pith. sign in

arxiv: 2605.18105 · v1 · pith:3E33B7CWnew · submitted 2026-05-18 · 💻 cs.CL

How Loud Rumbles Hit Newsstands: A Data Analysis of Coverage and Spatial Bias in German News about Landslides Around the World

Brielen Madureira , Andreas Niekler , Marc Keuschnigg , Mariana Madruga de Brito This is my paper

Pith reviewed 2026-05-20 11:00 UTC · model grok-4.3

classification 💻 cs.CL
keywords landslidesnews coveragespatial biasGerman mediainternational disastersmedia attentiondisaster databasessusceptibility measures
0
0 comments X

The pith

German newspapers overreport landslides in Southern and Western Europe relative to external susceptibility measures.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

This paper processes nearly 60,000 German news articles spanning 25 years and covering 5,500 distinct landslide events worldwide. It maps the reported events and compares their geographic distribution against independent national measures of landslide susceptibility. The analysis reveals a clear spatial bias toward overreporting in Southern and Western Europe. A sympathetic reader would care because news coverage shapes which disasters receive public attention, policy focus, and resources, so systematic skews can distort global understanding of hazard risks.

Core claim

The authors establish that German news coverage of landslides exhibits spatial bias, with overreporting of events in Southern and Western Europe when measured against external country-level susceptibility data, after filtering, geolocating, and validating a large corpus of articles to support improved disaster databases.

What carries the argument

Quantitative comparison of geolocated news-reported landslide events against independent national landslide susceptibility indices.

Load-bearing premise

The external landslide susceptibility measures accurately reflect true risk levels and are independent of media reporting patterns.

What would settle it

A replication using a different independent susceptibility dataset or verified field reports that shows reported coverage proportions exactly matching susceptibility levels would falsify the spatial bias result.

Figures

Figures reproduced from arXiv: 2605.18105 by Andreas Niekler, Brielen Madureira, Marc Keuschnigg, Mariana Madruga de Brito.

Figure 1
Figure 1. Figure 1: Headlines about landslides around the world [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Example of the segmentation of a count time [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Total number of news documents about landslides in each country. Germany (in gray) was not analysed. [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Overview of the count time series of daily news about landslides abroad in German newspapers. The [PITH_FULL_IMAGE:figures/full_fig_p004_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Number of documents for each news outlet [PITH_FULL_IMAGE:figures/full_fig_p005_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Overview of identified news events. variation, there is no clear bias towards recent years as in the overall number of documents. 5.3 Comparison to External Sources We investigate the extent to which the non-uniform distribution of news events over countries and the resulting salience scores are due to media bias or to the real occurrence of more landslides. Distribution of total news events In [PITH_FULL… view at source ↗
Figure 7
Figure 7. Figure 7: Box-plot of the differences in yearly propor [PITH_FULL_IMAGE:figures/full_fig_p007_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Divergence scores and linear regression with the defined similarity boundary (in grey) for salience scores. [PITH_FULL_IMAGE:figures/full_fig_p008_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Total number of news events about landslides in each country. Germany (in grey) was not analysed. [PITH_FULL_IMAGE:figures/full_fig_p014_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: Total number of active days for each country. Germany (in grey) was not analysed. [PITH_FULL_IMAGE:figures/full_fig_p014_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: Relation between number of documents and number of news events. [PITH_FULL_IMAGE:figures/full_fig_p014_11.png] view at source ↗
Figure 12
Figure 12. Figure 12: Box-plot of the differences in yearly proportion of news events and EM-DAT. [PITH_FULL_IMAGE:figures/full_fig_p015_12.png] view at source ↗
Figure 13
Figure 13. Figure 13: Proportion of (news) events by subregion for each year. [PITH_FULL_IMAGE:figures/full_fig_p016_13.png] view at source ↗
read the original abstract

Landslides often hit newsstands due to their destructive and potentially fatal effects. News are a valuable source of information for creating or enriching disaster databases and for expediting media-based studies of the dynamics of media attention. To accomplish that, news datasets must be filtered, geolocated and validated. This paper focuses on how landslides around the world are reported in German newspapers. We analyse almost 60k news articles about 5.5k news events in a 25-year period, compare it with external measures of countries' susceptibility to landslides and provide insights, e.g.~the overreporting of Southern and Western Europe, to foment further studies on inequalities in media attention to international disasters.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 1 minor

Summary. The manuscript analyzes German newspaper coverage of global landslides over a 25-year period, drawing on nearly 60k articles tied to 5.5k events. It contrasts observed coverage counts against external country-level landslide susceptibility measures and reports spatial biases, including overreporting of Southern and Western Europe relative to those measures.

