Survey advice

The recent surveys that have popped up here on the boards seem to all suffer some common errors. Here’s some of my advice and I’d like to hear some other comments.

  1. I’m assuming most of these surveys are student based. In the real world, everything you do will have a cost associated, including doing a survey. Try to keep your survey questions focused on what you really want to find out, not just jam it full of the normal questions.

  2. Proof-read your survey. I suggest reading your survey out-loud to yourself. Remember, ambiguous questions will only get you useless results.

  3. DO ask questions on how to improve your survey. There are many designers on these boards with years of research experience. They would love to hear precise questions relating to survey design.

  4. Think of your target market. If you are designing a stroller for single mothers, it is doubtful that your results garnered from a male-dominated professional website will be of much value. Do a google search for groups and forums made up of your target market.

  5. Really open ended questions are useless. The real value of research is in finding the links that people don’t normally make.

Some links:

How to design a survey : Survey Design Software : Design A Successful Survey System

Please stop bad survey design :

In my experience I think ANY survey I’ve seen on the forums has come off as “I’m a student, and I’m too lazy to really do research, so I’m going to draft up these 5 questions before going out to drink and hope that enough people on the internet have clicked OK by the time my assignment is due”.

They all read as if someone knows what the answer is to their question, and as if the survey responses of a fairly consistent demographic is going to give valuable insights into understanding a problem.

I still believe that even the most introverted of students need to experience going outside and doing real face to face research - even if it’s giving a survey in person. Otherwise it’s just checking a box to say “Yeah, I did research”

Cyberdemon: That’s what I wanted to say, but in a more positive manner;)

I’d like to think that #4 is a no-brainer, when doing a survey

Another tip.
Try to engage the audience a bit more then “Please do my survey of X”

2a. When proofing, read backward (from the end of the sentence to the start). It forces you to concentrate on the word, not the sentence. Its very easy to see what you think a sentence says when reading it “normal” versus what it actually says.

(Can you find the typo above?)

3a. Ask how to improve your survey BEFORE you publish your survey.

5a. An “other” catagory with a blank space for filling-in with multiple choice is OK. 1 or 2 open-ended wrap-up questions at the end of the survey is OK but not recommended. There really needs to be a reason for the respondent to complete them, make them important.

  1. Have an odd number of choices with the middle being neutral. 1=extremely dislike, 2=dislike, 3= neither dislike or like, 4=like, 5=extremely like. Have 5-7 choices, 3 is too little and 9 is too much.

  2. 1=negative attribute, 5=positive attribute, see example above. You can do it the other way but in the very least, be consistant with your ordering.

  3. Offer incentive to do survey. Can be monetary or a feel good.

  4. Realistically estimate the time needed to complete the survey.

  5. Indicate where they are in the survey (50% complete, etc.)

  6. Use questions to hit your target audience. If you are looking for single women ages 31-40, use the first questions to weed out those who aren’t in your demographic and give them a thank you page. Allow single women ages 31-40 through to the remainder of the survey.

  7. Use surveys primarily for high-level strategic questions (posistiong strategy, design strategy, value proposition, core values, etc.) and not for low-level tactical decisions (preference testing, usability testing, etc.).