Intention-to-Treat Error

Failing to analyze all cases in the group originally assigned — dropping out people who didn't complete the treatment — produces systematically misleading conclusions.

• A weight loss drug trial that only analyzes patients who completed the program excludes those who dropped out due to side effects — making the drug look far more effective and safer than it is.
• Speed cameras appear to make roads safer because the most dangerous drivers — repeat offenders who don't change behavior — eventually lose their licenses and leave the data set.
• A job training program appears highly successful because the most unemployable participants drop out early — the 'graduates' were always more employable.