中文

Faculty & Research

Insights

9886

Research background

Information is valuable in standard economic analysis because it improves decision-making. However, there are many situations in which people avoid useful information (see Golman et al. (2017) for a literature review). Even so, most studies focus on the overall information avoidance effect rather than the heterogeneous effect. Therefore, we cannot explain why some people seek or avoid health information. In particular, we do not have systematic evidence on whether high-risk individuals are more likely to avoid medical tests and how a tendency to avoid information varies with the test price and the type of disease. To investigate this issue, we conducted a randomized field experiment in rural China to find out the take-up rate for medical tests.

Literature Review

There are a few alternative explanations for the low take-up rate for medical tests. One explanation for the general tendency to avoid the test, based on the 4 neoclassical model, is the high price elasticity (Thornton, 2008); another behavioral explanation is procrastination generated by present bias. However, neither of the two explanations would predict that high-risk individuals are more likely to avoid the test. Our experimental design also excludes procrastination, because all individuals have already paid the upfront cost of being onsite.

Many empirical studies find that people tend to avoid important information regarding their health status (Lyter et al. 1987; Lerman et al., 1999; Sullivan et al., 2004; Thornton 2008; Oster et al. 2013; Ganguly and Tasoff, 2016). However, most previous studies focus on the overall information avoidance effect; only a few investigate the heterogeneous effect across the probability of having the disease. For the latter group, some find that people with higher risk of having cancer tend to delay a visit to the doctor (Caplan, 1995; Meechan et al., 2002; Persoskie et al., 2014). However, using elicited subjective beliefs, Oster et al. (2013) find that individuals with higher subjective belief about disease risk were more likely to pursue being tested for Huntington’s disease, and people were generally overly optimistic about the risk of having such disease. Meanwhile, Okeke et al. (2013) conducted a randomized trial in Nigeria with varying prices for cervical cancer screening. Despite the lack of statistics significance, they found that high-risk subjects (for both subjective and objective risk) tended to accept a higher test price in general.

To our best knowledge, this paper is the first field experiment to find that individuals with high subjective belief about the disease risk tend to avoid testing. In terms of cross-disease comparison, Ganguly and Tasoff (2016) find that more people are willing to forgo a $10 payment to avoid learning the results of the herpes simplex virus 2 (HSV-2) test than an HSV virus 1 (HSV-1) test, where HSV-2 is viewed as a more serious condition. Our comparison of diabetes and cancer shows a similar result.

Research Method

We collaborated with a local hospital to conduct a randomized field experiment with approximately 1,200 individuals in rural China to study demand for diabetes and cancer tests. The field experiment has two designs: price treatments and disease treatments.

In the price treatments, we vary the price of a diabetes test. Individuals were randomly assigned to one of three groups: the free group (T0), the 10 RMB group (T10), or the 30 RMB group (T30). Individuals chose one of three sealed envelopes offered by enumerators, and the voucher inside the envelope stated the price they would have to pay to receive the diabetes test. We then asked whether they would like to take a diabetes test. If they chose to do the test, nurses from the local hospital drew their blood after the physical examination.

In the disease treatments, individuals were randomly assigned to one of two groups: the diabetes group or the cancer group. Randomization was conducted by the researcher using a computer, and individuals were not aware of their assignment. We provided the disease test for free after blood had been drawn for another free blood test (so there was no additional cost of taking the test), but varied the disease type to be tested, diabetes or cancer. In all treatments, we elicited individuals’ self-reported beliefs about their corresponding disease risk before they made their testing decisions so that we could investigate the heterogeneous effect across the disease risk.

We are interested in (1) what is the impact of different treatments on take-up of the screening test; and (2) who selected to be screened under different treatments. The key information necessary to understand question (2) is diabetes risk, which can be determined by both objective and subjective measures. Objective measures include test outcomes (which are only available for those who take the test). The subjective measure is self-reported beliefs about diabetes risk and cancer risk. We asked participants the following question: “What do you think is the probability that you have diabetes/cancer?” To indicate their answers, participants were given 10 small paper balls and asked to distribute them across two areas: (1) No diabetes/cancer and (2) have diabetes/cancer. If participants put 2 paper balls into (2) and 8 paper balls into (1), the perceived probability that they have diabetes/cancer is around 20%.

Research Findings

We surveyed 664 individuals, with a response rate of about 93%, in the price treatment and 531 individuals, with a response rate of about 96%, in the disease treatment. Our research findings are as follows:

First, by using a field experiment, we observe that both low- and high-risk individuals are less likely to take the test—a phenomenon not revealed before.

Second, we find that as the test price increases, the average test outcome remains the same but the dispersion of the outcome decreases, indicating that both low- and high-risk individuals select out of the test as price increase.

Finally, we found an interesting heterogeneity: the pattern in which high-risk individuals avoid the test is more salient when the test price is higher and when the disease is more severe.

How the tendency to avoid information varies across the probability of having the disease has important policy implications. The test is more valuable for high-risk individuals. Under simple neoclassical intuition, they are more likely to take the test in any case; but according to our empirical results and under the optimal expectations model, they are less likely to take the test. If the latter is true, proper interventions that target the high-risk group create higher welfare gains than traditionally thought. Also, new policies that target the group that attaches higher weight to anticipatory utility can be more effective than traditional policies.

Juanjuan Mengis a professor of Department of Applied Economics at Guanghua School of Management, Peking University. She earned her Ph.D. in Economics from University of California San Diego.

About the author

Professor Meng Juanjuan graduated from the Guanghua School of Management, Peking University, and then went to the University of California San Diego to pursue her PhD in economics. Her research interests include behavioral economics and behavioral finance, and her researches have been published in leading international journals such as American Economic Review, Management Science, Journal of Public Economics, International Economic Review, Journal of Development Economics, Games and Economic Behavior.