[Paper Review] Food for Thought: Analyzing Public Opinion on the Supplemental Nutrition Assistance Program

Miriam Chappelka, Jihwan Oh+2 more|arXiv (Cornell University)|Oct 6, 2017

Sentiment Analysis and Opinion Mining5 references3 citations

TL;DR

This study analyzes public opinion on the Supplemental Nutrition Assistance Program (SNAP) using machine learning, sentiment analysis, and text mining on news and social media coverage across national and state levels. It reveals predominantly negative sentiment in media, with more extreme partisanship in outlets and regional clustering of negative reporting in the Midwest, informing advocacy through an online application for stakeholders.

ABSTRACT

This project explores public opinion on the Supplemental Nutrition Assistance Program (SNAP) in news and social media outlets, and tracks elected representatives' voting records on issues relating to SNAP and food insecurity. We used machine learning, sentiment analysis, and text mining to analyze national and state level coverage of SNAP in order to gauge perceptions of the program over time across these outlets. Results indicate that the majority of news coverage has negative sentiment, more partisan news outlets have more extreme sentiment, and that clustering of negative reporting on SNAP occurs in the Midwest. Our final results and tools will be displayed in an on-line application that the ACFB Advocacy team can use to inform their communication to relevant stakeholders.

Motivation & Objective

• To understand public perception of SNAP through analysis of media and social media coverage.
• To examine how sentiment toward SNAP varies across partisan and regional media outlets.
• To identify temporal and geographic patterns in media coverage of SNAP and food insecurity.
• To develop an online tool for advocacy groups to inform stakeholder communication based on data-driven insights.
• To link public opinion trends with elected officials' voting records on SNAP-related issues.

Proposed method

• Applied machine learning and text mining to extract and categorize discussions of SNAP from national and state-level news and social media sources.
• Conducted sentiment analysis to classify media coverage as positive, negative, or neutral, focusing on public perception.
• Used clustering techniques to detect geographic patterns in negative reporting, particularly in the Midwest.
• Correlated media sentiment with voting records of elected representatives on SNAP and food insecurity issues.
• Developed an interactive online application to visualize findings for use by advocacy organizations like ACFB.
• Employed natural language processing (NLP) techniques to handle large-scale textual data from diverse media sources.

Experimental results

Research questions

• RQ1 How does sentiment toward SNAP vary across different types of media outlets, particularly partisan versus non-partisan ones?
• RQ2 Are there geographic clusters of negative media coverage on SNAP, and if so, where are they located?
• RQ3 How has public sentiment toward SNAP evolved over time in national and state-level media?
• RQ4 To what extent do media portrayals of SNAP align with the policy positions of elected representatives?
• RQ5 Can machine learning and sentiment analysis effectively identify and track public opinion trends on social welfare programs like SNAP?

Key findings

• The majority of news coverage on SNAP exhibits negative sentiment, indicating a pervasive critical tone in media narratives.
• Partisan news outlets display more extreme sentiment, with significantly higher levels of negativity compared to non-partisan sources.
• Negative media reporting on SNAP is clustered geographically, particularly in the Midwest region of the United States.
• Temporal analysis reveals sustained negative sentiment in media coverage over time, suggesting persistent public perception challenges.
• The study successfully linked media sentiment trends with legislative voting records on SNAP, showing alignment in certain districts.
• An online application was developed to present findings and support advocacy efforts by the ACFB team.