6 minute read

Introduction

The proliferation of social media has profoundly influenced political communication, offering new platforms for interaction between politicians and the public. Twitter, in particular, has emerged as a crucial medium for political discourse, allowing Members of Parliament (MPs) to engage directly with constituents and peers. However, concerns about the creation of “filter bubbles” and “echo chambers”—where individuals are exposed only to information and interactions that reinforce their existing beliefs—have raised questions about the true democratic potential of these platforms.

The study “Bursting the (Filter) Bubble: Interactions of Members of Parliament on Twitter” by Sofia Ferro-Santos, Gustavo Cardoso, and Susana Santos investigates these phenomena within the context of Portuguese MPs’ Twitter interactions. By analyzing tweets and interactions over a specified period, the authors aim to determine whether MPs primarily use Twitter for broadcasting or for genuine interaction, and to what extent these interactions reflect political and status homophily.

This review critically examines the study’s methodology, findings, and contributions to the field, highlighting key limitations and gaps. Despite the valuable insights provided, the study’s reliance on a limited timeframe, outdated analytical tools, and a small sample size raises questions about the robustness of its conclusions. Additionally, the lack of novel advancements in understanding MPs’ social media behavior suggests a need for further research utilizing more modern methodologies and broader data sets. This review aims to contextualize these issues within the broader discourse on digital political communication, offering directions for future research to enhance our understanding of political interactions on social media.

Key Limitations and Gaps

  1. Short Timeframe: The analysis spans only four one-week periods over four months, which may not capture the full range of MPs’ interactions or account for variations due to political cycles, special events, or legislative sessions.

  2. Specific Political Period: Data collection coincided with significant political events, such as the final budget vote, potentially skewing interaction patterns and limiting the generalizability of findings to other times or political climates.

  3. Content Quality and Context: The study focuses on interaction frequency and type without analyzing tweet content or context, missing insights into the quality and substance of political communication on Twitter.

  4. Exclusion of Likes: The analysis does not consider “likes” as a form of interaction, ignoring a form of engagement that could reflect MPs’ preferences and influence.

  5. Limited Scope of Accounts: The sample size is reduced to 757 public accounts after excluding 33 deleted or private accounts. Coding based on public information and recent tweets may not fully capture account affiliations or influence.

  6. Bias in Coding: The subjective judgment involved in categorizing accounts into political affiliations and other groups could introduce bias, affecting the accuracy of identified interaction patterns.

  7. Focus on Twitter: The exclusive examination of Twitter interactions may not represent MPs’ overall social media presence or interactions on other platforms like Facebook or Instagram, where different dynamics might exist.

  8. General Linear Model Constraints: The statistical model used may not account for all variables influencing MPs’ interaction patterns. Factors like individual MP characteristics or external events might also significantly impact results.

  9. Potential for “Spiral of Noise”: The “spiral of noise” phenomenon, where dissonant voices fill MPs’ comment sections, might deter MPs from meaningful engagement. This was not directly measured or analyzed, potentially overlooking a key factor influencing MPs’ behavior.

  10. Homophily and Interaction Dynamics: While the study examines political and status homophily, it does not explore other homophily forms based on issues or demographics, which could provide a more nuanced view of interactions.

  11. Lack of Novelty: The subject matter is extensively studied, and this research does not introduce new advancements or insights. It largely reaffirms existing findings about MPs’ Twitter use.

  12. Outdated Technologies: The use of SPSS and other outdated technologies limits the study’s methodological robustness. More advanced tools and methods could yield better insights and more accurate analyses.

  13. Poor Statistical Analysis: Analyzing only 750 tweets from a population of 230 MPs is statistically inadequate, leading to potentially unreliable conclusions due to the small sample size.

Possible Future Research Directions

  1. Extended Timeframes: Future studies should analyze longer periods to capture comprehensive interaction patterns and account for political cycles and events.

  2. Content Analysis: Incorporate qualitative content analysis to assess interaction quality and context, providing deeper insights into political discourse on social media.

  3. Comparative Platform Studies: Examine MPs’ interactions across multiple social media platforms to identify differences and similarities in communication strategies and engagement.

