survey language advice

Feb 03, 2010 17:31


I'm editing questions for a study questionnaire that will be administered via computer, over the internet. Many of the people participating will be teenagers, so we're trying to make it as straightforward as possible.  For medicines (like certain antibiotics), we're asking them whether they take it, and then, in separate questions, how strong each ( Read more... )

Leave a comment

Comments 5

brainstorm questions liveonearth February 4 2010, 02:35:15 UTC
It's going to take several questions to get the info you're after. Just brainstorming here ( ... )

Reply


winterthunder February 4 2010, 04:04:01 UTC
I don't have experience with surveys relating to public health, per se, but I did do a series of surveys of poultry farmers and biosecurity a few years back. liveonearth has done a nice job of some specific questions, so I'll toss out a few general lessons I learned. Don't make it too long, for teenagers especially, because people will stop half way through and that really screws with your data set. Don't change your survey questions between rounds of administration, as that also screws with your data set. Look at each question and try to think of all the possible ways someone could misinterpret it. Hand it out to a few people to get their reactions. Every mistake made will show up in your survey responses, and probably a few more that you didn't think of. If possible, it's a good idea to test your survey with a small focus group and rewrite any troublesome questions before rolling it out for real.

Good luck!

Reply


polerbere February 4 2010, 04:50:11 UTC
What about "How many milligrams in each pill?"

Reply


brbrbrad February 4 2010, 12:21:31 UTC
I like the idea of doing pill size and number of pills as two different questions. Here's my suggestion for wording them:

"On the package your medicine came in, you can find information about the amount of medicine in each pill, e.g. 500 mg. What is the amount listed for this medication?"

and

"How many pills do you take each day."

This won't work very well though for liquid medications or medicines with two active ingredients.

Reply


meteoricpath February 5 2010, 08:38:28 UTC
I agree with most of the above. I would only add that it is important to keep in mind the goal of the analysis as much as possible when you are creating the survey. You need to balance the length of the questionnaire (longer it is the more likely people will either rush through or not complete) with the need for precise information. For example, why is it important that you need the exact dosing? Do you need to know exactly what people are taking and when? If so then a complex question tree with skips is the best way to go ( ... )

Reply


Leave a comment

Up