Designing surveys

Subject: Designing surveys
From: "Geoff Hart (by way of \"Eric J. Ray\" <ejray -at- raycomm -dot- com>)" <ght -at- MTL -dot- FERIC -dot- CA>
Date: Wed, 7 Oct 1998 08:10:12 -0600

Theodora Mazza is <<...designing a survey and would like to know if
anyone has any words of wisdom regarding surveys, sample questions,
questions to avoid, and so on.>>

"Surveying" is enough of an art and a science that you should do some
serious reading on the subject if you intend to do lots of it. I can
point you to <ahem> two of my articles in Technical Communication for
some relevent details:

Hart, G.J. 1996. The five W's: an old tool for the new task of
audience analysis. Tech. Comm. 43(2): 139-145.

Hart, G.J. 1997. Accentuate the negative: obtaining effective reviews
through focused questions. Tech. Comm. 44(1):52-57

I'm currently working (very slowly; sorry George) on another article
on experimental design that should provide more focused details. A
few very important things that I'll go into in some detail in that
article. Off the top of my head:

- Learn a little bit about experimental design so you understand how
to look for bias and errors of logic (among other errors) in your
questions.

- Ask some friends who don't know the subject as well as you do to
see whether they can find a way to misinterpret your questions and
provide answers that you weren't looking for. Reward them for
"breaking" your survey, then fix the problem. You'd be surprised how
hard it is to pose an unequivocal question.

- Answer the survey yourself, then sit down and see how difficult it
is to tally up and analyze the answers. Now multiply that degree of
effort by the number of completed surveys you expect to receive.
Ouch! Now go back and see if you can't structure the questions to
make analyzing the data less painful. If you can make a multiple-
choice survey work, there are ways to automate data processing (e.g.,
using OCR to scan the answers into a database so you don't have to
enter the data manually), but you're never going to get an entirely
painless survey.

- Figure out how to motivate people to respond (e.g., offer a chance
to win prizes; give _everyone_ something like a mousepad for their
efforts).
--Geoff Hart @8^{)}
geoff-h -at- mtl -dot- feric -dot- ca

"Microsoft Word: It grows on you... but with a little fungicide,
you'll be feeling much better real soon now!"--GH


From ??? -at- ??? Sun Jan 00 00:00:00 0000=



Previous by Author: Explaining click and drag
Next by Author: Blue chip and bored
Previous by Thread: Text for Course in SW/HW Doc
Next by Thread: designing surveys


What this post helpful? Share it with friends and colleagues:


Sponsored Ads