Sunday 28 April 2013
I am always looking for new ways to evaluate research. It is important that students realize that it not only about ecological validity and sample size, but there are many different lenses by which a study or theory can be evaluated. This winter I read “Thinking Fast and Slow” by Daniel Kahneman, where I was introduced to my new favorite evaluation strategy - the peak-end rule.
The peak-end rule is a heuristic in which people judge experiences largely based on how they were at their peak and at their end. I caught myself using this heuristic during Prague’s annual restaurant festival. A bunch of friends and I went to dinner. On the way home we talked about the restaurant and I said, “I wasn’t very impressed by the meal.” My friends started to laugh. They said, “John! You were smitten by the soup. You must have said five times how great it was! And then you said that you thought the main course was perfectly presented and amazing both in texture and taste. You only complained that the dessert wasn’t what you expected.” Aha. I had employed the peak-end rule. The fact that I was disappointed at the end of the dinner meant that my perception of the whole dinner was rather negative. The flip side could have been true. A mediocre dinner with a great dessert can be a really positive memory.
We often use this with movies as well. Think about the films we watch. We are more likely to recommend a movie that has a slow start but an amazing ending than a movie that has an amazing start but a rather lame ending.
Kahnemann et al (1993) asked participants to hold their hand up to the wrist in painfully cold water until they are invited to remove it. With their free hand, participants recorded how strong the pain was. The researchers used a repeated measures design. The two conditions were:
Condition 1: 60 seconds of immersion in water at 14 degrees celsius. End the end of the 60 seconds the experimenter instructed them to take their hand out and gave them a warm towel.
Condition 2: 90 seconds of immersion. The first 60 seconds the same as Condition 1. At the end of 60 seconds the researcher opened a valve that allowed slightly warmer water to flow into the tub. The water temperature rose about 1 degree Celsius.
The participants were then told that there would be one more trial - either a repeat of Condition 1 or repeat of Condition 2. Now, if you look at the two conditions, it makes sense that Condition 1 is the smarter choice. Both conditions have the same level of pain for 60 seconds - but after that time, Condition 1 gets a warm towel while Condition 2 gets a slight decrease in pain for an extra 30 seconds.
80% of the participants chose the second condition! This is a clear example of peak-end rule. The fact that the second trial was longer was not taken into account by the participants (something called duration neglect). They were basing their choice on how the condition ended, rather than making an overall assessment of the pain.
So, how can we use this? This heuristic is particularly problematic in the study of relationships. Much of the research done in retrospective - for examples, research on marriages that fall apart often are studied only “after the fact.” This means that the research is open to memory distortion on the behalf of the participants. In a study of social penetration theory the researcher may ask the participant to rate the level of disclosure in the relationship. If the couple was estranged during the last year of the relationship, it is very possible that due to peak-end rule, the perception will be that disclosure was “always a problem” in the relationship.
Can you think of more ways to use this heuristic as an evaluation of research?