Skip to main content
Category: Effects
Type: Preferring the move that counts as doing something
Origin: Anthony Patt and Richard Zeckhauser, Journal of Risk and Uncertainty (July 2000); Michael Bar-Eli, Ofer H. Azar, Ilana Ritov, Yael Keidar-Levin, and Galit Schein, Journal of Economic Psychology (October 2007)
Also known as: A preference for the normal action
Quick Answer — Action Bias is a preference for the move that counts as doing something, when staying still has the higher recorded success rate. Patt and Zeckhauser gave that name in 2000 to a penchant for action carried into a choice where the usual reasons for acting do not apply. Bar-Eli, Azar, Ritov, Keidar-Levin, and Schein counted one such gap in 286 penalty kicks that reached the goal: goalkeepers stayed in the center on 18 kicks and stopped 6, and jumped on 268 kicks and stopped 36.

​
What is Action Bias?

Action Bias is a preference for the move that counts as doing something, when staying still has the higher recorded success rate, or when the usual reasons for acting do not apply to the choice in front of the person. A goalkeeper who dives, and the same goal after he has stayed on his line, are the picture. The ball is in the net either way. The dive is the part a spectator can see. A coach who has just lost 4–0 and then fields three new players is the same picture in a different job. The next match can still be lost 3–0. The substitution is what looks like a response.
Individuals have a penchant for action, often for good reasons. But action bias arises if that penchant is carried over to areas where those reasons do not apply, hence is nonrational.
That is Anthony Patt and Richard Zeckhauser in 2000. The penchant itself can be useful. The bias, on their definition, is the penchant in a place where those reasons are missing. Status Quo Bias is a preference for the state already in place. Default Effect is the higher rate at which a pre-selected option is kept when doing nothing keeps it. On this page the normal move is the jump, and staying is the rare cell. Loss Aversion is a greater weight on a loss than on a matched gain. Patt and Zeckhauser name it as one relative of the penchant. Outcome Bias is a higher rating of a decision once the result is known to be good. Here the result held up for comparison is a goal, and the difference on the page is the dive.

​
Action Bias in 3 Depths

  • Beginner: Take one penalty that ends in a goal after a dive, and one that ends in a goal after the goalkeeper stays. Write down what is the same, and what a spectator can see.
  • Practitioner: Put the success rate of the normal move next to the success rate of staying still, and print how many times anyone stayed. In the 286 kicks, staying stopped 6 of 18. Jumping stopped 36 of 268.
  • Advanced: The 18 stays are the whole of the still cell. They were not assigned by a coin toss. A later count of World Cup kicks, described below, reports similar stopping chances for jumping and for staying. A separate study of coaches measures who is said to feel more regret. It does not measure who wins the next match.

