Skip directly to content

Zlata Tanović - Research Summary

Universitat Pompeu Fabra

The test effect: Behavioral change and potential biases due to (biomedical) testing in surveys

Does biomedical testing change health care seeking behavior of the research sample, and can it bias impact estimates of a related health care intervention? This paper is the first to rigorously analyze unintended effects of biomedical testing in surveys. Random assignment of blood pressure measurements in a 2013 household survey in Tanzania, and a second survey of the same individuals two years later, allows for the identification of this "test effect" on health care provider consultations for hypertension (chronic high blood pressure) and uptake of voluntary health insurance. As these were the baseline and follow-up surveys of a health insurance impact evaluation, the possible bias in the insurance impact estimates caused by the test effect can also be estimated. Since, complying with ethical standards, respondents who were tested were told their test result, the differential effect of high versus normal measured blood pressure can be determined. Having high measured blood pressure significantly increased the likelihood of health care provider consultations for hypertension. No evidence is found of a test effect for health insurance enrollment, nor of a bias in health insurance impact estimates due to the blood pressure measurements, for any of the outcomes.
 

Download the paper here