra 2 que generan rechazo: un camión en llamas bloqueando una vía y los campesinos arrojando leche a la vía. Una tercera de un hombre solo acomodando piedras, con la fuerza pública en el fondo. También se puede ver una imagen de las fuerzas especiales “cubriéndose” de los “ataques” de los manifestantes . Las imágenes positivas presentan la marcha de apoyo de los pueblos originarios marchando en Popayán y la del acuerdo, donde entre todos mueven un tronco que obstruía la vía. Por su parte el “El Espectador” se concentra en mostrar imágenes de los marchantes, incluyendo los mensajes de sus pancartas, y algunas situaciones de desventajas frente a la fuerza pública.
Este ejercicio es factible de profundizarse y está disponible para quien quiera hacerlo, sin embargo el balance está lejos de la realidad y revela ostensiblemente la importancia de las redes sociales en la protesta social: paralelamente a las publicaciones analizadas, circularon numerosos vídeos que reflejaban los graves abusos de la fuerza pública contra los manifestantes. El grado de violencia perpetrado, los convierten en material sensible que he decidido no incluir en este post, pero que pueden ser encontrados en Youtube y Twitter bajo las etiquetas “paro agrario” #paro agrario #esmad.
La Comisión nacional de derechos humanos de la Mesa Nacional Agropecuaria y Popular de Interlocución y Acuerdos emitió un comunicado informando que durante el paro se presentaron:
–660 casos de violaciones a los derechos humanos individuales y colectivos en todo el país
–262 detenciones arbitrarias
–485 personas heridas
–21 personas heridas con arma de fuego
–52 casos de hostigamientos y amenazas contra los manifestantes y líderes sociales
–4 personas desaparecidas
–51 casos de ataques indiscriminados a la población civil en diferentes departamentos del país
– la pérdida de 12 vidas a causa de impacto de arma de fuego.
¿Qué dijeron los periódicos colombianos de estas víctimas?
Mientras las negociaciones, los costos económicos y las consecuencias políticas del Paro superen en importancia las graves violaciones de derechos humanos, nuestra capacidad de construir la paz en Colombia, seguirá viciada.
Prioridades en desorden, como es usual.
 Agradeceré el dato si alguien puede facilitármelo.
 Me permití el uso de las comillas, por las razones que expongo al final de esta entrada.
Em 2012, começou a circular na internet uma imagem comparando como a mídia brasileira abordava uma manifestação no Brasil e outra na França. Os exemplos usados na imagem eram de notícias veiculadas em 2009:
No meio de tantos protestos no Brasil em junho de 2013, surge a dúvida de como esse tema foi abordado nas capas de jornais tradicionais. Estariam os jornais chamando essas pessoas de manifestantes, ou de vândalos?
Antes de responder a essa questão, vale a pena rever o que aconteceu no país em junho. Contra o aumento de 20 centavos no transporte público, manifestações ocorreram no início do ano em diversas cidades do Brasil, porém com pouca visibilidade na grande mídia.
Em São Paulo, através de uma mobilização organizada pelo Movimento Passe Livre os protestos nos dias 06, 07, 11 e 13 de junho contaram com 2.000, 5.000, 10.000 e 10.000 manifestantes respectivamente. Imagens e vídeos dessa última passeata foram massivamente divulgados pela internet, mostrando excessiva violência policial contra jornalistas, manifestantes e quaisquer pessoas que estivessem sob a mira deles. Isso foi o estopim para o surgimento de manifestações em outras cidades e para o aumento do apoio dos cidadãos aos protestos. Ao todo foram mais de 600 protestos no Brasil, onde os manifestantes citaram mais de 20 razões para estar nas ruas, segundo pesquisa do IBOPE.
