Causal Inference with Measurement Errors
摘要
In many empirical studies, it may not be available to observe the accurate data. For instance, [1] indicated that over 75% of the variance in fat intake data from the Epidemiologic Study Cohort in NHANES-I stems from measurement error. This chapter focuses on continuous treatment data measured with classical error, where researchers only observe the sum of the treatment and a random error, rather than the actual treatment received.