An email survey is one where the body of the email
contains the survey itself. This is in contrast to an Internet survey where you
send an email invitation and the body of the email contains a link to the
The problem with email surveys is that you don't know
what kind of email reader your respondents are using. A survey that looks
fine in one email reader might appear distorted in another email reader.
Therefore, we discourage you from sending email surveys.
Nevertheless, there are two kinds of email surveys: HTML
and plain text.
In an HTML Email survey, the email you send is the
survey itself. When the respondent views the email, they will see the survey.
Follow the same procedure as you would if creating an
Internet survey, including uploading the survey to the Internet. Then,
instead of sending the respondents a link to the survey, you send them the
first page of the survey (i.e., the _1.htm file).
When they open the email, they will see the first page
of the survey, and when they click the Submit button, the responses will be
stored on the server and the second page or thank you page will be displayed.
Data is retrieved from the server as if it were an Internet survey.
Conceptually, HTML email surveys are wonderful. However,
in reality, HTML email surveys can cause several problems. Some people have
their email readers set to not display HTML. Other people have their email
numbers to track who responded and who didn't.
For these reasons, we discourage you from sending HTML
email surveys. If you choose to send them anyhow,
please do so knowing that some respondents are likely to have
A plain-text Email is one where you create a survey
using ASCII text and include it as part of the body text in a regular Email.
The survey is Emailed to a potential respondent. They click their Reply
button, complete the survey, and then click the Send button to return the
completed survey to you. Upon receipt of their reply, you can immediately
import it into a StatPac data file, or filter it to a mailbox (i.e.,
file) with other returned Emails and import all the responses at once.
Importing returned Emails appends the respondents
answers to the end of a StatPac data file.
There are two kinds of variables that can be included in
a plain-text Email. One is where the respondent uses an X to check the
appropriate response(s). The other is where the respondent fills in a text or
numerical response. The following simple survey shows both kinds of
variables. The first five survey items are "X the box" and the last
five items are "fill in the blank":
TO PARTICIPATE, PLEASE CLICK THE REPLY BUTTON.
Type an X between the brackets to indicate your selection.
1. Gender [ ] Male [ ]
[ ] Black
[ ] White
3. What is your favorite brand?
[ ] Brand
X [ ] Brand Y
[ ] Brand
Z [ ]
4. Overall, how would you rate our product?
(Type an X between the brackets to
indicate your response.)
[ ] [
] [ ] [ ] [
] [ ] [ ]
5 6 7
5. Where did you hear about the product?
(Type an X between all brackets
[ ] Radio
[ ] TV
[ ] Newspaper
6. Overall, how would you rate
1 2 3
4 5 6
Type the number of your rating here.
7. How old are you? [
(Type your age between the brackets)
8. What do you feel would be a fair price for this product?
] Type the dollar amount between the
9. What is your favorite brand?
(Type the name of your favorite brand between
10. What could we do to make our product better?
(Type your answer between the
CLICK THE SEND BUTTON TO FINISH THIS SURVEY.
THANK YOU FOR YOUR PARTICIPATION.
Notice that the brackets are used to show respondents
where to type their answers. The text of the Email is completely free-form,
except that brackets may only be used to specify where respondents are
supposed to type their answer. You may not use brackets anywhere else in the
The numbering of the survey items is optional. The
purpose of these numbers is to guide the respondent from one question to the
next. They are not necessarily the same as the variable numbers in the
codebook. That is, some survey items might be multiple response and occupy multiple variables in the codebook.
For example, survey item 5 is multiple response and
takes three codebook variables. From that point on, the item numbering on the
survey is no longer is the same as the codebook variables.
Numbering the items on an Email survey also serves an
additional purpose. When a respondent replies to your Email, StatPac will
extract their answers from the Email by looking at what's between the
brackets. If a respondent inadvertently deleted one of the brackets, StatPac
might get confused on which variable it was evaluating. When item numbering
is not used, StatPac will report the error and not add any of that
respondent's answers to the data file. When item numbering is used, StatPac
will be able to identify the start of a new variable or series of multiple
response variables, and it will leave only the defective item blank. StatPac
will report the error, but the rest of the data for that respondent will be
added to the data file. Thus, item numbering allows StatPac to
"re-synch" with the variables in the codebook, even when a bracket
has been deleted.
If you do use item numbering on your Email survey, the
number may begin with the letter Q (a common abbreviation for question). It
must also end in a period followed by a space, and it may not contain
non-numeric characters. Each of the following would be correct usage of item
5. Where did you hear about the product?
Q5. Where did you hear about the product?
Q5. (a) Where did you hear about the product?
