The Civil Rights Act was stalled in Congress, facing fierce opposition from Southern Democrats.
U.S. military advisors in Vietnam increased from 900 to 16,000 during JFK's presidency.
The Cuban Missile Crisis had occurred just one year earlier, in October 1962.
Unemployment was at 5.5%, with GDP growth averaging 4.4% during JFK’s term.
Louis Harris & Associates surveyed 1,180 likely voters for Newsweek just before JFK's assassination.
57% of likely voters approved of JFK’s job performance.
Γ shows how much hidden bias would need to exist to change your conclusions.
Adjusts one factor at a time to show its impact on predicted approval.
Thousands of simulations based on the model's uncertainty, generating a full distribution of possible approval rates.
Adjust these key factors from November 1963 to see how they might have affected presidential approval.
Predicted Probability:
-
Probability Change from 57% Base:
-
95% Confidence Interval:
-
Standard Error:
-
Predicted # Approvers:
-
Approver # Change from ~40M Base:
-
Response Changes This chart shows how response frequencies change when modifying a key factor. Red bars indicate modified response levels, while green bars show values left unchanged.
Approval Comparison This chart compares the overall approval probability between the original estimate and the predicted outcome, including confidence intervals to show the uncertainty around each estimate.
Simulation Outcomes This chart visualizes thousands of simulated approval outcomes, providing a detailed picture of uncertainty and potential variability beyond simple point estimates. Tight distributions suggest stability, while wide spreads indicate greater risk or unpredictability.
Probability Distribution This chart compares the full distribution of voter approval probabilities between the original model and the modified scenario, revealing how voter likelihoods shift across the population when conditions change.
This table shows how much hidden bias (Γ) would be needed to overturn the simulated approval result. For example, someone might argue that JFK’s approval was inflated because his supporters were more likely to respond. The Γ value indicates how strong such unmeasured bias would have to be.
If the approval estimate remains stable even at higher Γ values (e.g., 2 or 3), the result is likely robust. This method mirrors techniques from early public health research—like studies linking smoking to cancer— which showed that only extremely strong hidden biases could undo the observed effects.
Γ helps evaluate how much hidden bias could alter the outcome. Higher values represent stronger potential for unmeasured confounding. Use the slider below to explore this risk.