Analyzing Patterns and Influence of Advertising in Print Newspapers

N. Harsha Vardhan, Ponnurangam Kumaraguru, Kiran GarimellaarXiv 2025 · DOI · arXiv

Introduction

While digital media dominates Western research, print newspapers remain a juggernaut in India, which boasts the world's second-largest newspaper market with over 110 million daily copies. However, analyzing print media at scale is notoriously difficult. Unlike web-based news, print archives are often locked behind proprietary formats or expensive databases, making cross-paper comparisons nearly impossible. This study addresses these challenges by developing an automated computer vision pipeline to analyze over 12,000 editions of major Indian newspapers, uncovering a significant correlation between corporate advertising spend and favorable news coverage.

Automated Analysis Pipeline

To transform raw page images into structured data, the researchers built a sophisticated multi-stage pipeline. The process begins with Segmentation, using a fine-tuned YOLOv8s object detection model to distinguish between articles and advertisements.

The automated pipeline for newspaper segmentation, OCR, and translation.

Following segmentation, the system employs a "best-tool-for-the-job" approach for Optical Character Recognition (OCR). Tesseract handles structured English articles, PaddleOCR manages the chaotic layouts of English ads, and Surya OCR is used for Indic languages like Hindi and Telugu. Non-English text is then translated via IndicTrans2. Finally, a RoBERTa-base model categorizes content into thematic topics such as Business, Politics, and Tech, while panel regression models analyze the relationship between ad spend and editorial sentiment.

Data Scale and Scope

The study utilized a massive, newly assembled dataset from five major Indian newspapers: The Times of India, Hindustan Times, The Telegraph (English), Dainik Bhaskar (Hindi), and Sakshi (Telugu). The scale of the data is unprecedented for this type of research, encompassing 12,358 editions containing approximately 800,000 articles and 465,000 advertisements from 2018 through mid-2024.

Strategic Ad Placement and the Premium Page Monopoly

A key insight from the data is the strategic nature of corporate advertising. Corporate entities dominate "premium" spots, specifically the 1st, 3rd, and last pages. While the government is the largest overall spender in the study's estimate ($1.02 billion vs. $890 million for corporations), its spending is distributed differently.

Comparison of advertising spend (in billion dollars) between companies and the government across different page types.

As shown in the data, the government relegates 88.4% of its ads to "inner" pages, often for legally mandated notices and tenders. In contrast, corporate advertisers prioritize high-visibility pages and the "odd-page" bias—the preference for right-hand pages that are more immediately visible to readers.

Quantifying Media Bias

The study provides empirical evidence for the "advertiser's halo" effect. For corporate entities, the regression analysis showed that a 1% increase in a company’s weighted ad ratio leads to an average increase of 0.0189 units in total sentiment score. While this number may appear small, on a scale of -1 to 1, it represents a statistically significant shift toward favorable coverage.

Furthermore, higher ad spend was a strong predictor of article volume. Companies that advertised more were written about more frequently, regardless of their actual "newsworthiness" as measured by external metrics like Google Trends popularity. Interestingly, this effect does not extend to the government; increased government advertising often correlates with more negative sentiment, likely because these ads are non-discretionary and do not purchase the same editorial goodwill as corporate campaigns.

Conclusion

This research demonstrates that while print circulation may be declining globally, the influence of advertisers on the "Fourth Estate" in India remains potent. By leveraging automated computer vision, the study reveals how corporate spend strategically shapes both the volume and the tone of news coverage. While proving direct causality remains a challenge, the strong correlation between ad spend and favorable sentiment highlights the ongoing tension between commercial interests and journalistic independence.

Summary updated

BibTeX
@misc{vardhan2025analyzing,
  author = {N. Harsha Vardhan and Ponnurangam Kumaraguru and Kiran Garimella},
  title = {{Analyzing Patterns and Influence of Advertising in Print Newspapers}},
  year = {2025},
  eprint = {2505.10791},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2505.10791},
}
Figure