​
Origin

The name on this page comes from Patt and Zeckhauser in 2000. The counted kicks come from Bar-Eli and coauthors in 2007. A 2002 study of coaches shows when people say inaction hurts more. That study is a judgment about feelings. Anthony Patt, then at the Belfer Center at Harvard, and Richard Zeckhauser, at Harvard, published “Action Bias and Environmental Decisions” in the Journal of Risk and Uncertainty, volume 21, number 1 (July 2000), pages 45–72. The lines above are from the abstract. The rest of the abstract is the study they ran. They used surveys of hypothetical environmental decisions. Apart from what an agent might do to look busy, people like to affect an outcome when a gain is reaped. Given a chance to help one of two sites, decision makers chose to foster improvement rather than prevent deterioration. The frame made it arbitrary which site was improved and which was preserved. They also report what they call strong action bias: choosing to reap gains even though losses must be imposed. They relate these patterns to loss aversion, status quo bias, omission bias for losses, and what they call bright-line behavior. This page does not carry a sample size or a percentage from that paper. The article PDF was not re-read past the abstract. Michael Bar-Eli, Ofer H. Azar, and Yael Keidar-Levin were at Ben-Gurion University of the Negev. Bar-Eli and Galit Schein were also at the Wingate Institute. Ilana Ritov was at the Hebrew University of Jerusalem. Their article, “Action Bias among Elite Soccer Goalkeepers: The Case of Penalty Kicks,” is in the Journal of Economic Psychology, volume 28, number 5 (October 2007), pages 606–621. It was received on 27 January 2006, revised on 28 November 2006, accepted on 5 December 2006, and posted online on 25 January 2007. The counts below are from the authors’ manuscript, MPRA Paper 4477, posted on 15 August 2007. They label it an early version of that article. The published abstract states the same 286 kicks, the same claim that staying in the center is the better strategy given how kicks were aimed, and the same observation that goalkeepers almost always jump. The journal PDF was not re-read cell by cell. They use the name for a bias in favor of acting. They reserve omission bias for a bias in favor of not acting. Omission Bias, in the work they cite, is a harsher judgment of harm that comes from acting than of the same harm from not acting. This page does not carry those experiment counts. Their own prediction flips the feeling when action is the norm. Daniel Kahneman and Dale T. Miller, “Norm Theory: Comparing Reality to Its Alternatives,” Psychological Review, volume 93 (1986), pages 136–153, is the account they use. A bad outcome feels worse when a better one is easy to imagine. An action is, in the usual case, easier to undo in the mind than a failure to act. When the norm is to act, the easy picture reverses. Standing still becomes the abnormal cause. The usual regret finding is the one they, and a later paper, both quote from Kahneman and Tversky (1982, p. 173). Paul kept his shares in company A. George switched from company B to company A. Each would have been $1,200 better off the other way. Both papers say a large majority thought George, who acted, would feel more regret. Neither paper prints that majority as a percentage. The 1982 chapter was not re-read for this page. Marcel Zeelenberg, Kees van den Bos, Eric van Dijk, and Rik Pieters, “The Inaction Effect in the Psychology of Regret,” Journal of Personality and Social Psychology, volume 82, number 3 (March 2002), pages 314–327, keep that scenario’s logic and add a prior match. Their claim is that a prior loss can make action the normal next step, so the coach who does not act is the one said to feel more regret. The case below is the kicks. The coach counts sit in the key points, because they are ratings of a story, not saves.

​
Key Points

Four comparisons keep the result fixed and move the action. A stopping rate needs the number of stays. A regret rating is not a league table. A feeling scale that piles up at the top cannot separate the dive from the stance. A rate computed from today’s kicks is a rate for today’s kicks.
1

Put the stopping rate next to the number of stays

In the 286 kicks that reached the goal, goalkeepers stayed on 18 and jumped on 268. The 18 stays stopped 6 balls and let in 12. The 268 jumps stopped 36 and let in 232. The chi-square comparison of those two stopping rates is 5.33, with N = 286 and p = 0.02. The table prints the stay rate as 33.3 percent and the jump rates as 14.2 percent to the left and 12.6 percent to the right. The 33.3 percent is 6 stops in 18 stays. It is the whole of that cell.
2

Let a prior loss change what looks normal

Zeelenberg and coauthors asked 165 students at Leiden University, 54 men and 111 women, who were paid, to read about two coaches. Steenland fields three new players. Straathof leaves the team as it is. Both teams then lose 3–0. When the prior match was absent, 41 of 55 said Steenland would feel more regret, and 14 said Straathof. Chi-square (1, N = 55) = 13.26, p < 0.05. When both had won the prior match 4–0, the split was 50 and 5. Chi-square (1, N = 55) = 36.82, p < 0.05. When both had lost the prior match 4–0, the split reversed: 18 named Steenland, and 37 named Straathof. Chi-square (1, N = 55) = 6.56, p < 0.05. These are answers about a written coach. They are not results of a match that was played.
3

Read the feeling scale with its ceiling

A separate questionnaire went to professional goalkeepers in Israel. Thirty-two of 36 answered. Twenty-five of the 32 ranked a jump as more normal than staying. Seven ranked staying as more normal. Chi-square (1, N = 32) = 10.13, p = 0.001. On a 1-to-10 scale of how bad a goal would feel, 15 of the 32 gave a 10 to the dive right, the dive left, and the stay. Two more gave the same score, below 10, to all three. Of the 15 who separated them, 11 said the stay would feel worse than either dive, and 4 said the opposite. Chi-square (1, N = 15) = 3.27, p = 0.07. A one-tailed binomial probability for 4 or fewer out of 15, at a fair coin, is 0.059. The authors call this compatible with their hypothesis. It does not clear 5 percent.
4

