What happens when people refuse to update their beliefs?


This page in:

Last week I wrote about “treatment as prevention.” Because being treated by a combination of ARV drugs effectively prevents the transmission of HIV from an infected person to his (her) uninfected partner, the idea is that if we were to test as many people as possible, find out who is infected, and offer them ARVs, we could make significant headway in preventing the spread of HIV. In other words, test and treat.

I offered my misgivings about this here, which had mainly to do with cost-effectiveness and offering people a drug that we have yet to convincingly show helps them (rather than their current and future sexual partners). However, it seems that I might have missed something much more basic. What if the people who are tested (and informed that they are HIV-positive) do not believe that they are actually infected with the virus?

Take a look at the figure below and read on:



As part of a study in Malawi, we have tested a sample of young women for HIV using rapid test kits during home-based voluntary counseling and testing sessions. These door-to-door sessions are a very effective method of getting people tested and informing them of their statuses as the refusal rates are very low and the rapid testing procedures are almost 100% accurate.

(The study followed the HIV testing algorithm described in the Malawi Ministry of Health guidelines. If you are interested in the details of testing procedures, you can find them at the end of this post.)

Some months later, different survey teams visited the same individuals and asked them about their subjective beliefs on their chances of being infected with HIV. The figure above shows the distribution of their answers to the question on the likelihood of being currently infected with HIV (on a scale of 1 to 10) among those who had tested positive for HIV during the VCT session, meaning that these respondents were informed that they were HIV-positive by a Malawian nurse, given post-test counseling, and referred to a health clinic.

  • Only 45% of the respondents thought that they are definitely infected with HIV.
  • Almost as many young women (39%) said that there was ZERO chance that they were infected with HIV.
  • The remaining 16% thought their chances of being HIV-positive as somewhere between 30 and 50 percent.

What is almost as perplexing is that we cannot find any baseline characteristics (such as household wealth or highest grade completed) or current indicators (such as achievement scores in math, language, or cognitive skills) that explain the variation in the subjective beliefs among this group of HIV-infected individuals.

Before we get to “test and treat,” we may first need to learn how to “test and convince.”

P.S. After I posted about the possibility of paying people to undergo circumcision for HIV prevention, the debate on banning circumcision in the U.S. has taken off in the Washington Post, the New York Times, and the blogosphere (although it seems that the Santa Monica effort to ban has been abandoned). I guess I know how to call them: I say "subsidize", they say "ban"!

P.P.S. You can find some more detail below about the HIV testing procedures discussed above.

Home-based voluntary counseling and testing was conducted by Malawian nurses and counselors certified in conducting rapid HIV tests through the Ministry of Health HIV Unit HCT Counselor Certification Program. Pre-test counseling, post-test counseling, and referrals to health clinics were conducted according to the study guidelines approved by the Malawian National Health Sciences Research Committee (NHSRC).

The collected blood sample was first tested using a Determine HIV/1-2TM assay (Inverness Medical, UK). If the test result was positive, then the sample was tested using a Uni-GoId® HIV assay (Trinity Biotech, Ireland). If these two test results were discordant, then the sample was tested with SD BIOLINE HIV 1/2 3.0 assay (Standard Diagnostics, Inc., Korea) for a tie-breaker. Participants who tested positive on the Determine plus either of the other two assays were interpreted to be HIV infected; those testing negative on the Determine or negative on the other two assays were interpreted to be HIV uninfected. Recognizing that HIV-negative participants could be in the window period, they were counseled to be retested after 3 months if they had any risk factors for infection.

All participating individuals provided informed consent at all stages of the study. Additional consent was obtained from parents or legal guardians of all unmarried girls under the age of 18. Informed consent for HIV, HSV-2, and syphilis testing was conducted separately at the time of biomarker data collection.




Berk Ozler

Lead Economist, Development Research Group, World Bank

Join the Conversation

Jason Kerwin
June 08, 2011

The same pattern exists in the MDICP data. This "I don't believe the results" effect gets at the mechanism behind the Thornton (2008) finding that learning one's HIV status has a negligible effect on prevention behavior. (http://www.aeaweb.org/articles.php?doi=10.1257/aer.98.5.1829)

Berk Özler
June 08, 2011

Thanks Jason,

I'll go back to take a look at Rebecca's results with a fresh set of eyes.


June 08, 2011

Maybe they don't believe the results or maybe they don't want to display their status to a stranger.

Berk Özler
June 09, 2011

Thanks. I think it is important to remember that the individuals are not being asked to report their HIV status. An anonymous enumerator is asking them (confidentially) about the likelihood that they are infected with HIV. It's hard for me to speculate about the pattern of misreporting on subjective probability of being HIV-positive.

Justin White
June 08, 2011

Being HIV-positive confers a lot of stigma in Malawi. Are you sure these people are reporting their true subjective beliefs. Life might be a lot easier, in the short run, if they live in denial or keep it a secret. Plus, if I were going to be lie about my HIV status, I'd probably lie with certainty, consistent with the heaping at zero.

You may just be dealing with a measurement problem. One way to investigate prospectively might be to ask about other sensitive behaviors or beliefs you could confirm and then see if subjects are honest about those.

Berk Özler
June 11, 2011

Hi Michelle,

Thanks for these very thoughtful comments. It's great that our infant blog has been able to gather such a nicely multidisciplinary audience!

I agree that denial is a likely scenario in this case -- especially during a survey. The question is whether this denial extends to real-life situations. It's OK if people are misreporting to enumerators as a coping mechanism. But, if they'd refuse a follow-up visit to a clinic, or treatment, or change behavior with their partners, that is much more serious for them and from a public health perspective.

I'll take a look at the paper by Peters -- thanks for the link.


Berk Özler
June 11, 2011

Slate Magazine is running this story by Sonia Smith (TexasMonthly.com), where it took five years for a women who was tested positive to visit a medical doctor: http://www.slate.com/id/2296486/entry/2296489/

Michelle Poulin
June 11, 2011

Hi Berk,

These results are intriguing. Thanks for posting (and the blog!). In my view, that such a high percentage of respondents report zero likelihood of being HIV positive isn't all that surprising. Why would a young woman in poor Malawi want to disclose to a stranger she has HIV? And this isn't just any stranger--this is an enumerator, who, as an employee of Chancellor College and of the WB, holds a higher social standing and receives income. Instead, the young woman has HIV, and thus the expectation of a shortened life expectancy. Or, this young woman, instead of planning for marriage and children or a job or more school (as her negative friends are doing) is thinking about how to manage the possibility of remaining unmarried, or of not having children, or of keeping her future husband or future children HIV negative, or about at what point she'll need to begin ARVs. This is stressful, and she may not be ready to deal with this reality. When asked by an enumerator about her likelihood of being HIV positive, she says zero.

It is really odd that no background characteristics predict those who "admit" to being HIV positive compared to those report zero likelihood. Perhaps this is due to some unobserved coping mechanisms, attributable to personality?

Pauline Peters and colleagues have an interesting paper that, in my opinion, offers a far more informative interpretation of how we (Western researchers) think of stigma. Their basic point is that it's not so much stigma that keeps people in denial, but rather it's people's attempt to simply live normal lives. This makes lots of sense when hardship is as pervasive as it is in southern Malawi.

The paper: http://journals.cambridge.org/action/displayAbstract?fromPage=online&ai…