How to Conduct an Online Survey
Our Online Survey
Software (Features and Samples)
The process involved in conducting a online survey is similar to a mail questionnaire. However, there a few special considerations. This page discusses the basic process of conducting a
online survey with special attention to the features that are different from mail surveys.
Clarify your research objectives
Decide on a sampling method
Invite the sample to participate in the Web survey
Host your Web survey on the Internet
Data collection and verification
Statistical analysis reports
If you are interested in trying a full-working version of StatPac's Online Survey Software, please click the link below.
Online Survey Software - Download
1. Clarify Your Research Objectives
All research begins by
defining the goals of the study.
What do you want to learn by
conducting an online survey?
Commit the goals
to writing. Doing so will help to
keep your survey focused. The
number one reason for the failure of surveys is a lack of
well-defined objectives. Poorly
defined objectives yields excessively long surveys and
ambiguous results. Clarify your research
objectives to focus on the
important information needed for
decision making. The online survey itself should be designed to directly address the research goals. In other words, stay away from "fishing expeditions".
2. Decide on a Sampling Method
Proper sampling is essential
for all research. If you select
the wrong people, then the
results will not reflect the true
attitudes of the population. If
you choose the right people, then
the results will provide a solid
foundation for decision making.
The goal in sampling is to select
people who represent the
Equally important is to get a high response rate,
If the response rate it too low, we question the reliability
of the data. Are the responders in some way different from the
nonresponders? So it's extremely important to choose a sampling
method that maximizes response rate.
Sending e-mail invitations seems to be the only method
that produces high (but wildly varying) response rates (15%-80%). Of course,
this requires that you have the e-mail addresses of the sample.
We strongly advise you not to send "spam".
While StatPac can be used to send bulk e-mail to any e-mail list, we
suggest that you send e-mail only to people with whom you've
had previous contact.
Publishing the URL to a web survey in written form (e.g., newsletters,
newspapers, postcards, letters, etc.) will likely elicit a low response
rate (<5%). Additionally, pop-up or pop-under windows also get
poor response (typically 2%-4%).
One sampling method that appears to work on the
Internet is called a snowball sample. In the e-mail invitation, you
ask potential respondents to forward the e-mail invitation to other
people who might also be interested in the topic. The potential
for bias is high with a snowball sample, so we discourage
the use of this method. However, it may be appropriate for some surveys,
especially when the researcher has a severely limited e-mail list.
There are four things that most affect response rates: 1)
the interest of the respondent in your research topic, 2) the quality of the
invitation, 3) the use of an incentive, and 4) the number of questions asked.
3. Questionnaire Design
surveys are more critical than
paper and pencil surveys. Our
research has shown that an online
survey must be shorter than
traditional mail questionnaires.
Respondents' willingness to
participate and complete a
questionnaire depends heavily on
its length. Thus, when conducting
a online survey, it becomes
imperative to ask as few
questions as necessary to get the
information you need. If you need help designing your questionnaire, here's a free designing questionnaires tutorial.
4. Invite the Sample to Participate
In a typical online survey,
potential respondents are sent an
e-mail inviting them to
participate by clicking on a link
in the e-mail. While you can use any e-mail program
to send the invitations, StatPac for Windows can
customize each e-mail with a
personal greeting or other
information you want to be
included. More importantly, it can
serialize each e-mail with
an ID number in order to track
who responded and who didn't.
The e-mail invitation you send can
be thought of as a "sales pitch".
Sell the potential respondent on why they should
complete your questionnaire. What benefit will they receive by completing
the questionnaire? One way to provide a benefit is to offer an incentive. Potential
incentives are drawings for prizes, donations to charities, a small tangible
gift, cash, a copy of the results. Altruistic appeals are usually not effective.
5. Host Your Surveys on the Internet
You need access to a server to host your
surveys on the Internet. Users of our software
can host their surveys on their own Web site or
our Web site (for free). StatPac for Windows
software has selections to use our free server.
If you are using the free version, the link
to your survey will be:
If you are using the registered version, you
can create a private folder on our server so the
link will be:
Note 1: Our server is Unix based so all links
are case sensitive.
Note 2: SSL secure surveys must contain the www
as part of the link.
All the server management tools are built
into StatPac. You'll be able to upload surveys
and download responses.
6. Data Verification and Collection
Internet surveys are different
from paper and pencil
questionnaires. Data checking and
verification are performed
immediately while the respondent
is taking the survey. All the data is captured
electronically and no manual data entry is
There are two basic methods of capturing
the data electronically. The first method is by e-mail.
When a respondent completes a survey his or her
answers are e-mailed to you and the data is captured
from the e-mail. The e-mail is method is not frequently
used. The second method is to store responses in a file on the
server and then download all of them to your local computer at once.
The file method is recommended because all the data is kept on the
server and it uses fewer Internet resources. If you have
more specific questions about the technical aspects of Internet surveys please
see our frequently asked questions.
7. Statistical Analysis Reports
The whole idea behind doing surveys is to
get decision making information... which means
clear and comprehensive reports. Look for
software that creates reports as MS Word
documents so you can easily customize the output.
Typical reports include
with graphics, and crosstab and
banner tables that show your key
variables broken down by the
demographics of the sample. Banner
tables give the information you need to
identify opportunities and to make
knowledgeable decisions based on the data.