Text and Causal Inference: A Review of Using Text to Remove Confounding from Causal Estimates

  title={Text and Causal Inference: A Review of Using Text to Remove Confounding from Causal Estimates},
  author={Katherine A. Keith and David Jensen and B. O'Connor},
  • Katherine A. Keith, David Jensen, B. O'Connor
  • Published 2020
  • Computer Science
  • ArXiv
  • Many applications of computational social science aim to infer causal conclusions from non-experimental data. Such observational data often contains confounders, variables that influence both potential causes and potential effects. Unmeasured or latent confounders can bias causal estimates, and this has motivated interest in measuring potential confounders from observed text. For example, an individual's entire history of social media posts or the content of a news article could provide a rich… CONTINUE READING
    13 Citations

    Figures, Tables, and Topics from this paper

    Uncovering Latent Biases in Text: Method and Application to Peer Review
    • 4
    • PDF
    Causal Effects of Linguistic Properties
    • PDF
    Controlled Analyses of Social Biases in Wikipedia Bios
    • PDF
    Quantifying the Causal Effects of Conversational Tendencies
    • Justine Zhang
    • Psychology, Computer Science
    • Proc. ACM Hum. Comput. Interact.
    • 2020
    • 2
    • PDF
    Unsupervised Discovery of Implicit Gender Bias
    • 3
    • PDF
    Robustness to Spurious Correlations in Text Classification via Automatically Generated Counterfactuals
    • PDF


    Adjusting for Confounding with Text Matching
    • 21
    • Highly Influential
    • PDF
    Using Text Embeddings for Causal Inference
    • 9
    • Highly Influential
    Causal Effect Inference with Deep Latent-Variable Models
    • 175
    • PDF
    Challenges of Using Text Classifiers for Causal Inference
    • 21
    • PDF
    Adapting Text Embeddings for Causal Inference
    • 8
    • Highly Influential
    • PDF
    Endogenous Selection Bias: The Problem of Conditioning on a Collider Variable.
    • 410
    • PDF