A hypothesis is a prediction or an idea that has not yet been tested.
Researchers must attempt to identify everything that might influence the results of an experiment, and do their best to neutralize the effects of everything except the topic of study.
Since social scientists do not seek to isolate variables in the same way that the hard sciences do, sociologists create the equivalent of an experimental control via statistical techniques that are applied after data is gathered.
A control is when two identical experiments are conducted and the factor being tested is varied in only one of these experiments.
Used loosely, a tentative conjecture explaining an observation, phenomenon, or scientific problem that can be tested by further observation, investigation, or experimentation.
To conduct an experiment, a scientist must be able to control experimental conditions so that only the variable being studied changes. For example, a scientist might have two identical bacterial cultures. One is the control and it remains unchanged. The other is the experimental culture and it will be subjected to a treatment, such as an antibiotic. The scientist can then compare any differences that develop and safely assume those differences are due to the effects of the antibiotic. But social life is complicated and it would be difficult for a social scientist to find two identical groups of people in order to have both an experimental and a control group. Imagine, for instance, that a sociologist wants to know whether small class size improves academic achievement. For a true experiment, she would need to find identical students and identical teachers, then put some in large classes and some in small classes. But finding identical students and teachers would be impossible! Instead, the sociologist can statistically control for differences by including variables such as students' socioeconomicstatus, family background, and teachers' evaluation scores in a regression. This technique does not make the students or teachers identical, but it allows the research to calculate how much of the difference in students' achievement is due to background factors like socioeconomic status and how much is actually due to differences in class size.
Scientists form a hypothesis, which is a prediction or an idea that has not yet been tested. In order to prove or disprove the hypothesis, scientists must perform experiments. The experiment is a controlled test designed specifically to prove or disprove the hypothesis . Before undertaking the experiment, researchers must attempt to identify everything that might influence the results of an experiment and do their best to neutralize the effects of everything except the topic of study. This is done through the introduction of an experimental control: two virtually identical experiments are run, in only one of which the factor being tested is varied. This serves to further isolate any causal phenomena.
Of course, an experiment is not an absolute requirement. In observation based fields of science, actual experiments must be designed differently than for the classical laboratory based sciences. Due to ethical concerns and the sheer cost of manipulating large segments of society, sociologists often turn to other methods for testing hypotheses.
Since sociologists do not seek to isolate variables in the same way that hard sciences do, this kind of control is often done via statistical techniques, such as regressions, applied after data is gathered. Direct experimentation is thus fairly rare in sociology.
Scientists must assume an attitude of openness and accountability on the part of those conducting an experiment. It is essential to keep detailed records in order to facilitate reporting on the experimental results and provide evidence of the effectiveness and integrity of the procedure.