How regional is the coverage in daily newspapers?
An Analysis and Visualization with Pandas and Tableau
About daily newspapers, the Austrian media landscape is divided into publications that describe themselves as regional media (newspapers of the federal provinces) and national media. In this article, I would like to examine whether this self-understanding is also reflected in the reporting. As a research method, I choose a data-based approach that collects and analyses the news articles published online in the period from 2 February 2021 to 3 May 2021.
The data
I have 217,595 contributions distributed among the following 15 publishers:
National media
- ORF / http://orf.at
- Österreich / http://oe24.at
- Kurier / http://kurier.at
- Die Presse / http://diepresse.com
- Kronen Zeitung / http://krone.at
- Der Standard / http://derstandard.at
- Heute / http://heute.at
- Neues Volksblatt / https://volksblatt.at
- Wiener Zeitung / http://wienerzeitung.at
Regional media
- Kleine Zeitung / http://www.kleinezeitung.at
- Salzburger Nachrichten / http://salzburg.com
- Vorarlberg Online / http://vol.at
- Tiroler Tageszeitung / http://tt.com
- Oberösterreichische Nachrichten/ http://nachrichten.at
- Niederösterreichische Nachrichten / http://noen.at
The articles are in the form of a csv file which, in addition to the article text and title, contains further information such as publication date, publisher, etc.
The analysis
The above chart shows on a map of Austria how many news articles in the regional newspapers dealt with the respective towns and cities. The color-coding indicates in which newspaper the article was published.
Regionality is measured by searching the articles for place names that appear in them. For each place found in the data, the article is noted with title, publication date, publisher, and the number of words. In addition, the postcode in which the place is located is saved for each article. I use this to determine the geocoding in the visualization software “Tableau”, as the place names directly are not suitable here. Tableau only knows the larger Austrian towns and not all the ones used here.
To be able to search for all Austrian municipalities, we use a list provided by Statistik Austria
That is 2,078 place names. Some place names that are common German words (e.g. “Bach” or “Berg”) have been removed from the list to avoid false hits.
The results for regional media
In the chart at the beginning of the article, the federal state newspapers Oberösterreichische Nachrichten, Niederösterreichische Nachrichten, Salzburger Nachrichten, Tiroler Tageszeitung, Kleine Zeitung and Vorarlberg Online are shown together. At first glance, it is noticeable that Oberösterreichische Nachrichten, Niederösterreichische Nachrichten, and Tiroler Tageszeitung have a strong connection to the respective federal province, whereas Kleine Zeitung is distributed over the entire federal territory. The Salzburger Nachrichten and Vorarlberg Online are difficult to identify in the presentation. If we look at them individually, we also see a strong concentration on the federal state in the case of Vorarlberg Online.
For the Salzburger Nachrichten, the orientation towards the federal state is not as clearly recognizable as the following figure suggests.
Results for national media
Let us now look at the national media. A joint presentation of these shows a distribution over the entire federal territory in each case.
If one looks at individual media alone, such as ORF, this impression is reinforced.
An interesting deviation here is the Volksblatt, which seems to have a lot of reference to Upper Austria.
Development over time — animation
If you want to see the development over time in the period under consideration, then it makes sense to look at an animation based on the maps. For this purpose, I have taken the publication date of the articles, summarised them into days, and compiled the individual graphics into an animation.
Here, too, the regional focus of the federal states’ newspapers is clearly evident over the entire period under consideration.
The implementation
In the following, I would like to show how and with which tools the analysis and visualization were carried out. The following Python script with the help of the package “Pandas” is used to search for the place names and the data is converted into the desired form for import into Tableau.
If we look at a histogram of the frequently occurring place names, it is noticeable that places with Bundesliga football teams often appear in the top ranks next to the large cities. This is understandable due to the extensive sports coverage.
In Tableau, we use the map display with the geocoding of the postcodes. The size of the circles indicates the number of articles per location, and the color marks the publisher. In the quick info for the rollover, the title of the article is also displayed.
I used Tableau’s animation functions to show the development over time.
Conclusion
I showed how to prepare the data with a few lines of “Pandas code” and how to easily create the corresponding graphics and animations with the standard functions of “Tableau”. Concerning regionality, clear differences between the publishers were very easy to recognize based on the visualizations.
More contributions to the analysis of news articles:
COVID19 Reporting:
Hidden information in news articles:
Coverage of political parties and elections:
To the stock analysis
