Powerful t-Tests in the presence of nonclassical measurement error

Arbeitspapier

This paper proposes a powerful alternative to the t-test in linear regressions when a regressor is mismeasured. We assume there is a second contaminated measurement of the regressor of interest. We allow the two measurement errors to be nonclassical in the sense that they may both be correlated with the true regressor, they may be correlated with each other, and we do not require any location normalizations on the measurement errors. We propose a new maximal t-statistic that is formed from the regression of the outcome onto a maximally weighted linear combination of the two measurements. Critical values of the test are easily computed via a multiplier bootstrap. In simulations, we show that this new test can be significantly more powerful than t-statistics based on OLS or IV estimates. Finally, we apply our test to the study of returns to education based on twins data from the U.S..

Object Details

• Language: Englisch

References and Relationships

• Bibliographic citation: Series: cemmap working paper ; No. CWP57/17

Classification and Topics

• Classification: Wirtschaft
• Subject: linear regression ; adaptive test ; power of test ; maximal combination of measurements ; repeated measurements ; multiplier bootstrap

Contributors, Places and Time

• Event: Geistige Schöpfung
• (who): Kim, Dongwoo ; Wilhelm, Daniel
• Event: Veröffentlichung
• (who): Centre for Microdata Methods and Practice (cemmap)
• (where): London
• (when): 2017

Further information

• DOI: doi:10.1920/wp.cem.2017.5717
• Handle: http://hdl.handle.net/10419/189798
• Last update: 10.03.2025, 11:42 AM CET