Impact Bias: Why Good and Bad News Fade Faster Than Expected

A forecast about feelings has two parts: how strong the reaction will be and how long it will last. Impact bias affects both, and the research finds the error larger in the second. Daniel Gilbert and Timothy Wilson named that part the durability bias: people are reasonably good at saying which way an event will move them and poor at saying when the movement will stop.

One reason is that a forecast is made about the event and nothing else. Asked how you would feel a year after moving to a new city, you picture the city. You do not picture the Tuesday in the new city when the dishwasher breaks, the commute, the work deadline and the phone call from your sister, all of which will be occupying the same mind. Wilson and Gilbert called this focalism. The focal event gets the whole stage in the forecast and a small share of the day in the life.

The second reason applies mostly to bad news. People do not anticipate how actively they will make sense of what happened. After a loss, explanations get built, meanings get found, the story gets retold until it is bearable. Gilbert and colleagues called this the psychological immune system and argued that its defining feature is that it works without being noticed, which is exactly why a forecaster leaves it out. The result is that negative events are survived faster than anyone predicts, and the person who recovers is often surprised by their own recovery.

Adaptation is real but it is not uniform, and the honest version of this page has to say so. Large panel studies find average life satisfaction returning to its earlier level within about two years of a wedding, but not returning fully after a divorce, after a long spell of unemployment, or after the onset of a serious disability. Impact bias is a claim about the gap between forecast and outcome, not a promise that everything is adapted to. Some events change a life, and the research can tell you which ones.

Daniel Gilbert, Elizabeth Pinel, Timothy Wilson, Stephen Blumberg and Thalia Wheatley published "Immune neglect: A source of durability bias in affective forecasting" in the Journal of Personality and Social Psychology in 1998. Across a series of studies they compared what one group of people predicted they would feel after an event with what a second group, who had actually been through it, reported feeling.

One study used university tenure. Current assistant professors were asked how happy they expected to be in the five years after a positive or a negative tenure decision, and predicted a large difference between the two. Former assistant professors who had actually been through the decision reported general happiness that did not differ reliably by outcome. Another study used romantic breakups, and compared separate groups of students rather than following anyone over time: on a 7-point happiness scale, the 36 students whose breakup had happened two months ago or more recently averaged 5.42, the 302 whose breakup was further back 5.46, and the 194 who had never had one 5.27, while those 194 forecast 3.89 for themselves two months after a breakup.

The oldest result in this literature is older than the term. In 1978 Philip Brickman, Dan Coates and Ronnie Janoff-Bulman interviewed 22 people who had won a major lottery prize within the previous eighteen months and 22 people from the same areas of the city who had not. Present happiness came out at 4.00 for the winners and 3.82 for the others on a 0 to 5 scale, a gap too small to be reliable, though the winners did report taking less pleasure in a list of ordinary activities. The sample was small and everyone was interviewed once, so the study cannot show how the winners changed over time, and it has carried far more weight in popular accounts than its design supports.

Later work with large national panels has filled that gap and complicated the picture. Richard Lucas and colleagues reported in 2003 that average life satisfaction rises around a wedding and returns to its earlier level within roughly two years. Erik Lindqvist, Robert Ostling and David Cesarini, following Swedish lottery winners for up to two decades, found sustained gains in life satisfaction years after a win, with little effect on day-to-day happiness or mental health. Both findings are consistent with the Brickman result on mood and inconsistent with the slogan it is usually quoted for.

Impact bias matters most where a forecast is doing the deciding.

Forecasts about feelings get better when they are checked against other people's experience rather than refined inside your own head.

• Ask someone who has been through it. In studies of affective forecasting, a stranger's report of how they currently feel predicts a forecaster's eventual experience better than the forecaster's own detailed imagining does.
• Forecast the day, not the event. Write out an ordinary Tuesday six months after the change, with the commute, the chores and the other people in it.
• Date your predictions. "On June 3 I expect to feel X about this" takes ten seconds and gives you a record that memory cannot revise.
• Expect to cope better than you think, and do not use that as a reason to accept something avoidable. The finding is about forecasting error, not about whether the loss matters.
• Treat recurring costs differently from one-off ones. The things people adapt to least well are the conditions that repeat every day, which is where a decision is worth more scrutiny than the headline event gets.
• Know which events the evidence treats as exceptions. Serious disability, long unemployment and bereavement show incomplete adaptation in panel data, and nothing here says otherwise.

Several biases concern the difference between an imagined future and the one that arrives.

• It is not optimism bias, which is expecting good things to be more likely for you than for other people. Optimism bias is about probability; impact bias is about how much a given event will change how you feel, and it inflates the forecast for bad events as well as good ones.
• It is not the focusing effect, which is the distortion that comes from attending to one feature of a situation while ignoring the rest. Focalism is one of the mechanisms behind impact bias rather than a separate prediction about duration.
• It is not projection bias, which is assuming your future preferences will match your current state, as when a full shopper buys too little food. That is about tastes carrying forward; impact bias is about the size and length of an emotional reaction.