1. **[[Design Pattern I - Instrumental Variable (IV)]]** $\quad$ 1.1. **Understanding the problem and solution** $\quad\quad$ 1.1.1. [[Confounding by intention]] $\quad$ 1.2. **Solving the problem using the IV pattern** $\quad\quad$ 1.2.1. [[Data and conceptual model]] $\quad\quad$ 1.2.2. [[Statistical modeling of IV]] $\quad\quad\quad$ 1.2.2.1. [[Replication of IV in Python]] $\quad\quad$ 1.2.3. [[(Double) Machine learning using IV]] $\quad\quad$ 1.2.4. [[IV the Bayesian way]] $\quad\quad$ 1.2.5. [[patterns/iv/Key takeaways|Key takeaways (IV)]] $\quad$ 1.3. **Oh my the IV!** $\quad\quad$ 1.3.1. [[Oh my! Different standard errors everywhere|Different standard errors everywhere]] $\quad\quad$ 1.3.2. [[Oh my! Larger effect size in the IV model|Larger effect size in the IV model]] $\quad\quad$ 1.3.3. [[Oh my! Within R-squared in the IV model|Within R-squared in the IV model]] $\quad\quad$ 1.3.4. [[Oh my! Priors in the Bayesian IV model|Priors in the Bayesian IV model]] $\quad$ 1.4. **Cases using the design pattern** $\quad\quad$ 1.4.1. [[Impact of recommendations on Amazon]] $\quad$ 1.5. **Where to from here** $\quad\quad$ 1.5.1. [[Data centricity in the IV pattern]] $\quad\quad$ 1.5.2. [[Using LLMs for IV discovery and data]] 2. **[[Design Pattern II - Regression Discontinuity (RD)]]** $\quad$ 2.1. **Understanding the problem and solution** $\quad\quad$ 2.1.1. [[Local identification by a cutoff]] $\quad$ 2.2. **Solving the problem using the RD pattern** $\quad\quad$ 2.2.1. [[Data and conceptual model (RD)|Data and conceptual model]] $\quad\quad$ 2.2.2. [[Statistical modeling of RD]] $\quad\quad\quad$ 2.2.2.1. [[Replication of RD in Python]] $\quad\quad$ 2.2.3. [[(Double) Machine learning using RD]] $\quad\quad$ 2.2.4. [[RD the Bayesian way]] $\quad\quad$ 2.2.5. [[patterns/rd/Key takeaways|Key takeaways (RD)]] $\quad$ 2.3. **Oh my the RD!** $\quad\quad$ 2.3.1. [[Oh my! Bandwidth sensitivity in the RD model|Bandwidth sensitivity in the RD model]] $\quad\quad$ 2.3.2. [[Oh my! Polynomial order and overfitting in RD|Polynomial order and overfitting in RD]] $\quad\quad$ 2.3.3. [[Oh my! The point estimate is not centered in the CI|The point estimate is not centered in the CI]] $\quad\quad$ 2.3.4. [[Oh my! DoubleML is worse for the RD design|DoubleML is worse for the RD design]] $\quad\quad$ 2.3.5. [[Oh my! Priors in the Bayesian RD model|Priors in the Bayesian RD model]] $\quad$ 2.4. **Cases using the design pattern** $\quad\quad$ 2.4.1. [[Yelp ratings and restaurant revenue]] $\quad$ 2.5. **Where to from here** $\quad\quad$ 2.5.1. [[Data centricity in the RD pattern]] 3. Design Pattern III - Parallel Trends $\quad$ 3.1. Understanding the problem and solution $\quad$ 3.2. ... 4. Design Pattern IV - Parallel Trends Extended $\quad$ 4.1. Understanding the problem and solution $\quad$ 4.2. ... 5. Design Pattern V - Factor Models and Imputations $\quad$ 5.1. Understanding the problem and solution $\quad$ 5.2. ... 6. Design Pattern VI - Matching and Weighting $\quad$ 6.1. Understanding the problem and solution $\quad$ 6.2. ... 7. Design Pattern VII - Policy Learning $\quad$ 7.1. Understanding the problem and solution $\quad$ 7.2. ... 8. Design Pattern VIII - Exposure Mapping $\quad$ 8.1. Understanding the problem and solution $\quad$ 8.2. ... 9. Design Pattern IX - Causal Decomposition $\quad$ 9.1. Understanding the problem and solution $\quad$ 9.2. ... 10. Design Pattern X - Controlled Experiments $\quad$ 10.1. Understanding the problem and solution $\quad$ 10.2. ... > [!info]- Last updated: May 14, 2026