This study uses a nationally representative survey to analyze a key survey design decision in child labor measurement: self-reporting versus proxy interviewing. The child/proxy disagreement affects 20 percent of the sample, which translates into a 17.1 percentage point difference in the national...