Thursday, August 9, 2018

Conversation and question asking are delicate exchanges, easily bruised.

A deeply unsatisfying article about an intriguing issue.

The article is How a question is phrased can drastically alter the answer from KnowledgeWharton. With due respect to the researchers from my alma mater, do tell. The underlying research is Eliciting the truth, the whole truth, and nothing but the truth: The effect of question phrasing on deception by Julia A. Minson, et al.

Anyone who has ever done even a modicum of survey design knows this. Context matters, wording of questions matters, sequence of questions matters. A halfway competent survey designer can get exactly the answers you want based on how the questions are posed. That is why surveys are so challenging and frequently wrong.
In strategic information exchanges (such as negotiations and job interviews), different question formulations communicate information about the question asker, and systematically influence the veracity of responses. We demonstrate this function of questions by contrasting Negative Assumption questions that presuppose a problem, Positive Assumption questions that presuppose the absence of a problem, and General questions that do not reference a problem. In Study 1, Negative Assumption questions promoted greater disclosure of undesirable work-related behaviors than Positive Assumption or General questions did. In Study 2, Negative Assumption questions increased disclosure of undesirable information in face-to-face job recruitment meetings, relative to Positive Assumption questions and General questions. Study 3 demonstrated that the relationship we identify between question type and the veracity of responses is driven by inferences of assertiveness and knowledgeability about the question asker. Finally, in Study 4, asking assertive questions with regard to uncommon behaviors led the question asker to be evaluated more negatively.
Fair enough. But what is new here? The rest is behind a pay wall so I cannot see what insight they are adding that hasn't been surfaced in the past fifty years of research into survey design.

Which is too bad because there are a lot of intensely interesting and useful issues in the topic.

Every conversation is a dance where the two parties are trying to understand and forecast the other. What are their respective bodies of knowledge, what are their assumptions, what are their interests, what are their desires, what is important to them, what are appropriate topics of conversation? The music of a conversation usually starts slow and builds, especially if it is between strangers. There is an ebb and flow, tentative assertions and modest retreats. A mutual seeking of a common ground.

Much the same is true of survey designs. You have to understand the audience or you have to have the time to discover them. As with a conversation, and this may be the point of Minson and her colleagues, there is a two way flow of intentional and unintentional information. Much is revealed of one to the other which in turn causes an evolution of expectations.

If they are taking their research in a useful direction, Minson et al will begin to build a structure by which we can understand how there is a meeting of minds between surveyor and surveyed. How do you reach a pertinent alignment of information and values and assumptions with a salient population in whom you are interested?

It is a lot more complicated and fallible than is usually assumed.

No comments:

Post a Comment