“most strongly agree to most strongly disagree” and
places a scale of one through 10 in between. Figure 9-4
is an example of a Likert scale question.
Exclusion/Inclusion
When you write scaled questions and especially
multiple choice questions, beware of overlapping
categories. This usually entails questions asking for a
numerical determination.
For example, suppose you want to determine how
often the base ticket office is used by base civilians. Your
survey question is multiple choice and asks the
respondent to fill in a block that fits into the categories
one to three times a month; three to five times a month;
five to seven times a month, and so on. The respondent
might have used the ticket office three times last month
and now must make a choice to put his or her response
in the first or second block.
When tabulating the responses, you will not be able
to know accurately, in which block the person should
really be. It would have been better to set up the choices
as: never; one to two times a month; three to four times
a month, and so on. Remember, the person who never
uses the facility is also a part of your general audience
and should have a place to put that information, as well
as the person who might use the facility more than a set
block might indicate.
RELIABILITY
The term reliability in survey research is a bit vague
as compared to reliability of research in chemistry for
instance. In chemistry, a quality control survey is
eliable when the same results appear time after time.
You have to be doing the same steps in a methodical way
to obtain similar results each time. To do the same survey
a number of times when dealing with human
respondents is certainly possible, but the material being
surveyed, human respondents, change minute by
minute. So checking for reliability becomes harder to
measure in a social science setting. With that
qualification being said, reliability still means having a
legitimate survey process that, given all other human
variables being the same, would produce similar results.
VALIDITY
The term validity deals with whether the
respondent’s real opinion is discovered or not. You may
have a wonderful survey process that has the questions
answered the same way on the pretest and the larger
complete sample survey, but the validity of the
responses comes into question when you realize that the
respondents had ulterior motives for answering some of
the questions the way they did. For example, what if you
asked your audience if they felt the government should
spend less money on defense. What if the vast majority
answered “NO.” Do they really feel the government
should not curtail defense spending, or are they afraid
property values in the area will drop if word gets out that
even the military people on base think less money
should be spent? The doubt of how valid the response is
to that question would certainly arise.
If all these potential problems in doing survey
research have you muttering about how we can ever
really know anything, welcome to the world of social
science research. All we can do is get as close to the facts
as we can through using as unbiased a process as
possible. Margins of error till always be with us when
Figure 9-4.—Likert scale.
9-9