Hold the rate inside the kicks that were taken

The authors’ own limit is the aim of the kicks they filmed. Staying has the higher stopping rate in that distribution. They write that if goalkeepers always stayed, kickers would aim at the sides, and staying would stop being the better choice. They leave the mixed-strategy equilibrium, in which each side mixes, for later work. Rishav Singh and Sumitava Mukherjee, Journal of Economic Psychology, volume 113 (March 2026), article 102886, look at 292 on-target penalty kicks from men’s World Cup matches, 1982 through 2022. Their abstract says goalkeepers still jumped more often than they stayed, that the chance of a stop was statistically similar for the two decisions, and that more kicks went to the sides than to the center. They conclude that jumping was not a suboptimal action in that sample. This page does not carry a percentage from their article. The PDF was not re-read past the abstract.

​
Applications

Use the still cell as a line you can read. Keep a regret story in the study that asked for feelings. Keep a stopping rate in the kicks that produced it.

Write the still cell beside the normal move

Before you copy a dive, a substitution, or a new plan, write how often the still option was used and what happened then. In the filmed kicks, the still option is 18 stays, with 6 stops. The normal move is 268 jumps, with 36 stops. The chi-square of 5.33 is a comparison of those two rates in this archive.

Separate a shared rating from a real gap

On the Israeli questionnaire, 15 of 32 goalkeepers gave the worst rating, 10, to a goal after every choice. That ceiling cannot show a preference. The 11-to-4 split sits in the other 15, with p = 0.07. The norm question is the sharper result: 25 of 32 called the jump more normal than the stay, p = 0.001.

Keep a coach story off the league table

After a written 4–0 loss, 37 of 55 Leiden students said the coach who left the team unchanged would feel more regret. After no prior match, 41 of 55 said the coach who made three changes would feel more. Both answers are about feelings in a vignette. Neither answer is a goal difference.

Name the kick distribution the rate uses

The 33.3 percent is the stopping rate while staying, given the aims in these 286 kicks: 92 to the left, 82 to the center third, and 112 to the right, from the goalkeeper’s view. The authors say a goalkeeper who always stayed would face different aims. Singh and Mukherjee’s World Cup abstract reports a sample in which the two decisions had similar stopping chances.

