Career Strategies for Librarians
So You Want to Prepare a Research Survey?
by Robert Perret

You have reached that proud day in every librarian’s life when it is time to prepare a real honest-to-
goodness publishable research survey.  But wait!  Has it been a year (or ten) since you computed your
last statistic?  Do you think of a chi square as the kind of biscotti that goes well with tea?  Can you still
tell a mean from a median?  Are you sure?  Not so long ago I too took the leap and conducted a survey
about the background and experiences of my fellow business librarians from around the world.  If, like
me, you are going into a research survey with a bunch of good intentions and fuzzy recollections, please
consider these insights as you move into the exciting world of original statistical research.

Do the literature review first (no, really): A funny thing happens when librarians do their own
research - they become their own worst patrons!  The sad truth is that sometimes other people have the
audacity to beat you to a good idea.  Before you spend weeks lovingly handcrafting the perfect survey
(and especially before you perform the survey), make sure it hasn’t been done by someone else.  
Research surveys are surprisingly time- and effort-intensive.   You don’t want to have to scrap a survey
because you didn’t take a day on the front end to see what else had been published on your topic.  Even
if you aren’t duplicating existing research, you want to know what else is out there related to your topic.  In
my case, I found that similar surveys had been conducted in 1989 and 2001.  I framed my survey as a
continuation of this research process, asking questions similar to those on the previous surveys.  This
way I was able to use the previous survey results and draw conclusions by looking at the data over time.

Stand on the shoulders of giants: Another benefit of performing your literature review early is
seeing how others processed similar data.  What statistical tests did they use?  How did they collect
their data?  If there is no similar study within library science, consider going outside the discipline to look
at papers in fields that do this sort of thing a lot, like sociology.  Once you have a feel for the tests you
need, grab a statistics textbook (from your local library, naturally) and give yourself a solid grounding in
the relevant material.  Pay particular attention to how the data should be collected.  You can always redo
your calculations, but it is probably not realistic to rerun your survey from scratch.

Know what the point of your survey is:  This may seem self-evident, but the whole point of doing
your own survey is getting exactly the information you need.  There is no excuse for guesstimating from
or reverse engineering your own data.  More important, it is potentially difficult and expensive to rerun the
same survey.  I can’t overemphasize the importance of thinking your survey through at the outset.

Keep it short and sweet: Create the most efficient survey possible.  People will only sit through so
much, so make all of your questions count.  Survey fatigue will affect the quality of your data and the
quantity of your responses.  Ask yourself honestly, “Would I sit through this survey?”  

Just rewards: If you must perform a long and/or involved survey, consider offering incentives for
your respondents.  Free food is a good way to attract people to an in-person survey.  A raffle may be
appropriate for an online survey.  Just be sure that the value of the reward isn’t skewing your results.

When anonymous is too anonymous:  You should be able to describe who did and did not
respond to your survey.  Because I hadn’t considered this, my survey was so focused on providing
anonymity that I was unable to answer this basic question.  For instance, it may be important to note that
80% of your respondents were under 20, or that no public librarians responded.  Any basic relevant
demographic information you can get without scaring off potential respondents will be very helpful in
framing your results.  Try to find a balance between providing anonymity (when appropriate) and
collecting enough information to describe your pool of respondents.

Random by design:  It takes a lot more work to get a random sample than you might think.  There
are a number of ways to slice and dice your survey population.  Make sure you conduct your survey  in a
way that fits your needs.  If you attract participants with a flyer in the school cafeteria you may only get
freshman, who are more likely to be eating there.  If you have a pop-up window on your library terminals,
you only get people who use the library’s computers and who are willing to respond to a pop-up survey.  
That’s usually not the kind of randomness you are looking for.  Conversely, if you actively seek out
participants, ensure you are not selecting only people who will reaffirm your own biases.  Either way,
have a clearly defined sample frame or set of criteria by which you selected or found participants.  
Sampling is an art unto itself, and you will definitely want to crack a statistics book at this stage.

 And then there were none:  On my first survey I failed to offer “None” as a response to the multiple
choice question “Which professional organizations do you belong to?”  My “No response” rate on that
question was tenfold that of any other question on the survey.  I can only infer that all of those people
were indicating no such membership, but, statistically speaking, there is no way to be sure.  Watch out
for questions that are overly restrictive or leading.  

 Allow for comments: If you are doing a quantitative study, allow a mechanism for respondents to
comment.  I got a lot of good material this way; in fact, a major theme for my article came from the
anecdotal comments in the field beneath each question.  

 It’s all a numbers game: Conversely, if you are performing a qualitative survey, be ready to
quantify the results and justify the way you did it.   Your friendly neighborhood statistics textbook will help
you here as well.  Be aware that it can be very labor-intensive to scale up qualitative results, especially if
you start cross tabulating.  For instance, imagine asking ten of your friends what their favorite movies
are.  It seems pretty easy to keep a running tally.  Now imagine asking them why they like their favorite
movies.  Now imagine keeping track of all the possible responses, all of the possible combinations of
responses, and who gave each answer.  Now imagine doing that with 100 people, or 1000.  It’s not that
qualitative research is bad; it’s just that it can snowball into a lot more work than you might be expecting.

If you take away one thing from this article, I hope it is that research surveys are a challenging,
rewarding, and entirely achievable way of generating scholarly research.   A lot of forethought and a little
research (you are a librarian, after all) are all you need to create professional, publishable surveys.  

About the Author

Robert Perret is a Reference and Instruction Librarian at the University of Idaho.  He holds an MLS from
the University of Denver and an MBA from Southwestern College.  His survey of business librarians led
to an article which has been accepted for publication in the
Journal of Business & Finance Librarianship.

Article published February 2010

Disclaimer: The ideas expressed in LIScareer articles are those of their respective authors and do not necessarily
represent the views of the LIScareer editors.