Blood donation is a perfect example. Blood shortages in developing countries are very frequent and have consequences so severe to constitute a global public health concern (WHO, 2011 ); high-income countries also face temporary emergencies and seasonal shortages. With less than 10 percent of the population donating in wealthy countries, and much fewer in poor countries, increasing the blood supply remains an important public policy concern.
For over 40 years, many national blood agencies worldwide have embraced the World Health Organization guidelines that oppose the use of economic incentives to attract blood donors (WHO, 2009 ). The opposition has been based largely on two notions. First, paying donors could decrease altruistic motivation for donating and result in fewer donations than a purely volunteer system. Second, economic incentives may attract donors with greater risk of carrying transfusion-transmissible diseases such as Hepatitis or HIV, thus potentially compromising the safety of the blood supply.
Over the past 40 years many studies have found evidence that support these concerns. Blood donors are often mentioned explicitly, both in academic discourse and in the blogosphere, as the chief example of the perverse effect of incentives on intrinsically motivated activities (see, e.g., The Economist’s “Buttonwood’s Notebook” of July 20, 2012  or Steffie Woolhandler and Dan Ariely’s October 11, 2012 post on Health Affairs Blog ).
However my co-authors Nicola Lacetera  (University of Toronto), Robert Slonim  (University of Sydney), and I argue in a recent Science article that such opposition should be reconsidered based on new more relevant evidence (Lacetera, Macis and Slonim, 2013 ).
First we note that the initial evidence for the standard view initially came from uncontrolled studies using non-random samples, and then later from surveys and artificial scenarios in lab settings that use hypothetical questions. Bob, Nico and I then ask whether the detrimental effect of incentives that was detected in these studies would be confirmed in actual blood donation situations.
We review an extensive body of very recent field-based evidence and natural field experiments that used large, representative samples, and the results clearly refute the previous findings: economic rewards have a positive effect on donations without negative consequences on the safety of the blood collected. This evidence, which includes our own extensive field studies, derives from studies examining the effects of actual incentive programs for actual blood donors in Switzerland, Italy, the United States and Argentina. Of the 19 incentive items examined in these studies using field-based methods, 18 had a positive effect. The incentives ranged from mostly small valued items such as t-shirts and $5 gift cards to a paid day off work. In addition, when researchers were able to compare different incentive levels in the same study, the effects were larger for items of higher economic value. The only incentive item that had no effect was a free cholesterol test (Goette et al. 2009 ); interestingly, free cholesterol tests and health checkups were among the most popular items according to the earlier studies that used survey data (Sanchez et al 2001 ; Glynn et al. 2003 ). Thus, in the context of blood donation, what people say they would do when faced with hypothetical incentives does not match what they actually do when faced with real-world incentives for actual blood donations.
We also find, in studies that could observe the safety of the donation as measured by whether donors were ineligible to donate due to risk factors (e.g., travel to malaria regions, promiscuous sexual activity, etc.) or whether the blood was later discarded because it tested positive for some transmissible disease, nothing to suggest a change in the usability of the blood (Goette and Stutzer 2008 ; Lacetera, Macis and Slonim 2012 ; Iajya et al. 2012 ). Thus, in contrast to the earlier evidence, the field-based evidence indicates that economic incentives can increase the blood supply without any negative consequences for safety.
Perhaps one of the most intriguing questions is why the previous evidence suggested rewards would diminish supply. There are many differences between the earlier approaches and the new evidence that can explain the radically different results. The most striking differences are methodological. The field-based studies we survey all use representative populations of actual blood donors (or non-donors in some cases) and observe actual blood donations as opposed to elicited responses to hypothetical scenarios. Moreover, in contrast to studies based on survey data, laboratory experiments, and framed or artefactual field experiments, in the observational field studies and the natural field experiments the donors were not aware that they were taking part in a study, and thus did not feel scrutinized by the researcher (Levitt and List, 2008 ; List, 2008 ). As a result, they might have been less concerned about their image and more excited about potential rewards. Moreover, past research often focused on getting paid cash to donate, whereas in the studies we surveyed the rewards were not framed as “payment” and thus they might have been seen as tokens of appreciation which can reinforce rather than undermine donors’ intrinsic motivation. Also, the rewards are typically provided merely for presenting at the blood drives, not for donating blood, which should reduce the risk that an ineligible donor might misrepresent health or other information.
A look at two early field experiments, Upton (1973)  and Ferrari et al. (1985)  further highlights the importance of carefully considering design details when conducting field studies and interpreting their results. Upton’s study found that subjects who had not donated blood in the past were more likely to donate when offered compensation, while the incentive had the opposite effect on those who had previously given blood. However, interpreting these findings is complicated by the fact that the researcher mentioned compensation when he recruited individuals to sign up for the study, which might have affected the selection of subjects into study participation. Ferrari and coauthors found large, positive effects of incentives. However, first, this study had a very small sample (N=80) compared to the thousands to tens of thousands of individuals in all of the recent field studies we surveyed. Second, the sample was not representative of the broader blood donor population and instead used a sample of college students. Third, the paper describes that ten female solicitors approached the 80 subjects on campus, but does not explain how the randomization was achieved (i.e., how the solicitors chose who to offer and not to offer incentives to). Fourth, the solicitors must have been aware of the purpose of the study (since they solicited subjects for both treatment and control). These factors raise potential concerns that do not exist in any of the other studies that we report. Finally, in this study there is no report on the subjects presenting or making an actual blood donation.
These (and other) methodological issues were likely not discussed in the 1970s and 1980s but are now considered quite important for empirical investigation, and my coauthors and I believe are essential for our confidence in the more recent evidence. Policies should rely on evidence, and the most relevant empirical evidence shows positive effects of offering economic rewards on blood donations. One immediate implication for policy is that guidelines prohibiting any economic reward should be reconsidered. More studies should be conducted, especially in poor countries, where many people die unnecessarily due to the lack of blood for transfusion. Of course, safety is a paramount concern, especially in resource-constrained countries. Advances in screening technology have greatly reduced the risk of tainted or otherwise unusable blood being used later in transfusions, although such advances are not widely available in poor countries yet. Ethical considerations should of course play a role, but evidence needs to be considered as well.
Of course, different methods have different virtues (including for instance laboratory studies offering tight controls and reducing potential participation risks and ethical concerns in some situations) and ideally should be complementary to each other. For instance, when estimates of the same parameter differ between field and laboratory studies, researchers can explore the differences and reconcile the results in the lab (Kessler 2013  provides a clear statement and demonstration on this point). Nonetheless, the contrasting results between the field-based studies and those relying on hypothetical scenarios, or based on framed experiments, indicates that the methods of inquiry matter a great deal for the results.
Mario Macis  (PhD, University of Chicago) is an Assistant Professor of Economics at Johns Hopkins University, Carey Business School and a Research Fellow at the Institute for the Study of Labor (IZA). His research interests are in labor and health economics, and he teaches classes in microeconomics, personnel economics and human resources management.