Donald Trump and Clip Art

Symbols versus Reality in Decision Making and Behavior

For this week’s episode of Human Factors in History, we went back into the archives to find this great study by Richard LaPiere from the 1930s. It is particularly useful to revisit studies like this one because it is unlikely that we could do anything similar today. Read on to find out why.

Dr. LaPiere was traveling extensively with a young Chinese student and his wife. While that doesn’t raise an eyebrow today, remember this was the 1930s. He was concerned that they might experience overt racism along the way. In most surveys of the day, people were not shy about expressing racist views. This is the source of my comment about not being able to reproduce the study today. Even where racism exists, response bias hides the magnitude.

During their travels, they stayed at 66 hotels and ate at 184 restaurants. In each case, Dr. LaPiere pretended to have some chore at the car so the Chinese couple had to go in to ask to be served. His objective was to evaluate the response of the proprietor and to count how many times they were refused service.

My Take

This is not the most controlled study, but he did his best. He remained unobtrusive and never told the Chinese couple about his plan. He got just close enough to see the response of the proprietor and he rated them along a six point subjective scale.

The results were the opposite of what he feared. They were only refused service at one very dilapidated auto-camp. The reason was clearly racist, with the proprietor referring to the couple by a term that I won’t repeat here. In almost half of the cases, the couple was received better than if LaPiere had gone in alone (in his estimation).

But the next step yielded even more surprising results. He waited six months until the visit would have faded from the proprietor’s memory. He surveyed the same establishments as well as a control group of similar but not visited establishments. He had two types of questionnaires. One asked specifically about Chinese guests. In the other version, he asked a series of questions about Chinese, German, French, Jewish, Indian, Japanese and other groups to see if this had any effect and to hide the Chinese focus.

The results amazed me. In over 90% of the questionnaires, for both versions and for both hotels and restaurants, the proprietor responded that he would not serve Chinese. The rest replied that it depended on circumstances. None said that they would serve them.

From my 2015 perspective, I would have guessed just the opposite. I would have guessed that survey respondents would deny their racism in the survey but at least some would behave this way in reality. Not only was it the opposite, but it was over 90% in the opposite direction in both cases.

LaPiere’s explanation is also interesting. He differentiates the symbolic response to the questionnaire to the behavioral response to the service in person. He concluded that the individuals were able to be racist against a theoretical Chinese person. But they could not be racist against a smiling and engaging Chinese couple in front of them. In one case, they were considering an abstract category. In the other, there was a living, breathing human (or two).

Which makes me wonder a little less why political opinion polls are often wrong. First, respondents may think of themselves as a “likely voter” even when they never vote. They may think of themselves as a “Democratic” or “Republican” voter but then be faced with a real person when choosing a candidate. And perhaps back to a symbol again when voting because there is only a name on the page, not a face (at least in the U.S.).

Your Turn

What would your expectation have been before reading the results? Would you have predicted more racism in person or in the questionnaire? Would you have predicted the extreme results of over 90% agreement in both cases? Also, what do you think about the approach? Could we do something like this in 2015?

Image Credit: Michael Vadon and ClkerFreeVectorImages

2 thoughts on “Symbols versus Reality in Decision Making and Behavior”

  1. Turns out, there is research like this going on as we speak. And using a method I am totally familiar with and embarrassed to say I didn’t think of before.

    Are you familiar with the research where they test bias by sending resumes with equivalent credentials but some with male names and some with female names? Or some with Caucasian and some with African-American names? It is really the same thing as the LaPiere study.

    A new one just came out that looked at the acceptance of various churches for race. They sent Catholic, Evangelican, and mainstream churches a letter supposedly from a potential new congregant asking about services. Some had Caucasian sounding names and some African-American sounding names. They measure the speed of the response, the length of the response, and a subjective rating of how welcoming it was.

    Perhaps counterintuitively, the mainstream churches were the ones that exhibited racial disparities in their responses. The Catholic and Evangelical churches did not.

Leave a Reply

Your email address will not be published. Required fields are marked *