A partir da ênfase que a grande mídia nacional deu a esses acontecimentos foi feita uma pesquisa, que consistiu na leitura e mapeamento da primeira capa de 6 jornais tradicionais. 3 de circulação nacional – O Estadão, A Folha de São Paulo e O Globo, e 3 jornais menores com distribuição local – Jornal Estado de Minas, O Liberal e O Correio Braziliense. 4 categorias foram criadas para classificar o foco de cada uma das matérias:
a) Violência (vermelho)
b) Protesto (verde)
c) Copa das Confederações (azul)
d) Ações Políticas de maior interesse do cidadão (amarelo)
(para uma descrição mais detalhada dos critérios que definem cada uma das categorias, veja o final do texto)
A partir dessa classificação, um gráfico foi gerado mostrando o assunto enfatizado pelas capas (estavam falando das manifestações ou não) e o tom (estavam enfatizando a violência ou não). Copa das Confederações foi incluída como categoria, pois ocorreu entre os dias 15 e 30 de junho, justamente no período de grande mobilização brasileira.
O gráfico revela ao menos 5 pontos interessantes:
1) Até o dia 13 de junho, o relato dos protestos focava na violência (vermelho) e no embate entre manifestantes e polícia. Note que a partir do dia 14 (dia seguinte ao protesto onde jornalistas foram feridos e presos em São Paulo), a mídia abre espaço para uma cobertura mais ampla (englobando as razões que motivavam os protestos) e sem focar somente na violência.
2) No segundo dia após a grande marcha do dia 20 de junho, onde mais de 1milhão e 700 mil pessoas marcharam por todo Brasil, surge uma mudança relevante nas matérias de primeira capa. A partir do dia 22 os jornais focam na discussão de reformas políticas (amarelo), trazendo temas cobrados nos protestos.
3) A abordagem das matérias com foco no protesto (e não na violência) aumenta a partir do dia de maior tweets em 17 de agosto. Logo após esse pico, há um decrescimento nos tweets que coincide com o crescimento do foco na violência nas capas. As hashtags estão listadas nesse documento.
4) O dia de maior manifestantes na rua não é o de maior tweets.
5) Por fim, há duas observações que não são possíveis visualizar através do gráfico gerado.
Primeiro, os jornais divulgaram nas primeiras páginas pesquisas sobre hábitos e vícios de jovens e estudantes. No dia 24 de junho, em um protesto próximo ao complexo da Maré (Rio de Janeiro), um grupo não identificado realizou roubos na área da manifestação. A polícia realizou uma busca aos assaltantes, onde 1 sargento e 9 pessoas foram mortas dentro da comunidade (3 não tinham antecedentes criminais). No dia seguinte, nenhuma capa relatou o ocorrido. Apenas dois dias depois, apareceu uma foto com o corpo de um dos mortos na capa de O Estado de São Paulo. O Globo, que é a mídia do Rio de Janeiro, não mencionou essas mortes na primeira capa em nenhum dia do mês.
E segundo, um episódio relevante de violência policial após um protesto foi ignorado pelos jornais analisados. Entre os dias 18 e 24 de junho nas capas dos jornais O Globo, Correio Braziliense, O Estado de São Paulo e o Estado de Minas, há chamadas para matérias informando que, entre estudantes e adolescentes:
houve crescimento no uso de drogas,
76% já ingeriram álcool,
se trata de uma geração sem ocupação (sem emprego, estudo ou atividade),
há uma nova droga usada na balada.
Segundo o levantamento, as primeiras capas de meses anteriores nunca tinham dado destaque para tais pesquisas.
Esses são apenas alguns dos muitos pontos que podem ser analisados a partir do gráfico acima. A conclusão é que a mídia brasileira têm oscilado na ênfase que dá aos protestos, sem nunca deixar de guardar o devido espaço para falar do futebol.
P.S. Esse é apenas o primeiro texto de um projeto maior de análise de dados.
O próximo passo é comparar a abordagem da mídia tradicional com a mídia social.
Entre em contato.
