Social scientists, like their colleagues in natural sciences, use a framework of laws to investigate the social world where researches are placed. Findings from the social science research should not contradict those laws, otherwise they are erroneous. The findings have also to be backed by a theory, or they may refute a theory.
The theoretical basis used in a research will have an impact on the produced hypotheses and the used methods. The interpretation of findings is influenced also by the focus on conflict, or consensus, on language or material existence, on constructionism or on realism. However, theoretical disagreements like these do not make the empirical results less adequate.
One of the main creative processes in science is the development of hypotheses. One of the most important methods of hypothesis generation involves the logical deduction of expectations from some established theory. Theory has its place in social science as it contributes to the development of hypotheses and this expands our knowledge of social behavior.
Another method for hypotheses development in social sciences involves conflicting findings. This method helps to precise theory by strictly specifying the conditions under which a certain outcome can be expected.
From a historic point of view, the social sciences are not far from the speculative character of their initial development. A key process that marked the transition to objectivity was the strong focus on operationalization which means the translation of abstract theoretical constructs into concrete procedures. Research procedures have to be described and this brings more clarity in theory and underscores the scientific character of social sciences. The theory determines what will be variable and what constant. Variables are characteristics that can be measured such as social class, age, sex, opinion and attitudes.
The categories used by social scientists could never be mutually agreed. The social scientists are interested in differences of several percentage points between categories in a variable. For example in the 1991 Census, in England and Wales, 72 percent of White people and 53 percent of Afro-Caribbeans lived in owner-occupation. However, in such cases scientists consider all other things a as constants, but the analysis is still useful as it provides a hypothesis that Afro-Caribbeans are more likely to have more housing problems than whites.
Multivariate analysis refers to the examination of several variables at the same time. Techniques used for this purpose are simple regression, log linear analysis, stepwise regression and logistic regression. After conducting the measurement, social scientists explain the results. The scientists are able to explain the way variables are constructed and the potential margin of error in any relationship. The reasoning is transparent.
However, research in many social disciplines does not aim to test theory as the theoretical background is often very loose. One of the widely used approaches is survey, which is often associated with naturalism in social sciences. Its two major forms: large scale surveys, such as the Labor Force Survey in Britain; and large-scale one-off surveys, focus on a certain topic. An example for the latter type of research survey is the National Survey of Sexual Attitudes and Lifestyles, carried out in Britain in the early 1990s. The purpose of the study was not to test any theory, but rather to examine a wide range of sexual practices and beliefs about sex.
Social sciences use statistics and statistical analyses as one of the key methods of research.
Statistics as a methodology are widely used in psychology, sociology, anthropology and political sciences. The wide use of statistics has led to the rise of quantitative social science. Quantitative social sciences use a variety of methods and techniques such as factor analysis, multilevel models, cluster analysis, item response theory, survey sampling.
Factor analysis is used to cut a big number of variables into fewer numbers of factors.
Multilevel models are placing all the components in a hierarchy of nested effects. For example, the behavior of students is influenced at several levels. Children in one class will have had similar teaching experiences. Classes in the school will have had also similar experiences and the school may be similar to other schools in the town.
Clustering is the classification of a group of observation into a subgroup, called cluster. Observations in one cluster have similarities. Clustering is mostly used in statistical data analysis.
The item response theory (IRT) is used for the design and analysis of questionnaires related to abilities and other variables.
Survey sampling refers to the selection of a sample of people to carry out a survey among them.