When you are doing an online survey project, you are on a road trip, of sorts. You are charting a route to a desired destination: accurate, objective insights about your target market. Unfortunately, you need to avoid the faulty detours that can put you off course.
One of the greatest challenges in designing online surveys is making sure they’re objective, and to that end question wording is critical. If bias slips into our questions, the responses will be also be biased. One of the common errors that happens in online survey design is that the survey creator inadvertently (we hope) writes leading questions, and as a result gets biased responses. We don't want people giving us the answers they think we want (most people are surprisingly polite). We want their actual, honest opinions and behaviors. Thus, a key to good surveys is writing questions that give the respondent no clues about what answer the researcher would prefer.
Three Tips for Staying on Course
• Avoid “yes or no” questions — By their very nature, yes-or-no questions are almost impossible to write without bias. Even something as simple as “Did you find our customer service satisfactory?” suggests that we’d prefer to hear a “yes” response. A better option would be to use a satisfaction scale along the lines of “Please rate your customer service experience”, offering choices from “not at all satisfied” at one end to “totally satisfied” at the other. This will reduce the likelihood of bias.• Avoid “leading” the subject — This usually happens when you try to include too much context in your question. Imagine that you're a bank doing an online survey, and you pose this question: “In order to improve customer service do you agree that our tellers should be bilingual?” We’ve already assumed a connection between good customer service and bilingual tellers in the phrasing of our question, so many will agree. Or how about this one: “To help lower our costs, do you think it's important for our company to…” and then insert your favorite action. Who’s going to say “no” to that? Just by having that preamble, you’ve already told them what the “right” answer is.• Avoid the agreement scale — This is the familiar five or seven point scale with responses ranging from ”Strongly disagree” to “Strongly agree”, and applied to a series of statements. Again using a customer service survey, those statements might include “The sales representative was courteous”, “The sales representative was knowledgeable”, and so on. Still inclined to be polite, it's pretty hard for people to disagree with such positive statements, but neither would you write those statements in the negative. “The sales representative was rude” and “The sales representative was not knowledgeable” are equally biased in the opposite direction. Rather than an agreement scale, try “Please rank our sales representative on the following items,” and then offer a continuum from, “Not at all knowledgeable” up to “Very knowledgeable”, and “Slow to respond to my questions” up to “Promptly responded to my questions.” This keeps it objective and avoids creating any sort of bias in the wording.
Poor Wording Will Lead You To Biased Data
Like driving a car after a New England winter, there can be potholes where you least expect them. Sometimes it seems like a pothole ravaged road is just determined to bust your tires; every time you dodge one pothole, you have to swerve to avoid the next. The same can apply to surveys; sometimes you review a survey draft and find that every time you look, there is another source of bias that needs to be corrected. But correct it you must or you won’t get to your desired destination: accurate, objective insights.