Experimental Design
Our cooperation partner registers all low-income households that participate in the energy efficiency assistance program in a program database and determines their eligibility for a refrigerator replacement voucher based on appliance age and estimated energy savings from replacement. Eligible households are part of our experiment and the computer randomizes households into one of seven experimental groups. Group assignment determines the version of the information letter that the household receives to inform about voucher eligibility and the steps required to redeem the voucher in cash. A control group receives the standard information letter without any of our interventions. The “Gain Frame”-group is informed about the energy savings from replacing the old refrigerator and sees the abstract image of a purse with money dropping into the purse. The “Loss Frame”-group is informed about forgone savings of not replacing the refrigerator and sees the purse icon with money dropping from the purse. In the “Simple Gain Frame”-treatment the information content of the “Gain Frame” is presented with additional graphical icons (a fridge, the purse, a happy smiley). In the “Simple Loss Frame”-treatment the information content of the “Loss Frame” is presented with additional icons (crossed-out fridge, the purse losing money, a sad smiley). In a “Checklist”-treatment we present the necessary steps to redeem the voucher in cash in form of a checklist. In a “Checklist and Simple Gain”-treatment we combine the checklist presentation, with the gain framing presented with icons. After having received the information letter, households decide whether they request the voucher and whether they replace their old refrigerator against a new, energy-efficient refrigerator. Both the request- and replacement-choice are the outcome variables of interest. Our hypothesis is that low-income households may lack the cognitive capacity to process the information presented in the letter, thus both the icons and the checklist are intended to simplify the information content. The hypothesis of the gain and loss framing follows classic behavioral economic research according to which loss framings enable greater behavioral change.