When beginning a project, you must decide which type of research is the best fit. You can use exploratory research to study a topic and identify a problem. This leads into constructive research, i.e., following identification of the issue, you can look into new solutions. For example, exploratory research might involve researching trends and statistics in energy consumption over the last 50 years. The constructive research would be a search for ways to reduce energy. Through the use of direct observation, an empirical researcher could determine the potential success of these ideas.
Once you have finished your research, you can turn that information into quantitative statistics. As statistical data, you can represent knowledge through numerical information that can in turn be analyzed through charts and graphs to study distribution. A common tool is the bell curve, which can be used to determine norms. By plotting the information in this way you make it possible to spot outliers and trends, both of which can be used to identify a potential area of interest.
A correlation is a value showing the degree of relationship between two variables. This commonly used statistical analysis technique is often used in medicine, e.g., to determine whether smoking habits are related to lung cancer. This brings a critical aspect to mind: causality. Although a correlation does not always indicate causality (e.g., many girls under 150 pounds watch "Jersey Shore," but that does not mean that watching "Jersey Shore" makes you skinny), it is important to determine if any causality exists, and how.
Probability and trend analysis involve the practical application of research and interpretation to finding more data and solutions. Probability refers to the likelihood of an event occurring, based on existing data. It can be used to predict and prepare for certain outcomes. Trend analysis is less bound in numerical data and often involves the written interpretation of patterns that emerge in statistics. It is most appropriate for studies over a long period of time, or for trials undertaken multiple times.