Critérios utilizados para as categorías criadas:
a) Violência (vermelho):
imagens com carros ou ônibus ou objetos pegando fogo
imagens de policiais fazendo barreira, atirando ou jogando spray de pimenta
descrição da matéria com uso das palavras: vândalo, depredação, medo, quebra-quebra, quebradeira, violência, saque, criminoso
descrição da matéria com apoio à ação violenta da polícia
matéria abordando o medo de participação aos protestos
b) Protesto (verde):
foto da passeata
explicação sobre a motivação do protesto
c) Copa das Confederações (azul):
fotos ou texto sobre os jogos da Copa das Cofederações
“Protesters set fire to more than 300 cars in France”
“Vandals set fire to a train in Rio”
How is this topic addressed on the front page of traditional newspapers, in the midst of so many protests in Brazil during June? Were the newspapers calling these people protesters or vandals?
Before answering this question, it is worth reviewing what happened in Brazil that month. The 20 cent hike in the Real (less than $0.10) for the transportation fare triggered demonstrations in the main cities in Brazil. Little coverage was observed in the mainstream media. In São Paulo, demonstrations were mobilized by the Movimento Passe Livre on June 6th, 7th, 11th and 13th, with 2,000, 5,000, 10,000 and 10,000 protesters, respectively. Images and videos of the demonstrations were massively disseminated over the Internet, showing excessive police brutality against journalists, demonstrators and anyone under police target. This flood of videos and images was the catalyst for the demonstrations in other cities, and increased public support for the protests. At the end of the month there were over 600 protests. Protesters have cited more than 20 reasons to be on the streets, according to research by Brazilian statistic research institute called IBOPE. Some of these reasons were used to classified the category of the research below.
The research consisted of reading and mapping the front pages of 6 traditional newspapers; Three newspapers with national circulation:
O Estadão, São Paulo
A Folha de São Paulo, São Paulo
O Globo, Rio de Janeiro
Three newspapers with local distribution:
Jornal Estado de Minas, Minas Gerais
O Liberal, Para
O Correio Braziliense, Brasilia (the capital of Brazil)
Four categories were created in order to classify the focus of each article:
Confederation Cup – soccer tournament that acts as a rehearsal for the World Cup (blue)
Political issues raised by the protesters (yellow)
(For a detailed description of each category, see end of text)
Based on this classification, a chart was generated showing the emphasis of the front pages (whether they were referring to the demonstrations or not) and tone (emphasizing violence or not).
The Confederations Cup was included as a category because it was happening at the same time (June 15th – 30th). The coverage referred to the tweets of the protesters on the streets the day before.
The graph reveals at least 5 interesting points: 1) Up to June 13th, the major focus of the coverage was violence (red) and the clash between protesters and police. On the 13th, several journalists were injured and arrested in São Paulo. The day after, the media seemed to make room for wider coverage (covering the reasons that motivated the protests), without focusing only on violence.
2) After the big march on June 20th, the protests dominated the newspapers’ front pages. The focus on violence was much more prevalent than the focus on the protest itself (green). By June 22nd, a significant change takes place giving us a discussion about policy reforms (yellow) rather than the protests. The president of Brazil, Dilma Rousseff, proposed a wide range of actions to begin reforming Brazil’s political system.
3) The front page focus on the protests grows in direct correlation to the citizens participation on Twitter for August 17th (hashtags used in this graph are here). After these 2 days, the tweets decrease and the focus on violence in the coverage increased.
4) The day with more protesters on the streets was not the day with more tweets about it.
5) Finally, there are two observations that are not covered in the graphic. First, the newspapers began to report about habits and vices of youth and students. Second, a relevant episode of police violence occurring after a protest was ignored by the newspapers.
Between June 18th and 24th, on the front pages of 4 newspapers, there were articles about students and teens pertaining to:
increase in drug use,
76% had drunk alcohol,
a lazy generation: no jobs or studies,
new drug in dance club.
Analysis on previous newspapers’ covers show that there was no focus on teenagers lifestyle previous to the protests.
On June 24th, there was a protest close to a poor community in Rio de Janeiro where an unidentified group robbed protesters and people around the area of the demonstration. The police then began to look for the assailants, and by the end of the “hunt”, 1 Sergeant and 9 people were killed. All of the people killed were residents in this poor community and three had no criminal record. The next day, this story did not make any of the front pages! Just two days later, a photo related to the deaths appeared on the cover of O Estado de São Paulo. O Globo, which is the media of Rio de Janeiro, did not mention the deaths in the front page at all.
