Sensitivity analyses that quantify the robustness of inferences to concerns about omitted variables and other sources of bias.
Questions? Issues? Suggestions? Reach out through the KounFound-It! Google Group.
Latest News
Announcing the NEW AND IMPROVED KonFound-it app and upcoming workshops
12/19/2024
Happy holidays to you!
We have released new versions of the konfound package in R (1.0.2) and in Stata. New and updated features include:
- Conditional Robustness of Inference for Replacement (CRIR) in which there is no relationship between predictor and outcome in the replacement data conditional on other terms in the model (e.g., for use with interaction effects when models include main effects as applied in diff in diff).
- 2x2 and logistic regression added to Stata (already in R).
- More options for Fragility index (e.g.,
switch_trm)
- More options for Fragility index (e.g.,
- Unconditional Impact Threshold for a Confounding Variable is provided when possible.
- See Lonati, S., & Wulff, J. N. (2024). Hic Sunt Dracones: On the risks of comparing the ITCV with control variable correlations. Journal of Management. DOI: 10.1177/01492063241293126
- Coefficient of Proportionality – how strong would selection on unobserved covariates have to be relative to observed covariates to nullify an estimated effect.
index = "COP" - Directly specify a threshold for inference (other than statistical significance) via
eff_thr - Specify a non-zero null hypothesis for significance testing via nu
- Application to What Works Clearinghouse Benchmarks for educational research
- All raw results provided (in R,
to_return = "raw", in Stata use"return list"after command) - Improved statements in print out
- Improved KonFound-It app interface: https://konfound-project.shinyapps.io/konfound-it/
Read an overview of the package at https://konfound-it.org/konfound/ and read through the Introduction to konfound vignette.
Also, be sure to check out an AI-generated podcast, created by Google’s NotebookLM, discussing the KonFound-It! team’s article “Quantifying the robustness of causal inferences: Sensitivity analysis for pragmatic social science” in Social Science Research (Frank, Lin, Xu, Maroulis, & Mueller, 2023):
Sign up today for the next KonFound-It project workshop facilitated by Dr. Ken Frank: Sensitivity Analysis for Causal Inference. The workshop will take place fully online March 5–6, 2025 (10:30am–12:30pm and 1:00pm–3:00pm Eastern time both days).
Start KonFounding
Try out KonFound-It! to calculate sensitivity analyses through an interactive web app.
Benchmarks
Explore sensitivity analyses calculated for the What Works Clearinghouse through an interactive web app.
Learn More
Browse through a variety of resources and guides.