In my view, data are guilty until proven innocent. By that I mean we cannot assume they are error free and honest in what they seem to be telling us. Justice in statistics can be harsh. :-)

Statisticians working with consumer survey data for the first time are often alarmed when they dig into the data. They can be especially critical of the questionnaire, which may have been designed by a person or persons with no formal background in survey research.

Many questions may make little sense to most people, or have different meanings to different people. Answer categories may overlap or be ambiguous in other ways.

People vary in response styles too and, furthermore, their attention may wander during sections of the questionnaire of little interest to them.

In the worst case, important marketing decisions may be made on the basis of how respondents interacted with the survey instrument.

Fortunately, besides professional questionnaire design, there are statistical means psychometricians have developed which can help reduce the noise in survey data.

This is a big topic and I can't do justice to it in a short LI post, other than to say statisticians familiar with survey research now have a multitude of tools to help them clean survey data and avoid being "conned." .Kevin Gray

✴️ @AI_Python_EN