These are just some of the many points that can be analyzed from the graph above. The conclusion is that the Brazilian media have changed their emphasis on the protests, always keeping considerable space for soccer coverage.
Note: This is only the first article of a larger project of data analysis. My next step is to compare the approaches of traditional media versus the social media.
Criteria used for the categories created: a) Violence (red):
Images with cars buses or objects on fire
Pictures of police barrier, police shooting or pepper spraying
Description of the article headline using the words: vandal, vandalism, fear, smashing, crash, violence, looting, criminal
Description of the article headline supporting violent police action
Text about the fear of participating in the protests
b) Protest (green):
Photo of the marches
Explanation of the motivation of the protest
Opinion about the protest without mention the words: vandal, vandalism, fear, smashing, crash, violence, looting, criminal
Buy two copies of the same edition of one newspaper. You need two copies to be able to display both sides of every page. We used the exterior side of the papers from one copy, and the interior from the other copy.
To be cautious, we marked (draw a thin line) in the side of the paper that we were not going to use, to avoid having a piece of paper and not do not know which side is the one to use.
Cut and separate Ads and News.
Once you have the two piles with Ads and News, you have to make the bar charts. Keep’em straight and make them have the same width. To make the puzzle easier we put all the full (uncut) size pages together at the bottom of the bars.
The front page images in the display view are displayed differently if they have been coded or not. It makes easier for the user to know intuitively which images have been coded (more transparent) with the ones that hasn’t (more opaque).
I’ve been testing with different opacities: make both uncoded and coded images lose opacity, and color codes to lose transparency. Related commit.
PageOneX, as it is now, doesn’t support online newspaper front pages, but I decided to give it a try using the recently discovered pastpages.org archive. I used it to track the coverage of he #occupygezi protests in Turkey by the BBC.
This article will be part of the forthcoming book Accountability Technologies: Tools for Asking Hard Questions edited by Dietmar Offenhuber and Katja Schechtner. It is based on previous published texts about PageOneX. It wants to be a short summary of the possibilities of the tool that has just been released: PageOneX.com There is an article in preparation with Sasha Costanza-Chock that will provide a review of scholarship in this area (print newspaper coverage) and will describe our approach in-depth to the topic.
Newspaper Front Page Analysis: How Do They Tell the Story?
Social Networking Sites (SNS), especially Twitter and Facebook, have played an important role in the current global social movement wave, from the Arab Spring to the Occupy movement in the United States. Social movement actors use SNS, among other tools, to document, promote, support and, in some cases, to coordinate movement activity. SNS have also become key sources of information for observers, both supporters, detractors, and non-participants, as well as for professional journalists working within print, broadcast, and online media, whose reporting strategies have, in some cases, shifted towards curating and incorporating content first circulated via SNS. In the context of increasingly transnationalized media firms and cross-platform convergence, information rapidly flows back and forth between SNS and mass media. Print newspaper coverage also has an impact far beyond paid subscribers or readers, since it greatly influences broadcast (television and radio) news agendas; newspaper coverage also plays an important agenda-setting role for blogs and social media. The complexity of the converged media ecology requires new approaches to newspaper content analysis, which has long been used by social movement scholars as a proxy for social movement activity. This essay describes an innovative approach to the analysis and visualization of front page newspaper coverage.
2. Background: How Did I Start Coding Front Pages?
On Sunday May 15th, 2011, one week before the municipal elections that were taking place all around the country, there was a grassroots-organized march in many cities in Spain under the motto “Real Democracy Now” (Democracia Real Ya[ref]Democracia Real Ya was not only the motto, but also the citizen platform to deepen in democratic values http://democraciarealya.es[/ref]). By then, I was living in Boston and closely following all these events through the Internet.
