Haas Newsroom


Gambling: a game of emotions, not odds

UC Berkeley marketing professors find that even when gamblers predetermine a spending budget, all bets are off after the first loss


January 27, 2009


Haas School of Business Media Contact:


Pamela Tom

(510) 642-2734



Ute Frey

(510) 642-0342



When UC Berkeley marketing professor Eduardo Andrade and his wife went to Vegas for the first time, Mrs. Andrade insisted that they should make a responsible plan. After a short debate, they decided to gamble no more than $200 per day. When Mrs. Andrade lost at the table, broke the deal and spent more than planned, Andrade made her inconsistent turnabout the subject of his next research. What made her change her mind?

 “Planned versus Actual Betting in Sequential Gambles,” by Andrade and co-author Haas Marketing Professor Ganesh Iyer, finds that in the planning phase, people behave conservatively—they say they will bet less in a following gamble in the event they lose a previous one. However, after experiencing an actual loss, individuals open their pocketbooks wider than planned. Andrade is an assistant professor at the University of California, Berkeley’s Haas School of Business, Marketing Group and studies the effects of emotion on behavior. Iyer is the Edgar F. Kaiser Professor of Business Administration and Associate Dean of Academic Affairs at Haas.

 “When gamblers haven’t experienced the actual pain of loss, they make cold and deliberate assessments of how much to bet in case of a future loss. Many think, ‘well, since I’ll have less money, I should bet less’. When they actually lose, however, the resulting negative emotion makes them overreact,” says Andrade.

The paper suggests emotions can be manipulated to alter spending behavior and its implications may reach beyond the casino to state-owned lotteries, public policy and the issue of consumer overspending.

The researchers set up a series of three revealing experiments. Participants planned and placed a sequence of gambles. The game resembled roulette. Blue and red squares appear on an electronic screen. An “X” flashes on the screen for 15 seconds and randomly stops on one of the squares. If the X stops on a blue square, you win; on a red square, you lose.

Each experiment followed three steps: (1) a trial version (no betting involved), (2) a planning phase, (3) and an actual betting phase. The trial version simply allowed people to see how the gamble worked. Then, in a planning phase, participants were required to make betting “commitments”, stating how much they would bet in the first game and in subsequent games given a gain or loss in previous gambles. Furthermore, the participants actually believed that they would bet as planned. Finally, during the actual phase of the gamble, participants were offered an unexpected chance to revise their bets if they so desired.

Undergraduate students, each paid $15, participated in the experiment. They could use part of this money in the gambles. Since it represented their own participation fee, each decided on whether and how much to gamble within a given range (for example, from $0 to $5 per game).

The results showed that, at the planning phase, individuals said they would bet less after losing a previous gamble and roughly the same amount after winning a previous gamble. 

However after the first game, 40 percent of the losers changed their minds and deviated from their plan. In addition, 90 percent of that 40 percent group bet a larger amount than they planned in the next game. In contrast, winners, on average kept their promises. 

Andrade explains, “The deviations from the plan after losses surprised us for two reasons: First, participants knew everything about the gamble before planning their bets. Second, the time delay between planning and actual phases did not take longer than one minute. Why can’t people correctly anticipate how they would behave?”

Andrade and Iyer reasoned one main aspect differed from planning to action: emotions. When planning, people are not experiencing the loss. When the loss is finally felt, people overreact by increasing their stakes. The authors show that after a gain people also deviate from the plan, but not in a particular direction: some bet more than planned and some bet less. On average, no deviation shows up.

When the researchers asked participants to also predict how they would feel if they lost. Many gamblers underestimated how bad they would feel after losing, and often were more likely to bet more than planned. Andrade points out, “That was my wife’s reaction. She was actually astonished by how frustrated she felt after losing, and said the only way to feel better was to try to recover the loss and make some extra money.”

The third experiment confirmed the emotion hypothesis.  After the first game, researchers showed videos to the losing gamblers. One group watched the TV sitcom, “Friends”; another watched a relatively innocuous documentary; the third group watched a depressing five-minute clip from the 2001 drama, “Life as a House”. Then, the second gamble began. Participants could keep or deviate from their plans.

The results proved negative emotions affect the pattern of decisions. Those who watched the depressing scenes showed the strongest pattern of deviations from the plan. Those who watched the documentary also deviated but not as much. Those exposed to the “feel good” sitcom did not deviate from their original betting plan in the second game. The conclusion: once the pain was detached from the loss, the deviation disappeared.

The paper reports consumer spending in commercial casinos increased from $17.1 billion in 1996 to $32.4 billion in 2006.  Andrade adds, “From a public policy perspective, the open question now is how much of that spending is unplanned and whether such deviations might hurt individuals’ wellbeing in the long run.”

This paper is forthcoming in the American Marketing Association’s

Journal of Marketing Research, June 2009.