​
Case Study

The kicks come from televised matches. The instructions to the judges say the games were played in Israel and worldwide during the preceding years. The abstract says top leagues and championships worldwide. Three independent judges marked each kick on a nine-cell diagram of the goal. Directions below are from the goalkeeper’s perspective. The center is cells 4, 5, and 6 of that diagram, a band across the middle, not a single point on the line. The judges started with 311 kicks. They set aside 18 that hit a post or the crossbar or missed the goal, because those balls are neither saves nor ordinary goals. They set aside 7 more on which the three marks were not even on adjacent cells, or all three differed. When two judges agreed and the third chose an adjacent cell, they kept the majority cell. The analyzed set is 286 kicks that reached the goal. The question is which choice stopped more of those kicks. The ball went left 92 times, to the center band 82 times, and right 112 times. Goalkeepers stayed on 18 kicks, jumped left on 141, and jumped right on 127. That is 6.3 percent staying and 93.7 percent jumping. In the full 311, before those exclusions, the shares were 7.1 percent staying, 44 percent jumping right, and 48.9 percent jumping left. The authors discuss the 286. The joint counts show the choices were not a late read of a ball already in flight. A stay met 1 kick to the left, 10 to the center, and 7 to the right. A jump left met 54, 41, and 46. A jump right met 37, 31, and 59. The goalkeeper’s choice matched the kick on 43 percent of the 286, counting a stay against a center kick as a match. A random choice among three directions would match on 33.3 percent. Always jumping to the right, the most common aim, would match on 39.2 percent. The authors read the gap over those benchmarks as a noisy cue, not as a goalkeeper who watches the ball and then picks a side. They cite a flight time of about 0.2 to 0.3 seconds from Chiappori, Levitt, and Groseclose (2002) and from Palacios-Huerta (2003). Those two papers were not re-read here. Bar-Eli and coauthors also write that in the Chiappori data the goalkeeper stayed in the center on about 2 percent of kicks, so that paper’s check of mixed-strategy play is mostly right against left. Palacios-Huerta, as they describe him, analyzes a game of right and left only. Stops follow the match of directions, with a large difference in the center band. The table prints 29.6 percent stopped for the 54 kicks that went left and met a jump left, 60.0 percent for the 10 that stayed in the center and met a center kick, and 25.4 percent for the 59 that went right and met a jump right. A jump still stopped a few balls aimed at the center band: 9.8 percent of the 41 center kicks met by a jump left, and 3.2 percent of the 31 met by a jump right. The other off-direction cells print 0.0 percent. The text says the 141 jumps left stopped 20 balls. Combining both jump directions gives 36 stops and 232 goals. The 18 stays give 6 stops and 12 goals. The chi-square test of equal stopping rates is 5.33, with N = 286 and p = 0.02. The overall row of the table is 14.2 percent to the left, 33.3 percent in the center, 12.6 percent to the right, and 14.7 percent for all 286. The paper also says that in their data about 80 percent of penalty kicks resulted in a goal. That sentence uses a denominator this page does not rebuild. The 14.7 percent is the stopping rate inside the 286. Their reading is that, given these aims, the choice with the higher stopping rate is to stay, and that goalkeepers almost never take it. Probability matching would have them stay about as often as kicks arrive in the center, 28.7 percent. They stayed on 6.3 percent. The authors set probability matching aside on that gap. The account they propose instead is the norm. On that account, a goal after a jump can be told as bad luck, and a goal after a stay can be told as having done nothing, where the visible move is to dive. The questionnaire tests that telling. It was written in Hebrew. Half the goalkeepers received the answer order reversed. The list was the two goalkeepers on each of the 12 teams in the Premier League, the top league in Israel, and the first-choice goalkeeper on each of the 12 teams in the National League, the second league. That list has 36 names. Thirty-two took part. Asked which action is most normal and which is least probable, 16 ordered the choices right, then left, then center. Nine ordered them left, right, center. Five ordered them center, left, right. Two ordered them center, right, left. No one put the center between the two jumps. Collapsing right and left, 25 of 32 treated a jump as more normal than a stay. The feeling question, in an important match, did not clear 5 percent, as the key points record. The lesson is a real archive in which the rare choice stopped a larger share of the balls it faced, beside a profession that treats the common choice as normal. The boundary is the design. Nobody assigned a goalkeeper to stay. The center is a band, and a jump can still reach part of it. Eighteen stays are a small cell, and the p-value of 0.02 is the comparison of 6 stops in 18 with 36 stops in 268. The authors say the advantage depends on kickers aiming as they did in this footage. Singh and Mukherjee’s later World Cup count, in the same journal, reports similar stopping chances for the two decisions. The Israeli feeling scale is a rating of a goal already scored. It is not a count of saves.