Tens of thousands of people throughout Spain marched on the streets during what later would be known as #15M, the #SpanishRevolution or the ‘indignados’ movement. The movement was inspired by the Arab Spring and the Icelandic Revolution and was preceded by previous social mobilizations in Spain. The country was in the midst of a huge economic crisis and suffering from austerity measures that were dismantling the welfare state. The crisis was related to the world economic recession and to the bursting of the real estate bubble. In the short term, this new mobilization was related to #nolesvotes[ref]#NoLesVotes (Do not vote for them) was the hashtag used in Twitter to promote this cyber movement. First used against the “Ley Sinde” law (Ley de Economía Sostenible [Sustainable Economy Law], March 2011), which opened the possibility of Internet censorship. That led to a second stage of the online movement that urged a change in the electoral system to fight bipartisanship, “Do not vote for them” was their motto, referring to the parties that approved the law. “Les” also meant “Ley de Economía Sostenible.”[/ref], a mobilization launched and forged on the Internet against Internet censorship that a new law was enforcing. In the years before, there had also been other mobilizations similar to the Democracia Real Ya platform involving demonstrations on the streets throughout the country, among others: Juventud sin futuro[ref]Started in Madrid in February 2011, “Youth without a future” is a citizens’ platform to address the problems of the precarious situation of the youth during the economic crisis juventudsinfuturo.net.[/ref], to protest the precarious situation of the youth; Por una vivienda digna[ref]Started in Madrid in 2003 during that year’s real estate bubble, “Right to housing” is a citizens’ platform to address housing problems viviendadigna.org.[/ref], a citizens’ platform created during the real estate bubble crisis to address housing problems.
Democracia Real Ya was a grassroots movement that received no supported from any party or union. It was a decentralized organization based on the Internet that opted for a multi-site demonstration, instead of the traditional march in the capital (Madrid). No flags were used and the “ghosts” of the two Spains from the Civil War; today’s two major political parties were also absent. Despite the movement’s success, the mainstream media did not pay much attention: It was just another demonstration in an electoral period.
That Sunday night after the march, around 40 people decided to stay in Puerta del Sol, the central square of Madrid. They wanted to keep on with the mobilization, and managed to convince the police to let them stay in the square during the night. After their first general assembly, they started organizing the first working groups. On the next day, many more people came to show support, and around 300 stayed on to sleep in the square. At 5 a.m. the police evicted them, an act that proved to be the tipping point of the movement. On Tuesday evening, more than 10,000 people crowded into the Puerta del Sol to protest the eviction. By that time, I was absorbed with the #acampadasol hashtag in Twitter, the 24-hour live streaming from soltv.tv, and trying to follow every bit of information about what was happening in my hometown of Madrid.
Some major newspapers were covering the events, but people in social media, mainly Facebook and Twitter, argued that they were not getting enough coverage. On Wednesday, May 18th, the electoral board in Madrid denied the right to march or gather in the square, Junta Electoral Provincial de Madrid (2011), and that made the protests even more successful than the previous days: Puerta del Sol Square was full again. I tried to visualize the subsequent reactions of the mainstream media to this new event and tweeted my first newspaper front page data visualization.
On Thursday morning, May 19th, media attention was completely centered on the occupied squares: The protests started filling all the front pages only after the fourth night at Puerta del Sol, after the “tent cities” had already spread all around the country. The front page data visualization became popular after appearing on Menéame, a Spanish news aggregator site, as it showed the very moment when the protest entered the mass media loop in an easy and comprehensible way. Indeed, it was used by bloggers to reflect on how the media had waited too long to cover the protests. Some journalists also used it to think about how the media should cover these kinds of protests. The following Sunday, one week after the mobilization had started, the elections took place, but the camps and the spirit of change from the 15M movement remained.
The 15M movement was now live on the streets and ready to grow and take infinite shapes and ways of protests. My front page analysis was also making its first steps, and I was ready to start diving into it and testing its possibilities.
How Are Social and Mass Media Related?
Twitter is the tool researchers most frequently use to study information flows that Social Networking Sites enable, because these streams of information are open, allowing anyone to study their data. Facebook is arguably a more important site for spreading news, based on its far higher penetration rate, but as a closed network, the study of the data has not been as easy as with Twitter. We have used front page newspaper coverage analysis to compare newspaper coverage of recent mass protests with social media attention, in particular, in Twitter.
