Most social research starts out with a hypothesis or research question.
Example:
What happens to a certain behavior if ---- or what causes people to react a certain way. In most
social research projects there are two hypothesis, the research hypothesis (which is the one the
researcher wants to support) and the null hypothesis which the researcher wants to reject. The
null hypothesis proposes that there will be no change in behavior due to the researchers
experiment or that the research question is not valid.
Research Hypothesis: People in North County are more likely to vote for a republican candidate
than a democrat.
Null hypothesis: People in North County are more no more likely to vote for a republican
candidate than one who is not republican.
After you decide on the research question or hypothesis you need to determine how to test it.
This is usually done by examining variables. A variable is a characteristic that varies from one
person to the next or from one point in time to the next. Ex: age, social class, population, crime
rate. Though some variables (called constants) such as sex will not change.
Social scientists are mainly concerned with dependent and independent variables. The
independent variable is the factor that is controlled/tested or manipulated by the researcher. It
can also be the presumed cause of some problem or existing condition.
The dependent variable is the effect caused upon the subject or groups of subjects, which are
being studied. Ex: One may hypothesize that 1-parent family structure leads to greater
delinquency than a 2-parent family structure. To test this hypothesis one might interview
samples of delinquents and non-delinquents to see if they came from 1 or 2 parent homes. In
this example, family structure would be the independent variable and delinquency (the young
persons reaction to family structure) the dependent variable.
Social researchers use many different ways to investigate social problems. These are the
experiment, the survey, content analysis and participant observation. Each is useful and has its
own application.
In an experiment the researcher actually manipulates one or more of the independent variables
that the subject is exposed to. The manipulation occurs when the subject is exposed to one or
more conditions and tests the subject’s reaction to the condition they are assigned (Ex: One
group of students get mentors to help them in school and another does not). The researcher
measures the degree of change in gpa/attendance/behavior of the group with mentors (the
experimental group) and the group without mentors (the control group).
In contrast, survey research is a type of research that is not manipulated by the researcher. It is
retrospective because the effects of the independent variable (neighborhood density) on the
dependent variable (crime) or family size (1 or 2 parent) and delinquency) are recorded after
they have occurred. Surveys attempt to reconstruct these influences and consequences by
means of verbal responses to self-administered questionnaires, face to face or telephone
interviews. Since surveys lack the same controls as experiments, it is harder to determine cause
and effect. Additionally, because they don’t involve experimental manipulation surveys can
investigate a larger number of independent variables in relation to dependent variables and
generalize to a broader range of people.
Content analysis is a research method that studies and describes the content of previously
produced messages. Ex: A content analysis might look at how books magazines and
newspapers of a particular era control public opinion or even manner of dress and what topics
are particularly important in what era.
Participant observation is a method in which the researcher actually participates in the lives of
the people they are studying either openly or covertly. Ethically is best to be up front about this
but may not be possible with certain populations. Participant observation is useful for it provides
in-depth information about a situation or series of events as defined by the people involved in the
situation. This is important since outsiders have a different and often under-informed
perspective of other people’s lives.
Now that we know what types of methods we can use to conduct our research we can talk about
how we can measure our data once it’s collected.
The data you collect will be measured on one of four scales: nominal, ordinal, interval or ratio.
A nominal (the word means name) scale of measurement is the simplest form of measurement.
For example..are you an introvert or extrovert? rich or poor? democrat or republican? When we
collect data based on a name or category we are collecting nominal data. The problem with the
nominal scale of measurement is it can’t be used to record anything but qualitative comparisons.
Knowing you are an extrovert or introvert doesn’t let me make a qualitative comparison because
it doesn’t tell me how introverted or extroverted you are.
Ordinal scales are the simplest of the quantitative scales. They require that the data is ranked
from highest to lowest. They tell whether one subject ranks higher or lower than another, but don’
t not tell you how much lower or higher. Ex: social status tells you if one is lower, middle or upper
class but does not give the degree between intervals like an income level scale would.
An Interval scale of measurement not only tells us about the ordering of categories, but also
indicates the exact distance between them. Interval measures use constant units of
measurement (ex: yards, feet, minutes or seconds)
A ratio scale has all the properties of the scales just mentioned but also has a meaningful zero
point. A point at which there is a total absence of the variable being measured. Most of the
research you will conduct will be based upon the interval or ratio scale of measurement. No
matter what scale we use, they all offer a way to measure some type of behavior.
Statistics are also useful for generalizing findings from a small sample to a large population with a
high degree of confidence. In this course, we will learn how to use statistics to tell if the results of
our small samples are due strictly to chance or sampling error (sampling a non-representative
population).
Back to HPS 620