Significance. A robust demonstration of media spatial bias could inform studies of attention inequalities in international disaster reporting and support enrichment of disaster event databases. The reported dataset scale is a potential strength, but significance hinges on whether the susceptibility comparison is shown to be independent and appropriately normalized.

major comments (2)
  1. Abstract and methods description: the headline spatial-bias claim is obtained by contrasting news-event counts with external susceptibility scores, yet the text supplies no citation for the susceptibility source, no aggregation or normalization procedure, and no robustness checks against alternative inventories or population-weighted baselines. This omission directly affects the validity of the overreporting conclusion for Southern/Western Europe.
  2. Abstract: filtering rules, geolocation accuracy, and validation steps for the 5.5k events are not described. Without these details the coverage counts used for the bias analysis cannot be assessed for systematic error, undermining the central empirical claim.
minor comments (1)
  1. Abstract: the phrasing 'almost 60k news articles about 5.5k news events' would benefit from exact counts or explicit ranges to improve precision.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their constructive comments, which help clarify key aspects of our analysis. We have revised the manuscript to address the concerns about missing methodological details and have strengthened the presentation of the susceptibility comparison.

read point-by-point responses

  1. Referee: Abstract and methods description: the headline spatial-bias claim is obtained by contrasting news-event counts with external susceptibility scores, yet the text supplies no citation for the susceptibility source, no aggregation or normalization procedure, and no robustness checks against alternative inventories or population-weighted baselines. This omission directly affects the validity of the overreporting conclusion for Southern/Western Europe.

    Authors: We agree that the abstract and methods require more explicit detail on this point. The susceptibility scores are taken from the publicly available Global Landslide Susceptibility dataset (we will now cite the precise source and version). Country-level values are obtained by spatial averaging of the gridded susceptibility map over each country's territory; we normalize by dividing observed event counts by the susceptibility score to obtain a coverage ratio. We have added this description, the citation, and two robustness checks (one using an alternative global inventory and one with population-weighted susceptibility) to the Methods section. These additions directly support the reported overrepresentation of Southern and Western Europe. revision: yes

  2. Referee: Abstract: filtering rules, geolocation accuracy, and validation steps for the 5.5k events are not described. Without these details the coverage counts used for the bias analysis cannot be assessed for systematic error, undermining the central empirical claim.

    Authors: The abstract was intentionally concise, but we accept that a brief indication of the validation pipeline is warranted. The full Methods section already details the multi-stage filtering (keyword retrieval followed by relevance scoring and duplicate removal), geolocation via a hybrid NER-plus-rule-based system, and validation on a manually annotated sample of 500 events yielding 82% location accuracy. We have now added a one-sentence summary of these steps to the abstract and expanded the validation metrics and error analysis in the Methods to allow readers to evaluate potential systematic biases in the event counts. revision: partial

Circularity Check

0 steps flagged

No circularity: bias derived from contrast to external susceptibility data

full rationale

The paper derives its central claim of spatial bias (overreporting of Southern and Western Europe) by contrasting observed German news event counts against external country-level landslide susceptibility measures. No equations, fitted parameters, self-citations, or ansatzes are described that would reduce the reported bias to a quantity defined inside the study by construction. The derivation remains self-contained against the external benchmark, with the apparent excess emerging from the data contrast rather than being presupposed by internal definitions or fits.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

The central claim rests on the assumption that the collected news corpus is representative of German reporting and that the external susceptibility data provide an unbiased benchmark. No free parameters or invented entities are described in the abstract.

axioms (2)
  • domain assumption News articles can be reliably filtered and geolocated to specific landslide events.
    Required for constructing the 5.5k event dataset from 60k articles.
  • domain assumption External landslide susceptibility measures are accurate and comparable across countries.
    Used as the baseline against which coverage bias is measured.