For example, we have compared the use of popular hashtags, the way message “tweets” are tagged in Twitter, in the 15M movement. The most popular hashtags were #acampadasol, #15M, and #spanishrevolution. We have compared the total number of tweets using these hashtags with the front pages of 7 major newspapers in Spain. These hashtags had been selected among those most used at the time regarding this movement according to research by the Complex Systems and Networks Group at the University of Zaragoza (BIFI). The Twitter data came from Trendistic, an online Twitter data visualization provider.
A similar data analysis has been made with the Occupy movement with US newspapers. In this case, the chart above displays both the percentage of surface dedicated to the Occupy movement (red) and the number of tweets from the various related hashtags like: #ows, #occupywallstreet, #occupyLA, #occupyOakland or #occupyBoston.
Tweets are instant messages that cover events almost in real time. On the contrary, front pages cover the news of yesterday. Therefore, it is normal to see how different media react to the same event: Twitter coverage peaks the day that it occurs, and press coverage peaks the day after. Following this idea, it is more interesting to explore when Twitter and the press differ: When enormous tweet traffic about an event or action is not followed by that big coverage, or when a week-long, sustained conversation in Twitter around a hashtag doesn’t find its way into the print.
3. The Front Page Data Visualization: Building a Tool
The first data visualization started as “manual” process: I downloaded the images one by one from kiosko.net, a website providing newspaper front pages from all around the world, and built the array of images with Inkscape, a vector graphics program. Later on, after the good acceptance of the visualization, I started building a script to allow others to replicate this kind of analysis. Around a year later I started PageOneX, a free software tool (code available in github), designed to aid the coding, analysis, and visualization of front page newspaper coverage of major stories and media events.
PageOneX constitutes a straightforward way of interpreting front pages, a very specific piece of the media ecosystem that has direct influence on radio and TV broadcasts. In the past, researchers trying to analyze front pages needed to obtain copies of newspapers, measure column-inches by hand (with a physical ruler), and manually input measurements into a spreadsheet or database. This laborious and time-consuming process would be followed by calculation, analysis, and sometimes data visualization. Some of these steps can now be automated, while others can be dramatically simplified.
The process described here is designed to facilitate the relatively rapid creation of “small multiple” data visualizations of front page newspaper coverage. “Small multiple,” a term popularized by Edward Tufte, is a kind of data visualization that allows the visual comparison of multiple series of data: Series are displayed separately and placed next to each other. In the case of PageOneX, the displayed data include small images of the newspaper front pages themselves, as well as the surface area on the newspapers’ front pages dedicated to a specific news story (a “thread”) that runs chronologically along the x-axis (time). Coders manually select the portion of front pages dedicated to that thread. The selected areas are then displayed as shaded or colored transparencies on top of their original front pages. When viewed at sufficiently close range (zoomed in to a story), the text of the original stories that have been coded can still be easily read. However, this form of graphical information display is most compelling on a larger scale: Small thumbnails of the newspaper front pages are arrayed chronologically from left to right, and the viewer is then easily able to see at a glance the evolution of the coverage – or lack of coverage – of a particular news thread over time, as well as to compare different newspapers.
The visual approach allows the reader to have a sense of the coverage and, at the same time, have a look at the very raw data of the analysis, which are the components of the article, whether they be images, headlines or text. These highlighted areas could also be quantified to produce a bar chart. To make the measurements of the areas of news coverage comparable among different newspapers sizes, we use the percentage of used surface, and not the real area in mm².
The basic visualization of this type of graphic displays both the data itself (front pages and highlighted surfaces) as well as the measurement (the percentage of surface area) in a bar or line chart. Bar charts give a more accurate view of the surface devoted to the story, whereas line charts (see below) appear as a more continuous timeline, although they run the risk of suggesting a visual interpolation of discrete data. In other words, newspaper front pages provide one surface area datum per day, so a line connecting 2 surface area data points (from two consecutive days) produces a false sense of continuity. Line charts are useful, however, to compare more than 2 news threads over a long period of time, like the Arab Spring coverage (Egypt, Syria and Libya). When different topics are displayed one beside the other, it is a good way to study the battles for attention in the media; how related news “fight” for a position on page one, like in the Arab Spring coverage (Egypt, Syria, Libya and others) in Spanish newspapers.
