The scientific equivalent of "When did you stop beating your wife?"

Illustration for article titled The scientific equivalent of When did you stop beating your wife?

There's a famous joke question: "When did you stop beating your wife?" The structure of the question is funny — or disturbing — because any response condemns you. You'd assume that a quick-witted person could see through it. But this study proves that you're wrong.


"Do you think your contempt for America might influence your judgement?"

"I don't know. Have you ever noticed a disregard for personal freedom characterizing your questions?"

These are both examples of questions that we have all, unfortunately, heard during debates on one tv show or another. Sometimes it is a deliberate attempt to characterize the responder's actions as negative no matter what they say. Sometimes it's someone tipping their hand and showing their unconscious bias. When we hear it, we like to think that we're above such simple verbal trickery. A couple of studies show that we're not.


Elizabeth Loftus and John Palmer gained fame in the 1970s with two experiments that show that how a question is framed can alter a person's opinion, not just shading it with bias, but quantitatively changing the way they remember a past event. The first study involved showing people different videos of car collisions, ranging from five seconds to half a minute long. After each clip, the students were asked a series of questions, including one that required them to judge the speed of the cars. The question used different verbs to describe the car wrecks: smashed, collided, bumped, hit, and contacted.

Students tended to respond with different speeds depending on the word used to describe the collision. "Smashed," had the students estimate the cars going at just over forty miles per hour. The estimated speed was reduced for "collided," further reduced through "bumped" and "hit," until it reached a low of 31.8 miles per hour at "contacted."

In a second experiment, students were shown a single video of a crash, and then asked questions. Loftus and Palmer had them come back a week later to answer a few more questions. They again used varied description words for the collision, but this time they were more interested in a later question. They asked the students if there was broken glass on the ground at the scene of the accident. Those who had gotten the questionnaire with the word "smashed" were more likely to remember broken glass than those who had gotten the word "hit," and were nearly three times as likely to remember broken glass than those who hadn't gotten a question which described the accident. There was not broken glass at the scene in the film.

Loftus and Palmer concluded that, "When the experimenter asks the subject, 'About how fast were the cars going when they smashed into each other?', he is effectively labeling the accident a smash." A smash sounds more severe than a hit, and so people conclude that there had to have been broken glass. There had to be. The cars were smashed. To be fair, the largest portion of the group correctly remembered the accident as having no broken glass - only sixteen out of fifty remembered broken glass in the "smashed" experiment.


Still, the experiment shows that language communicates far more effectively than we consciously think that it does. There's a reason that we have come up with so many synonyms for words, each of which is shaded with slightly different meaning. We want to communicate information. The problem is that we communicate that information even when we don't mean to.

Source: Journal of Verbal Learning and Verbal Behavior

Image: Shuets Udono


Share This Story

Get our newsletter



I'm not sure that any response to "when did you stop beating your wife?" condemns you. "I have never beaten my wife" is a fairly safe answer, I think.