Deep Dive into Anonymity: A Large Scale Analysis of Quora Questions

Binny Mathew, Ritam Dutt, Suman Kalyan Maity, Pawan Goyal, Animesh Mukherjee

TL;DR

This large-scale study of 5.1 million Quora questions reveals that anonymity is a mainstream, normative tool rather than a niche for trolls. While global linguistic differences between anonymous and non-anonymous questions are negligible due to "topical mixing," a deep dive into specific clusters shows that anonymity is primarily used for seeking support on sensitive personal issues—and surprisingly, these questions receive faster community responses than general ones.

What Problem Does This Paper Address?

Anonymity on the internet is often associated with the "online disinhibition effect," frequently manifesting as trolling, cyberbullying, or toxic behavior on platforms like 4chan or Reddit. However, Quora presents a unique case: it is a mainstream Q&A site where experts and laypeople coexist, and anonymity is an explicitly supported, socially acceptable feature.

The authors set out to understand the mechanics of anonymity in this "normative" environment. They address several key questions:

  • Does the linguistic structure of a question change when the asker is anonymous?
  • What topics drive people to hide their identities?
  • How does the community's perception of a topic (reflected in their answers) differ from the asker's perception?
  • Does anonymity affect how quickly a question is answered?

Methods

The researchers employed a multi-stage analytical pipeline:

  1. Linguistic Analysis: They used SpaCy for Part-of-Speech (POS) tagging, VADER for sentiment analysis, and LIWC/Empath for psycholinguistic categorization (e.g., identifying words related to "family," "sexual," or "pain").
  2. Statistical Rigor: Because the dataset is massive, standard t-tests would flag even tiny, meaningless differences as "significant." Instead, they used Cohen’s $d$ to measure effect size, ensuring they only reported differences that are practically meaningful.
  3. Topical Clustering: They grouped questions by Quora’s user-assigned topics to calculate an Anonymity Ratio ($a_r$) for each topic.
  4. Personality Mapping: By correlating LIWC categories with established psychological research (the Big-Five traits), they inferred the personality profiles of users who post anonymously.
  5. The Anonymity Grid: A novel $2 \times 2$ framework comparing the anonymity of the question vs. the anonymity of the answers to identify mismatches in social stigma.

Datasets

The study utilizes a massive dataset of 5,160,765 Quora questions and 488,122 topics, originally collected by Maity et al. (2015). The data includes timestamps, user IDs (where available), and logs of anonymous vs. known activity for both questions and answers.

Key Insights and Ideas

  • The "Topical Mixing" Effect: At a global scale, anonymous and known questions look linguistically identical. The authors discovered this is because the "mask" is used inconsistently. People sometimes ask about sensitive topics (like depression) openly, while asking mundane questions (like "What is the best camera?") anonymously. This "mixing" washes out statistical differences until you zoom into specific topical clusters.
  • Anonymity as a Support Tool: When zooming into "Anonymous Clusters," the language shifts dramatically. These questions use significantly more first-person pronouns ("I," "me") and verbs, focusing on personal struggles like anxiety, sexual health, and relationship advice.
  • The Anonymity Grid Mismatch: The paper identifies fascinating social dynamics through its grid:
  • Question Anon / Answer Known: Personal health/grooming (e.g., "How do I get rid of acne?"). The asker wants privacy, but experts are happy to provide advice openly for social recognition.
  • Question Known / Answer Anon: Politics and Religion. These are "safe" to ask but "dangerous" to answer due to potential backlash or controversy.
  • The Speed of Empathy: In a counter-intuitive finding, anonymous questions on highly sensitive topics (the "Anon Positive" group) have the lowest First Response Time (FRT). The community appears to prioritize these vulnerable queries, providing a high level of engagement and support.

Key Results

  • Prevalence: Approximately 38.7% of all questions on Quora are posted anonymously.
  • Linguistic Shift: In anonymous clusters, the use of personal pronouns is 124% higher than in known clusters.
  • Sentiment: Anonymous questions carry 75% more negative sentiment compared to known questions, reflecting their use for discussing pain and distress.
  • Response Speed: The median FRT for the "Anon-Anon" group (sensitive topics where both asker and answerer are anonymous) is just 1.28 hours, compared to 4.01 hours for general "Known-Known" topics.
  • Personality: Users posting anonymously show a positive correlation with Extraversion and Agreeableness, but a negative correlation with Openness.

Limitations and Future Work

The authors acknowledge that their personality analysis is inferred at the cluster level rather than the individual level, as they do not have direct access to private user data.

Future directions include: * Temporal Dynamics: Investigating how the anonymity of a topic changes over years as social norms evolve (e.g., does a topic become "less taboo" over time?). * Demographics: Using name-based inference APIs to study how anonymity usage varies across different genders and ethnicities. * Answer Content: Moving beyond metadata to perform deep linguistic analysis on the content of anonymous answers.

Summary updated

Abstract

Anonymity forms an integral and important part of our digital life. It enables us to express our true selves without the fear of judgment. In this paper, we investigate the different aspects of anonymity in the social Q&A site Quora. The choice of Quora is motivated by the fact that this is one of the rare social Q&A sites that allow users to explicitly post anonymous questions and such activity in this forum has become normative rather than a taboo. Through an analysis of 5.1 million questions, we observe that at a global scale almost no difference manifests between the linguistic structure of the anonymous and the non-anonymous questions. We find that topical mixing at the global scale to be the primary reason for the absence. However, the differences start to feature once we "deep dive" and (topically) cluster the questions and compare the clusters that have high volumes of anonymous questions with those that have low volumes of anonymous questions. In particular, we observe that the choice to post the question as anonymous is dependent on the user's perception of anonymity and they often choose to speak about depression, anxiety, social ties and personal issues under the guise of anonymity. We further perform personality trait analysis and observe that the anonymous group of users has positive correlation with extraversion, agreeableness, and negative correlation with openness. Subsequently, to gain further insights, we build an anonymity grid to identify the differences in the perception on anonymity of the user posting the question and the community of users answering it. We also look into the first response time of the questions and observe that it is lowest for topics which talk about personal and sensitive issues, which hints toward a higher degree of community support and user engagement.
BibTeX
@inproceedings{mathew2019deep,
  author = {Binny Mathew and Ritam Dutt and Suman Kalyan Maity and Pawan Goyal and Animesh Mukherjee},
  title = {{Deep Dive into Anonymity: A Large Scale Analysis of Quora Questions}},
  booktitle = {SocInfo},
  year = {2019},
  doi = {10.1007/978-3-030-34971-4\_3},
}

Loading related content...

Figure