We can see how the media space is usually focused on one “revolution” at a time, and that there is not much juxtaposition of news about this topic.
We observed a big drop in the Arab Spring coverage on April 13th, because the Japanese tsunami was breaking news: Note how the green area (Japanese tsunami) “hides” all the other news related to the Arab Spring for a week. It is not the case that “nothing” related to the Arab Spring happened during that week, but rather that the media decided where to pay attention. News on front pages “fights” for space and its areas are related to one another.
Comparing Asynchronous News Coverage in Time
The relative measurement of the surfaces lets us not only compare different news from different newspaper sizes, but also from asynchronous timelines. This method would allow the measurement of the news cycle: enabling us to understand how long news lasts on the front pages and how similar events attract the attention of the media.
In the comparison of the surface coverage dedicated to the Haiti earthquake (day 1 = January 12th, 2010) and the Japanese tsunami (day 1 = March 11th, 2011) in 4 Spanish newspapers, we could measure and compare the length of the coverage. Remarkable is the rapid drop of the Japanese crisis coverage after day 8 (March 18th, 2011), caused by the UN’s authorization allowing the international community to enter the war in Libya.
Semantic Analysis of News Content
It is also possible to make an automatic estimation of the article’s content based on semantic technologies. This process would benefit from the automatic transcription of scanned newspaper front pages through optical character recognition (OCR). The graphic displays the number of times that certain words have appeared during the fall of Mubarak and Ali. Grouped under “democratic” and “dictatorial” groups, January and February 2011.
As a case study, we used the fall of Mubarak and Ben Ali to compare how newspapers framed their last government/regime days. We analyzed the number of times that certain words appeared on the front page, and grouped them into two pools:
The aim of this study was to measure how newspapers are shifting their vocabulary along time, depending on their support / alignment with the issue they are covering. In Mubarak’s fall, we can see how newspapers shifted their frame during the crisis from a predominant use of democratic words (president, government) to more dictatorial ones (dictatorship, dictator, regime). In the case of Ben Ali, more “dictatorial” types of words were employed to describe his last days in power.
Content and Frame Analysis
We have seen how articles can be categorized, then colored and quantified, by multiple taxonomies related to their content: a particular thread or categorized by country. To understand how newspapers are addressing one topic would be not enough to understand how much they are covering it. We will need a framing analysis of their coverage: Are they positive/neutral/negative about it? Are they supportive? How biased is the information they provide?
We used this approach to analyze how different Spanish newspapers were covering corruption cases. The fast-growing list of corruption investigations in Spain contained almost every institution in the country. Everyday, new information and a new cases were being unveiled, provoking an increasing sense of indignation. It was also true that the levels of indignation had reached an apex, that they caused either incredulity or insensitivity.
The hypothesis and common ground knowledge is that every Spanish newspaper is aligned with a political party. Newspapers follow the corruption cases of the “opponent” parties. Measuring front page coverage seemed to be a good proxy to analyze their political agendas and to verify the hypothesis. I selected the most important Spanish newspapers and highlighted the stories related to the different corruption cases during a one-month period.
In January 2013, the El Mundo and El País newspapers reported that the ex-treasurer of the ruling party (Partido Popular, PP), Luis Bárcenas, had a $29 million Swiss account, and then published excerpts from the party’s parallel financial account. Both scoops impelled all the other newspapers to talk about the same PP corruption case, although other corruption cases affecting the main parties were also on the media menu. We coded the corruption cases from the major three parties: PP (conservative), blue; PSOE (center-left social democratic), red; CiU (conservative Catalan nationalist), orange; the royal house, green; other cases, purple.
Both newspapers, El Mundo and El País, are in opposite political spectrums, but made a common effort in this case to cover the Bárcenas story affecting the conservative party (PP) in the government.