​
Boundaries and Failure Modes

Action Bias is the wrong name when a rate from 18 stays is retold as a rule for every penalty, when a vignette about regret is retold as a win, and when a preference for the option already in place is given this page’s name. The first boundary is the footage. The advantage of staying is computed from the aims in these 286 kicks. The authors write the equilibrium point themselves: if staying became the constant choice, kickers would send the ball to the sides. They also write that they are not estimating the mix both players would accept. The 18 stays were the kicks on which a goalkeeper chose to stay. They are not 18 kicks assigned to a stay. A goalkeeper who stays may be facing a different kicker, a different score, or a different cue. The match of directions, 43 percent against a random 33.3 percent, is their evidence that some cues are present. The chi-square of 5.33 does not remove that selection. In the full set of 311, stays were 7.1 percent, close to the 6.3 percent in the analyzed set, and that larger set is not a fix for the small cell. The second boundary is the other archive. Singh and Mukherjee’s abstract covers 292 on-target kicks in men’s World Cup matches from 1982 through 2022. Goalkeepers again jumped more often than they stayed. The abstract says the chance of stopping the ball was statistically similar for jumping and for staying, and that more kicks went to a side than to the center. Their conclusion is that the high rate of jumping in that sample is not an action bias. This page has no cell counts from their tables. A World Cup shootout and a televised league penalty are different lists. One abstract does not erase the 6 stops in 18. It does block any sentence that treats 33.3 percent as the save rate of standing still in every competition. The third boundary is the export. The Leiden split, 37 of 55 after a 4–0 loss, is a judgment of who feels worse in a story about two coaches and a Sunday match. Zeelenberg and coauthors also ran the same story with a prior win and with no prior match, and the majority ran the other way. Escalation of Commitment is a further decision to continue because of what has already been spent. The coach item does not measure that spend, and it does not measure the next result. Patt and Zeckhauser’s environmental surveys are a different operation: a stated preference for causing an improvement rather than preventing a deterioration, in a frame that swaps the two. Their percentages are not on this page. A clinic, a trading desk, and a ministry do not inherit the 33.3 percent.

​
Common Misconceptions

Three retellings travel with the dive. Standing still is treated as the rule for every kick. Regret is treated as if it always follows action. The preference for a jump is treated as the same thing as keeping whatever is already selected.
In these 286 kicks, the 18 stays stopped 6 and the 268 jumps stopped 36, with p = 0.02. The authors say that rate belongs to the aims in the footage. If every goalkeeper stayed, they expect kickers to aim at the sides. Singh and Mukherjee’s World Cup abstract reports similar stopping chances for jumping and for staying, in 292 on-target kicks. The center in the 2007 diagram is a band of three cells. A jump left stopped 9.8 percent of the center-band kicks it met.
Kahneman and Tversky’s 1982 pair, as both later papers quote it, has a large majority assigning more regret to George, who switched shares and missed $1,200. Zeelenberg and coauthors reproduce that direction when the prior match is missing: 41 of 55 name the coach who changed the team. After a 4–0 loss, 37 of 55 name the coach who left the team unchanged. The direction of the regret story follows what the prior result makes normal. The papers that quote the 1982 item do not print its percentage.
Status quo bias is a preference for the state already occupied. The default effect is a higher rate of keeping the option a form has already marked. In the kicks, the rare choice is the stay, 18 of 286, and 25 of 32 Israeli goalkeepers called the jump more normal. A pre-checked box can favor doing nothing. The pattern on this page favors the move that looks like a response. Omission Bias, in the literature Bar-Eli and coauthors cite, judges harm from action more harshly than the same harm from inaction. They use the present name for the reverse, when action is the norm.
Action Bias sits beside a preference for the state already in place, beside a mark that doing nothing keeps, beside the extra weight of a loss, beside a rating that follows the result, and beside another decision made because of what was already spent.

Status Quo Bias

A preference for the option already in place. In the filmed kicks the common choice is the jump, and the stay is 18 of 286.

Default Effect

A higher rate of keeping the option that doing nothing selects. A pre-checked box can favor inaction. The norm in the penalty study is the dive.

Loss Aversion

A loss weighs more than a gain of the same size. Patt and Zeckhauser name it among the relatives of the penchant for action. Their survey percentages are not on this page.

Outcome Bias

A higher rating of the same decision when the result is good. Here the compared result is a goal either way, and the recorded difference is 6 stops in 18 stays against 36 stops in 268 jumps.

Escalation of Commitment

A further commitment because the earlier spend would otherwise look lost. The coach vignette changes who is said to feel more regret after a 4–0 loss. It does not price that loss.

​
One-Line Takeaway

Before you treat a dive, a substitution, or a new plan as the responsible move, write the success count for staying still and the number of times anyone stayed.