The Customer Service Survey
VocaLabs' weblog providing news and commentary on the challenges of providing good customer service.
Question and Questioner Bias
Wednesday - September 07, 2005 01:58 PM in
One of the little understood facets of any quality survey has to do with controlling or eliminating biased results either because the questions asked can mislead or the survey administrator (usually without realizing it) causes false answers.
Certainly blatant examples exist that we'd all recognize as causing biased feedback. "Do you prefer our luscious melt in your mouth candy versus that other brand?" But what I'm referring to are more subtle phrasings and even the voice inflection of the survey giver that can easily skew the study results.
It is human nature to want to please, and survey participants are no different. If the survey taker gets some signal that there is a preferred response, they will tend to give the "correct" answer even if it isn't the more truthful one. This is particularly significant in face to face questioning where the test administrator also has to give no facial expression or body language that sends a signal as to which answer is preferred.
And it doesn't have to be a visual clue as even a jaundiced, tired or under trained outbound telemarketing call center agent can influence responses with voice inflections and accent.
Question bias is just as insidious, but fortunately easier to control. One example: "How satisfied were you..." (skews toward higher satisfaction ratings) versus the more balanced "How satisfied or dissatisfied were you..." Peter uses a great example: "Do you like milk?" gets different answers if the previous question was "Do you like cookies?"
Most anyone can write survey questions and there are inexpensive aids on conducting studies available. But writing and presenting truly objective questions isn't easy. The internal do-it-yourself route may be adequate for your purposes. Just understand that if objectivity and truthful results are paramount; reputable professionals deserve serious consideration.
Posted by Rick Rappe
It is human nature to want to please, and survey participants are no different. If the survey taker gets some signal that there is a preferred response, they will tend to give the "correct" answer even if it isn't the more truthful one. This is particularly significant in face to face questioning where the test administrator also has to give no facial expression or body language that sends a signal as to which answer is preferred.
And it doesn't have to be a visual clue as even a jaundiced, tired or under trained outbound telemarketing call center agent can influence responses with voice inflections and accent.
Question bias is just as insidious, but fortunately easier to control. One example: "How satisfied were you..." (skews toward higher satisfaction ratings) versus the more balanced "How satisfied or dissatisfied were you..." Peter uses a great example: "Do you like milk?" gets different answers if the previous question was "Do you like cookies?"
Most anyone can write survey questions and there are inexpensive aids on conducting studies available. But writing and presenting truly objective questions isn't easy. The internal do-it-yourself route may be adequate for your purposes. Just understand that if objectivity and truthful results are paramount; reputable professionals deserve serious consideration.
Posted by Rick Rappe
Posted at 01:58 PM by | | | |