The following would be incorrect usage of item
numbering. The first example does not have a period following the number. The
second example has a space after the Q. The third example has the letter
"a" before the period and space.
5 Where did you hear about the product?
Q 5. Where did you hear about the product?
Q5a. Where did you hear about the product?
Codebook Design for a Plain Text Email
The codebook for an Email survey is the same as any
other codebook. The rule is, if you want the respondent to use an X to show
her response, the codebook must specify a value label for each response. If
you want the respondent to fill in an answer, do not specify any value labels
for that variable.
Variables can be alpha or numeric. When multiple sets of
brackets (boxes) are required, they can be placed horizontally (items one and
four) or vertically (items two and five). If
you specify more than one column of boxes (item three), they must be
positioned horizontally. That is, the first box is associated with value
label 1 and the box to the right of it is associated with value label 2. In
the second row, the first box is associated with value label 3, and the box
to the right of it is associated with value label 4.
Here is the codebook for the first five items:
V1. Gender (A1)
V2. Ethnicity (N1)
V3. What is your favorite brand? (N1)
V4. Overall, how would you rate our product? (N1)
V5. Where did you hear about the product? (N1)
V6. Where did you hear about the product? (N1)
V7. Where did you hear about the product? (N1)
The last five items on the sample survey are fill in the blank items. The variable formats in the
codebook specify the maximum response length for each of the items. The
actual distance between the opening an closing
bracket on the Email survey can be any length. If the respondent
"stretches" the space between the brackets by typing a longer
response, the data will be recorded properly in the data file provided the
answer does not exceed the field length specified in the codebook. If the
respondent's answer does exceed the field length, StatPac will allow you to
adjust the codebook so that all fields are sufficient to hold the responses.
Also note that the opening and closing brackets do not need to be on the same
line of the Email. Item ten in the example offers the respondent several
lines for their answer. Here is the codebook for the second five items
V8. Overall, how would you rate
our product? (N1)
V9. How old are you? (N2)
V10. What do you feel would be a fair price for this
V11. What is your favorite brand? (A45)
V12. What could we do to make our product better? (A500)
If a respondent prefixes a numeric response with a
dollar symbol (as might readily happen for V10), the dollar symbol will be
eliminated from response when importing the Email file.
The basic text for an e-mail survey can be created by
selecting Design, Email Survey. This will provide a foundation for the
body of your email. It is expected that you will edit the text as necessary
for your specific application and audience.
Capturing a Respondent's Email Address
To capture the respondents' Email addresses, include a
variable in the codebook that has the name EMAIL. Give it a format sufficient
to hold all Email addresses (e.g., A70) The
EMAIL variable may be placed anywhere in the codebook. Do not indicate
anything on the Email survey itself. StatPac will properly capture the
respondent's Email address when ever the EMAIL variable name is specified in
Unlike plain-text Email, CGI Email cannot automatically
capture the respondent's email address. Therefore, in order to capture the
respondent's Email address, and you must explicitly
specify it as a field on your web page.
Email to a Mailbox
All e-mail programs allow you to automatically or
manually send a received e-mail to a mailbox or text file. Automatic
"filtering" is often used to sort the incoming Email into a series of
mailboxes. These mailboxes are simply text files containing all the Emails
that have be routed to the mailbox. The naming
convention for the mailboxes is different for each Email program. Eudora and
Outlook Express programs use a .mbx extension.
The first step is to determine the name of the mailbox. That is, the name of
the file that contains the completed surveys (i.e., that. holds the filtered
Emails). This is the file that you will import into StatPac. If you do not have an Email program that
has filtering to mailboxes, you could import the incoming Email survey's into
StatPac one at a time from the clipboard, but this might be slow. A better
technique might be to use notepad to copy/paste all the Emails into one file
that can then be imported into StatPac.
Considerations for Plain Text Email
Because of the wide diversity in e-mail readers, we
suggest that you limit every line of your e-mail survey to 60 characters.
This will avoid unintentional word wrap. Create your email using a
non-proportional font (e.g., Courier) to make it perfectly clear as to you
how long a 60 character line will be.
We suggest that HTML tags not be used to enhance the
appearance of your e-mail. Many e-mail readers do not recognize the control
sequences used by HTML tags, and these potential respondents will see control
codes as part of their e-mail. Make your e-mail surveys generic looking so
they will display properly with all e-mail readers.
Before actually sending out the surveys, send several
copies of the e-mail survey to yourself. Make sure the surveys look
exactly the way you want the respondent to see them. Then reply to each of
the surveys. Now check your mail to make sure that they get filtered to the
correct mailbox. and check the import procedure to
make sure that the data is being properly imported. Checking the entire
import process is crucial for e-mail and Internet surveys.