El País centered its coverage on the PP (blue), leaving little space to the CiU (orange) or the monarchy (green), and none to the PSOE (red), following the expected behavior of a party related to the PSOE and the center-left. Meanwhile, El Mundo had a colorful timeline illustrating all the major corruption cases. Since the newspaper tends to defend the PP and attack the PSOE, it represented a change in their agenda.
However, some things did not fit into the alignment of mass media with certain major political parties that we could have expected. All of the newspapers were talking about the PP corruption case, but not in the same way. To understand how they were framing the story, we needed to add a new layer to the visualization: The issue of whether newspapers were framing corruption neutrally or negatively (plain areas) or defending and giving a positive view (striped areas).
A conservative newspaper like ABC (2nd row) dedicated their entire front page to the PP corruption case, but after the first news proclaimed that “PP must tell the truth about the Bárcenas case,” they started defending the actions the PP was taking. As expected, other conservative newspapers (La Razón, La Gaceta, but not El Mundo, which had published the scoop) took a defensive position (Jan. 19-20). It is also worth noting that once El País released the excerpts of the parallel financial account (Jan. 31), La Gaceta, a far-right newspaper, stopped defending the PP.
I published a blog post about this case study that garnered a lot of attention in the social media sphere (Twitter). It is worth mentioning that El Mundo published a piece on this research: “El Mundo, the newspaper that covers all corruption cases” in a four-column headline[ref]“A study by the prestigious MIT in the US shows that this newspaper stands out in the Spanish press for investigating the political parties without exception”. . [/ref]. They were interested in showing how the data visualization supported their unbiased work, and redrew the graphic to show El Mundo in first place.
In Spain, mass media are usually blamed for maintaining and supporting the bi-partisan system[ref]PP-PSOE, both parties are the ones entangled in more corruption cases; check the Corruptódromo map[/ref]. This time, however, some newspapers were playing a fundamental role in unveiling corruption cases, but most of them only kept reporting about certain corruption scandals. Only a well-informed citizen who escapes the biased point of view of his/her reference newspaper can get the full picture of what is happening and react accordingly.
The method for analyzing newspaper front pages that we have described here, as well as the PageOneX.com platform, provide a straightforward way to analyze and visualize news attention to specific stories over time, across or between newspapers. In the future, we imagine either extending this tool or connecting it to others focusing on other media platforms, such as TV[ref] Archive.org is making the TV searchable by recording TV images and captions in a public archive: http://archive.org/details/tv .[/ref] and radio broadcasts, or to the online versions of newspapers and social media.
Content analysis, and newspaper content analysis by amount of coverage, has had a long history in the field of communication studies. New tools are enabling novel approaches to this kind of analysis. Some aspects of newspaper content analysis can be automated, some can be distributed (crowdsourced), some can be done much more quickly by human coders using sophisticated new interfaces. The data produced by such analysis can also be visualized in new ways, and compared against data from other new sources of information flow such as social media. This essay has explained the approach we have taken with PageOneX, demonstrated its application to several specific stories, and suggested areas for further work. We hope that PageOneX provides a valuable example to anyone interested in monitoring the media: communication scholars, social movement activists, or advocacy organizations.
We invite anyone interested to contact us and to participate in the future development of the platform.
I finally installed the PageOneX real size project with stories related to guns, wars and terrorism. Come visit the display this week to the second floor of building E14 at MIT Media Lab. Thanks all of you that helped in this process: Adrienne, Luis, Álvaro, Nacho and Matjaz!
What follows is a photographic guide to the PageOneX real size installation.
After the post Guns vs Sports vs Wars I wanted to test news by location: local, for news related to the city or region where the newspaper is published; national for US news; and continent (as Ethan suggested). This visualizations fits pretty well with the previous one, as it fills the gaps of the more locally oriented newspapers like The Boston Globe and LA Times.
I do not think I will use this approach in the final PageOneX real size version, but it is interesting to view the same newspapers coded by different ways. I can imagine a par of special glasses that would allow the viewer shift from one visualization to another.
Note: I have only coded the main news (only the big ones).