Negativity bias
Claim assessed: Negative events weigh more than positive ones of the same size.
Published · how we work
In short
Negativity bias means that bad things affect us more than good things of the same size. Losing 50 pounds upsets you more than finding 50 pounds pleases you. A criticism stays with you longer than a compliment. This doesn’t mean bad always wins. Good can make up for it, but it needs numbers: several good things for one bad one.
What it says
The general pattern. Baumeister, Bratslavsky, Finkenauer and Vohs (2001) gathered studies from very different areas and found the same pattern almost everywhere. In one daily diary study, a bad day still showed in well-being the next day, while a good day left no trace the day after. Unpleasant events affected both good and bad mood, while pleasant ones affected only good mood. Negative information about someone new weighs more in the final impression than positive information. The authors say they looked for areas where good was consistently stronger and found none. The closest exception was optimism about the future, but even that didn’t hold when people made concrete decisions.
Four forms. Rozin and Royzman (2001), in the same year, described the bias in four forms. A negative thing has more effect than an equivalent positive one. The negative grows more unpleasant as it approaches, faster than the pull of the positive grows. A mix of good and bad is judged worse than the sum of the two would suggest. And negative things are more varied and call for more kinds of response. They propose that the negative also dominates because it “spreads” from one thing to another more easily than the positive.
News, in 17 countries. Soroka, Fournier and Nir (2019) showed real BBC World news stories to 1,156 people in 17 countries on 6 continents and measured their physiological reactions: skin conductance and heart rate variability. On average, people were more activated by negative stories than by positive ones. The variation from person to person was large, though. Taken one by one, the bias was significant in only 9 of the 17 countries, by skin conductance, and the country explained under 3% of the differences between people.
Clicks, in 12,448 tests. Robertson et al. (2023) analysed the tests in which the site Upworthy showed readers different headlines for the same story, at random. The analysis was registered in advance, as a registered report. Across 12,448 tests, with over 205 million impressions, each extra negative word in a headline of average length raised the click-through rate by 2.3%. Each extra positive word lowered it by 1.0%. This was despite headlines containing slightly more positive words than negative ones.
The 5-to-1 ratio. It is the most quoted number linked to this bias. Gottman and Levenson (1992) filmed 73 married couples in 1983 while they discussed a problem in their relationship for 15 minutes. At every turn of speech, they counted positive behaviours, such as agreement or humour, and negative ones, such as complaining, criticism or defensiveness. The couples were split into two groups by this very balance: “regulated”, where the positive consistently outweighed the negative, and “nonregulated”. Four years later, the nonregulated couples had considered separating more often and had separated more often. Gottman and Levenson note that, in the regulated couples, the ratio was about 5. Baumeister et al. quote the number, from a 1994 book by Gottman, as a rule: in a relationship that works, good behaviours must outnumber bad ones at least five to one.
Example
At their annual review, an employee gets ten positive comments and one critical one. That evening they think almost only about the criticism. According to negativity bias, this isn’t a sign that they are unusually sensitive, but the ordinary reaction: a negative comment weighs more than a positive one of the same size. The example is an illustration, not a study cited here.
How to apply it
The steps below are a practical way to apply it that we propose, based on the studies cited.
- When you judge a day, a month or a person, count the good too. The bad comes forward on its own. If you want an accurate picture, you have to look deliberately for what went well.
- In a relationship, avoiding the bad matters at least as much as adding the good. Baumeister et al. conclude that a relationship depends more on not doing bad things than on doing good things. Good gestures help, but they don’t automatically erase a criticism or a defensive reply.
- When you give negative feedback, keep in mind that it will weigh more than you think. Say it clearly once, don’t repeat it, and put it next to what actually works.
- Watch what your clicks bring you. Negative headlines get more clicks, even for the same story. A news feed that seems ever gloomier may also reflect what you choose to open, not just the world.
- Don’t use 5 to 1 as a recipe. The number is an average observed in some couples who were doing well, not a tested target.
Limits and nuances
- The 5-to-1 ratio is a description, not a cut-off. Gottman and Levenson split the couples by the very balance between positive and negative and reported that, in the regulated couples, the ratio was about 5. The sample was 73 couples, 9 of whom divorced within the four years. The study didn’t test whether a ratio of 4 or 6 changes anything. Nor does it show that one negative behaviour “is worth” five positive ones; that is Baumeister et al.’s interpretation. We haven’t read Gottman’s 1994 book, where the number becomes a rule.
- The 2001 review is broad, not systematic. Baumeister et al. started from about 100 replies to an email sent to an association of psychologists, and say openly that they aimed for a variety of areas, not the strength of the evidence in each. Some examples, such as lottery winners compared with people paralysed in an accident, compare events whose size can’t be equated.
- The average hides large differences. In the 17-country study, many people didn’t react more strongly to negative news, and in some countries the average wasn’t significant. The authors conclude that positive news can find an audience too.
- There is an alternative explanation. Alves, Koch and Unkelbach (2017) propose that positive things are more alike than negative ones, because the positive tends to be moderate while the negative is extreme in different directions. Negative information would then be processed more because it is more varied, not necessarily because it affects us more strongly emotionally.
- The bias is weaker in memory. Baumeister et al. note that people sometimes remember pleasant things better, especially as time passes.
- The link with loss aversion. The example of money lost and gained resembles loss aversion, a disputed explanation for the endowment effect. See the endowment effect entry.
- What we read. Baumeister et al. (2001), from the copy on the Carlson School of Management site (University of Minnesota), Gottman and Levenson (1992), from Levenson’s lab site at Berkeley, and Soroka et al. (2019) and Robertson et al. (2023), from PubMed Central, we read in full. For Rozin and Royzman (2001) and Alves et al. (2017) we read the abstracts. We haven’t read Gottman’s 1994 book.
Sources
- John M. Gottman, Robert W. Levenson (1992). Marital processes predictive of later dissolution: Behavior, physiology, and health
- Roy F. Baumeister, Ellen Bratslavsky, Catrin Finkenauer, Kathleen D. Vohs (2001). Bad is stronger than good
- Paul Rozin, Edward B. Royzman (2001). Negativity bias, negativity dominance, and contagion
- Hans Alves, Alex Koch, Christian Unkelbach (2017). Why good is more alike than bad: Processing implications
- Stuart Soroka, Patrick Fournier, Lilach Nir (2019). Cross-national evidence of a negativity bias in psychophysiological reactions to news
- Claire E. Robertson, Nicolas Pröllochs, Kaoru Schwarzenegger, Philip Pärnamets, Jay J. Van Bavel, Stefan Feuerriegel (2023). Negativity drives online news consumption
See also: Availability heuristic, Clickbait (curiosity gap), Endowment effect and status quo bias, Framing effect, Loss aversion, Peak-end rule