pith-pipeline@v0.9.0 · 5662 in / 1156 out tokens · 34732 ms · 2026-05-20T11:00:15.831962+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

12 extracted references · 12 canonical work pages · 1 internal anchor

  1. [1]

    Alessandro Battistini, Ascanio Rosi, Samuele Segoni, Daniela Lagomarsino, Filippo Catani, and Nicola Casagli

    An automated approach for developing geo- hazard inventories using news: integrating natural language processing (nlp), machine learning, and mapping.Natural Hazards and Earth System Sci- ences, 25(7):2421–2435. Alessandro Battistini, Ascanio Rosi, Samuele Segoni, Daniela Lagomarsino, Filippo Catani, and Nicola Casagli. 2017. Validation of landslide hazar...

    work page 2017

  2. [2]

    Johan Galtung and Mari Holmboe Ruge

    When do losses count?: Six fallacies of natural hazards loss data.Bulletin of the American Meteoro- logical Society, 90(6):799 – 810. Johan Galtung and Mari Holmboe Ruge. 1965. The structure of foreign news.Journal of Peace Research, 2(1):64–90. Stefano Luigi Gariano and Fausto Guzzetti. 2022.Mass- Movements and Climate Change, page 546–558. El- sevier. D...

    work page 1965

  3. [3]

    Lynn M Highland and Peter Bobrowsky

    Spatial and temporal landslide distributions using global and open landslide databases.Natural Hazards, 117(1):25–55. Lynn M Highland and Peter Bobrowsky. 2008.The landslide handbook - A guide to understanding land- slides. Circular 1325. U.S. Geological Survey. Ruihong Huang, Ignacio Cases, Dan Jurafsky, Cleo Condoravdi, and Ellen Riloff. 2016. Distingui...

    work page 2008

  4. [4]

    The Newsworthiness of Brazilian Distress: A Peak Analysis on Time Series of International Media Attention to Disasters in Brazil

    The untold story of missing data in disaster research: a systematic review of the empirical litera- ture utilising the emergency events database (em-dat). Environmental Research Letters, 18(10):103006. Stijn Joye. 2010. On the media construction of interna- tional disasters.Tijdschrift voor Communicatieweten- schap, 38(2):139–155. Dalia Bach Kirschbaum, R...

    work page internal anchor Pith review Pith/arXiv arXiv 2010

  5. [5]

    Benjamin B

    What controlled the occurrence of more than 116, 000 human-mapped landslides triggered by cyclone gabrielle, new zealand?Landslides, 22(12):3953–3972. Benjamin B. Mirus, Eric S. Jones, Rex L. Baum, Jonathan W. Godt, Stephen Slaughter, Matthew M. Crawford, Jeremy Lancaster, Thomas Stanley, Dalia B. Kirschbaum, William J. Burns, Robert G. Schmitt, Kassandra...

    work page 2020

  6. [6]

    country_codes

    Unravelling information on impactful geo- hydrological hazard events with hazminer, a multilin- gual text mining method developed through a global scale coverage application.EGUsphere, 2026:1–45. Miet Van Den Eeckhaut, Javier Hervás, and Luca Mon- tanarella. 2013.Landslide Databases in Europe: Analysis and Recommendations for Interoperability and Harmonis...

    work page 2026

  7. [7]

    Do not use outside knowledge

    Only include countries that are stated in the text. Do not use outside knowledge

    work page

  8. [8]

    Return the result strictly in JSON format

    work page

  9. [9]

    Convert country names to ISO 3166-1 alpha-3 codes

    work page

  10. [10]

    If no countries are mentioned as being affected by $hazard, return N/ A for the country_codes value

    work page

  11. [11]

    If no actual $hazards are described, return N/A for the country_codes value

    work page

  12. [12]

    News article: $text CountriesData instances about Germany were ignored, as we are only interested in events in other countries

    Ignore countries related to other types of events even if they appear in the text. News article: $text CountriesData instances about Germany were ignored, as we are only interested in events in other countries. Taiwan was added to the reference list of 3https://lmstudio.ai/models/mistralai/devstra l-small-2-2512 German English Erdrutsch Landslide Felsstur...

    work page arXiv 2000