Continued from page 1
* When using multiple choice answers to jog
respondent's memory, will it create bias towards available choices? Is that okay?
* It might seem easiest to use a Yes/No or True/False question, but be sure that these two answers are only ones that a respondent might possibly want to use or they will be forced to choose, diluting truth of their answer.
*When using ranges, for example "strongly agree" to "strongly disagree," how many choices are necessary? Five is typical - but is it possible that many users will be "lazy" and chose middle, and neutral, position?
*Before giving using text fields, known also as "free form" fields, consider how you will summarize these responses.
As you develop survey, think about different types and picture results that will come from using each type. With your audience in mind, determine which type would provide most effective results for what you need to find out. You can further refine these decisions when you test survey, as explained later.
Protect Against Non-Response
It is unavoidable that a percentage of those surveyed are non-responders. The best way to minimize this element in survey results is to keep survey short and to point, follow a clear outline with stated goals, and accurately target who survey recipients need to be.
Believe it or not, research states that a 10-15% response rate on any survey is considered a success.
Dangle a Carrot
Even with every precaution taken, creating surveys is never an exact science because there is little control over human element. For whatever reason, you will often scratch your head wondering why people looked at survey and dropped out, or just never even looked. A carefully executed survey can minimize this - and incentives can take it a step further. Do you have something you can give away, to encourage folks to take your survey? Providing a free item or bit of information, providing a free copy of results, or some other lure can go a long way towards gathering results.
Fly a Pilot
When you, yourself, develop a survey, it is easy to overlook elements that might stop your respondents from completing survey and wording that might not be clear. Always be sure to test survey on a few people prior to sending it out to your entire audience. This will also allow you to review results and see if questions asked and way they were answered will result in effective results.
Cut it Off
Be sure to tell your respondents when survey will be closed. Typically, most respondents will respond within 5 days of receipt and so unless you know for sure that a significant portion of your audience needs longer, a week is a fair amount of time. Do not send out survey on Fridays or over weekends - we all know what we do with mail received on those days! Try Tuesday morning for optimal response, and expect to see approximately 50% of your responses in first 24 hours.
It may seem like you need a dual degree in cognitive psychology and statistics to develop an effective survey. This not being true for most of us, these guidelines will provide a foundation to improve both survey and results of your efforts. There is no better way to find out what your customers want than to ask them, and if they are already online, web surveys are an ideal way to do it.
Eileen Parzek, better known on the web as "Turtle", is the owner of SOHO It Goes! supporting entrepreneurs and small business through design, communications and technology.
