The study, led by Roland Fryer, Steven Levitt (yes, of Freakonomics fame), John List (I think John List is famous!) , and Sally Sadoff, ran in Chicago Heights and tested two versions of the same financial incentive (Fryer et al., 2012). One group of teachers was told they'd receive a bonus at the end of the year if their students improved. A second group was paid the bonus up front, in September, and told they'd have to give it back if the improvement didn't materialize. The dollar amount was identical. The only thing that changed was the reference point: was this money already theirs, or was it still out there to be won? The loss-framed version raised math scores by 0.201 to 0.398 standard deviations, enough to functionally replace an average teacher with a considerably stronger one. A separate arm of the study, identical in every way except that it was framed as a standard forthcoming bonus, produced smaller and statistically insignificant gains.

The same mechanism shows up in charitable giving, not just earning. A 2023 study in PLOS ONE by Maximilian Späth gave participants a fixed pot of money to split between themselves and a charity, but varied the instructions (Späth, 2023). Some were told they could "give" a portion to charity. Others were told they could "take" a portion for themselves from money that otherwise belonged to the organization. Same pot, same options, same charity. When the decision was framed as taking, people kept less and the charity received significantly more. The frame changed where the money was assumed to sit before anyone touched it, and that assumption did most of the persuasive work.

Neither study is about doom or urgency. Nobody threatened the teachers with a closing school, and nobody told Späth's participants a nonprofit was about to shut its doors. The lever is quieter: which side of the ledger a piece of money starts on in someone's head, not how frightening the copy sounds.

Loss aversion is one of the cognitive biases I write about in Neurogiving, alongside anchoring and status quo bias, and this pair of studies shows why it belongs in that company rather than standing alone. A loss frame only works because it first moves the anchor: the teachers had to experience the money as already theirs before losing it could sting, and Späth's "taking" participants had to experience the charity's share as already spoken for before keeping it felt like subtraction. Loss aversion is the amplifier. Anchoring decides what counts as a gain or a loss in the first place.

That distinction matters for how appeals get written. A lot of fundraising copy manufactures urgency, closing wards, unmet need, a mission at risk, and calls it "loss framing." It can work, but through fear and vividness, not through the reference-point mechanism these studies isolate. Stay tuned for an upcoming podcast of episode of The When Bearing Witness Podcast with Maria Bryan, where we dive deep into the science of urgency, empathy, and compassion as well as the risks that it poses. The more ethical (and effective) version doesn't ask a donor to imagine a catastrophe. It quietly repositions what the donor already thinks of as theirs against what was already spoken for. That's a subtler, more honest tool than most of what gets built in its name, and easier to get wrong, since it depends on the donor accepting the reference point rather than rejecting it as manipulation.

Here's a specific thing to check, not a template to fill in. Pull your last three appeals and find the sentence that describes what the gift does. What does it assume about the money's status a moment before the ask: new money entering the picture, a gain the donor creates, or money that already has a claim on it, a status the donor is asked to honor or withhold?

Most annual fund copy defaults to the first frame without anyone deciding to. "Your gift provides..." treats the dollar as if it didn't exist until the donor produced it. Späth's study points to a different sentence nearby, one that treats the need as the default and the donor's decision as honoring it or withholding from it. That isn't the same as writing "your inaction will hurt someone," which is a threat. It's closer to "here is what already has a claim on this," a reframing of ownership rather than an appeal to fear.

Before rewriting anything, sit with a harder question: does your donor actually experience the beneficiary's need as a legitimate prior claim, the way Späth's participants experienced the charity's share as already allocated? If yes, the taking-versus-giving frame has real room to work. If your cause is one donors experience as pure discretionary generosity, forcing this frame onto it may just read as manipulative, because there's no believable prior claim to invoke.

Research Sources

In Lab Notes Pro

Lab Notes Pro goes deeper this week on how solid the underlying science actually is, including a global replication of prospect theory across 19 countries, a study showing loss framing doesn't always beat gain framing (sometimes an aspirational frame wins outright), real donation data on when a harder-edged frame wins volume but loses average gift size, and a systematic review of matching gifts and other donor incentives that explains why some "leverage" appeals work and others quietly erode trust. You'll also get the breakdown across five fundraising contexts, three specific ways organizations misuse this research, a real-time account of what a donor experiences when a reframed ask lands well or badly, and a runnable experiment you can build on your own file.