Information Technology Reference
In-Depth Information
(Babbie, 2007). This, of course, implies that sampling strategies are important.
To be able to make a statement from findings about a given population with
some level of assurance, the larger the sample the better. In addition, the more
variables on which you would like to analyze a population, the larger your sam-
ple needs to be. For example, let's imagine that you are studying access to
healthcare. One of your research questions is to see if there are differences in
access to services. You might decide you want to analyze on a socio-economic
variable to see how access to healthcare services changes by income level.
When you begin to break the population into sub-levels of income, the samples
in each category decrease considerably. Let's say you have a sample of 500.
When you begin to break it by sub-levels of income, you may end up with only
20 or 30 in your upper and lower categories, depending on how you decide to
break down income levels. These small numbers in some of the categories will
seriously constrain your analysis. As a rule, the more complex the analysis, the
larger the sample size should be.
Another aspect of traditional surveys that facilitates the ability to gener-
alize to a larger population is a standardized questionnaire (Babbie, 2007).
Everyone participating in the survey needs to answer the same questions.
Questionnaires can be broad. But it's important to be mindful that if it is
too broad it presents a limitation for the meaning of an answer. In other
words, the broader the question, the less meaningful the answer will be.
Questions can also be designed to test an idea or conceived notion of the
researcher, based on his or her own analytical work before designing the
survey instrument.
A limitation of survey research, however, relates directly to question
design. Sometimes what we think is important from our perspective may not
be what is actually important to others. Designing an instrument that relates
to our perspective runs the risk of missing valuable information and insights
about a subject. In addition, “surveys cannot measure social action; they can
only collect self-reports of recalled past action or of prospective or hypothet-
ical action” (Babbie, 2007: 277). One of the challenges in survey design is to
be clear whether or not the topic is measurable through a survey instrument.
It is also important to note that it's possible that, by studying a particular sub-
ject, there may be an influential effect on a person's opinion (Babbie, 2007).
A person may not have formed an opinion until challenged to think about a
particular topic in a particular way through the research instrument. These
are among several reasons why some researchers prefer qualitative research
to survey methods.
Typically, when researchers speak of the ability of their questionnaire to
measure a particular social phenomenon, what they are referencing is the
concept of
validity
. When speaking of a high level of validity, researchers
are making the statement that their questionnaire measures what it was
meant to measure. In other words, the findings from a questionnaire can be
