Intention-to-Treat (ITT) Instrumentation

In Simple Words

When some assigned groups fail to receive a promised intervention, using initial random assignment as an indicator to adjust overall outcomes is like measuring a fitness program's impact based on everyone invited, then mathematically adjusting for those who actually attended the gym.

Research question

In field experiments, non-compliance—such as administrative failures where assigned schools fail to receive tutors or participants decline intervention—biases direct treatment estimates.

Participants

N = 9,986 Grade 3 and Grade 4 students across 77 municipal primary schools in Mumbai, India.

Study Design

Instrumental Variables (IV) / Two-Stage Least Squares (2SLS) regressing test score improvements on actual treatment status using random assignment intention-to-treat (ITT) as an instrument.

Results

In Year 2 in Mumbai, 33% of schools assigned to receive a balsakhi tutor did not receive one due to administrative difficulties; 2SLS instrumentation successfully eliminated non-compliance bias to yield unbiased direct treatment estimates.

Takeaway

Use initial random assignment as an instrumental variable in Two-Stage Least Squares regression when evaluating educational interventions affected by program non-compliance or implementation failure.

Related findings

• Catchment Cohort Size Instrument →Methods & evidence

  To measure how school grant size truly affects family spending without getting confused by family choices, researchers look at how many children of school age happen to live in the surrounding area. Since the total grant amount is fixed per school, having more nearby children automatically shrinks the per-child funding without relying on family preferences or behavior.

• Financial Outcome Winsorization →Methods & evidence

  Imagine scoring a gymnastics competition where one judge gives a score of 100 while everyone else scores between 5 and 10. To prevent that single extreme score from ruining the average, you cap any score higher than 10 down to exactly 10 before calculating the final result.

• Indirect Inference Structural Estimation →Methods & evidence

  Imagine trying to map the unseen internal gears of a complex clock. Instead of opening it, you run experiments—shaking it at different speeds and measuring how the hands move.

• Instrumental Variable Estimation under One-Sided Non-Compliance →Methods & evidence

  Imagine testing whether wearing a new type of safety goggles reduces eye injuries in a factory. If workers randomly offered goggles choose to wear them only half the time, while unoffered workers can never get them, researchers can use the initial random offer as a lever to isolate the pure effect of actually wearing the goggles.