  4. Broader Account Inclusions: Include “likes” and a broader range of accounts to provide a more holistic view of engagement and influence.

  5. Advanced Coding Techniques: Employ objective and automated coding methods to minimize bias and improve account categorization and interaction analysis accuracy.

  6. Influence of External Factors: Investigate external events, individual MP characteristics, and other contextual factors’ impact on social media interaction patterns.

  7. Utilize Modern Analytical Tools: Use more advanced and up-to-date analytical tools and methodologies to improve the study’s robustness and accuracy.

  8. Larger Sample Sizes: Ensure larger sample sizes for more reliable statistical analyses, enhancing the validity and reliability of findings.

Analysis from a Data Science Perspective

From a data science perspective, the study “Bursting the (Filter) Bubble: Interactions of Members of Parliament on Twitter” presents a mixed bag of strengths and limitations. The study employs social network analysis and statistical modeling to investigate MPs’ interactions on Twitter, providing valuable insights into the communication patterns of political actors. However, several aspects of the methodology and data analysis warrant closer scrutiny.

Strengths

  1. Social Network Analysis (SNA): The use of SNA is a robust choice for visualizing and understanding the complex web of interactions between MPs and other Twitter users. Tools like Gephi effectively map these relationships, offering a clear picture of the network’s structure and the centrality of various actors within it.

  2. Categorization of Interactions: By differentiating between types of interactions (retweets, replies, quote-tweets, mentions), the study provides a nuanced view of how MPs use Twitter. This granularity is crucial for understanding the varying degrees of engagement and the potential democratic value of these interactions.

Limitations

  1. Outdated Analytical Tools: The reliance on SPSS for statistical analysis is a significant drawback. Modern data science often leverages more powerful and flexible tools such as Python (with libraries like Pandas, NumPy, and SciPy), R, or machine learning frameworks like TensorFlow and PyTorch. These tools offer advanced capabilities for data manipulation, visualization, and statistical modeling, potentially leading to more robust and insightful findings.

  2. Limited Sample Size and Timeframe: Analyzing only 750 tweets from 69 MPs over four one-week periods significantly limits the study’s statistical power and generalizability. A larger dataset over a more extended period would provide a more comprehensive understanding of MPs’ Twitter behavior and improve the reliability of the results.

  3. Lack of Content Analysis: The study focuses on interaction types without delving into the content of the tweets. Natural Language Processing (NLP) techniques could be employed to analyze tweet content, sentiment, and topic modeling. This approach would provide deeper insights into the nature and quality of interactions, beyond mere frequency and type.

  4. Potential Bias in Account Coding: The manual coding of accounts based on public information and recent tweets introduces subjectivity and potential bias. Automated classification techniques, such as machine learning classifiers, could enhance accuracy and consistency in categorizing accounts and their affiliations.

  5. General Linear Model (GLM) Constraints: The use of a General Linear Model to analyze interaction patterns may not capture the complexity of the data. Advanced statistical methods, such as generalized additive models (GAMs) or mixed-effects models, could provide a more nuanced understanding of the factors influencing MPs’ Twitter interactions.

Recommendations for Future Research

  1. Utilize Advanced Data Analytics Tools: Future studies should leverage modern data science tools and frameworks to enhance analytical rigor and flexibility.

  2. Expand the Dataset: Collect data over a longer period and include a larger sample of MPs and their tweets to improve the robustness and generalizability of findings.

  3. Incorporate NLP Techniques: Use NLP to analyze tweet content for sentiment, topics, and other qualitative aspects of communication, providing deeper insights into the nature of interactions.

  4. Automate Account Classification: Employ machine learning algorithms to classify accounts, reducing bias and improving the accuracy of account categorization.

  5. Apply Advanced Statistical Methods: Utilize more sophisticated statistical models to better capture the complexities of interaction patterns and their determinants.

While the study offers valuable insights into MPs’ Twitter interactions, adopting more advanced data science methodologies and tools could significantly enhance the depth and reliability of future research in this area.

You may also enjoy