Several issues arise in all the above settings for raising the SNAP question when people apply for health coverage. First, Marketplaces need to decide which consumers will be presented with the option of seeking SNAP. In theory, a Marketplace could offer all applicants for health coverage the opportunity to connect with SNAP. In practice, it will probably be more effective to limit this opportunity to consumers who are likely to qualify. Raising the issue of SNAP with Marketplace applicants whose income is clearly too high to qualify could trigger questions and negative reactions from consumers who may be sensitive to being characterized as potentially eligible for public benefits.84 It could also lead to opposition from Marketplace officials anxious to maintain a strong distinction between traditional means-tested programs and the Marketplace, which targets consumers of all income levels. Further, it would be an inefficient use of SNAP agency resources to contact consumers who are highly unlikely to qualify for SNAP.
Second, it could be important for Marketplaces to link to SNAP records so that the opportunity to submit a SNAP application is presented only to people who are not already receiving SNAP or do not have a pending application. Otherwise, consumers who already participate in the program or who recently applied could become concerned about the status of their SNAP eligibility or could otherwise become needlessly anxious about their situation.
Third, the way that SNAP is framed is likely to influence consumer responses. Much research suggests the importance of even background visual imagery in priming various responses and influencing behavior. One study, for example, compared two groups’ responses to a web site sequence that began with an explanatory web page featuring certain background imagery followed by a neutral web page in which experimental subjects could choose between a sofa that offered the advantage of greater comfort and a sofa that offered the advantage of lower cost. For the group where the background imagery on the introductory page featured fluffy clouds, “designed to prime comfort,” only 38.7 percent preferred the cheaper sofa. Another group was shown an explanatory web page with background imagery consisting of pennies embedded in a green background, “designed to prime price.” In that second group, 55.8 percent preferred the cheaper sofa—a 44 percent relative increase, compared to the first group. Similar results occurred when subjects were offered a choice between safer and cheaper cars, with background imagery priming those two concerns.85 If SNAP is presented along with images of appealing food, for example, or imagery that reinforces a parent’s role as provider for children, consumers may be more likely to seek assistance.
Higher responses could likewise result if receipt of SNAP is framed as a potential loss (“don’t miss out on this chance to get help paying for food”) rather than a potential gain (“get help paying for food”). Behavioral economics research shows that behavior can be more powerfully affected by the risk of loss than the possibility of gain, even if the two are identical in reality but only described differently. A summary of this area of research noted that “people typically require a potential gain of at least $100 to make up for exposure to a potential loss of $50 because the subjective impact of losses is roughly twice that of gains.”86
One illustrative study analyzed the effects of showing 15-minute videos about mammography to women who had not received the recommended number of procedures in the past. Two groups saw different versions of the same video. One version, entitled “The Benefits of Mammography,” described mammography’s gains. The other, called “The Risks of Neglecting Mammography,” provided the same information, framed in terms of the harm that could result from missing a mammogram. In their immediate responses to the videos, the two groups did not differ in what they learned or how much they liked the videos. However, 12 months later, 51.5 percent of those seeing the first video had received a mammography, compared to 66.2 percent in the second group.87 Reframing from gain to potential loss was thus associated with a 20 percent relative increase in the likelihood of receiving a mammogram 12 months after seeing a brief educational video.88 As with the earlier study showing the impact on physician choices when cancer treatment results were described in terms of mortality rather than survival, this example illustrates the effect of framing, as opposed to factual content, on decision-making.
Fourth and more broadly, program administrators could consider field-testing the many options described above. Such testing could investigate optimal timing for posing questions about SNAP as well as the language and visual images that evoke the greatest responses from consumers. For example, field-testing could assess the impact of opt-out language like the following: “It looks like you might be eligible for help paying for food. I assume you'd like to have the state food agency contact you to see if you qualify. If that's right, what's the best way for them to reach you?”
It could also investigate the impact of language suggesting that most people in the consumer’s position agree to such contacts; much behavioral economics research suggests that descriptive norms powerfully influence behavior. For example, one study asked women, if they were to contract breast cancer, whether they would be willing to add chemotherapy to hormonal treatment. Two groups of women were told that adding chemotherapy would raise five-year survival rates by only 1 percent. Among those who were also informed that “a few women” in such a situation chose chemotherapy, 26.6 percent said that they too would make that choice. By contrast, 35.9 percent preferred chemotherapy among those who were instead told that “most women” added chemotherapy to hormone treatment—a 35 percent relative increase.89
Returning to the issue of how to structure the link from health coverage to human services programs, many of the policy choices discussed above ultimately involve empirical questions about which approach will prove most effective. As suggested earlier, randomized controlled experiments, subject to strong ethical safeguards, could go beyond informal “market testing” to rigorously assess the impact of particular methods for giving health applicants an opportunity to seek SNAP benefits.
In all of this analysis, it is important to remember the heterogeneity of low- and moderate-income consumers. People vary on every important dimension—tolerance of risk, speed of cognitive overload, the kinds of stimuli that create cognitive overload, preferences for receiving and giving information in various ways, total cognitive capacity at particular points in time, and more. The most effective system for connecting health applicants to SNAP would allow different types of consumers to find the channel of assistance best suited to meeting their needs.