Package: aihuman 1.0.1

Overview | Data Preparation & Descriptive Analysis | Nuisance functions | Contingency table of human decisions and PSA recommendations | Agreement between human decisions and PSA recommendations | Human+AI v. Human comparison | How the human overrides the AI recommendation? | AI v. Human comparison | Alternative AI Recommendations | Different loss functions | Policy Learning | Helper functions | Whether to provide PSA | Whether to follow PSA

Overview | Synthetic data | Data 1: synth | Data 2: psa_synth | Data 3: hearingdate_synth | Interim data | Descriptive statistics | Distribution of the judge's decisions (Figure 1) | Replication codes | Intention to treat effects of PSA Provision (Figure 2) | Main analysis: Average Principal Causal Effects (APCE) | AiEvalmcmc() | CalAPCE() or CalAPCEparallel() | APCEsummary() | PlotAPCE() (Figure 4) | Principal strata (Figure 3) | Principal fairness (Figure 5) | Optimal decision (Figure 6) | Comparison between the judge's decisions and DMF recommendations (Figure 7) | Test of spillover effects: Conditional Randomization Test (CRT) | Frequentist analysis | Sensitivity analysis