Define the research objectives. Decide on the scope of the research, meaning what is being investigated and how; and its relevance, meaning why anyone should care about the results. The scope often depends on the existing body of knowledge in the field, the available resources and the schedule. Research studies are usually funded by public or private sector donors who set the budget and the timetable for results. (See Reference 1)
Select the research type. The scope determines if the research design should be exploratory, descriptive or hypothesis testing. Research methods may be combined. Exploratory does not mean "first" -- for example, the first research study in a particular university on student-teacher ratios would not be exploratory because similar studies have been conducted elsewhere. The simplest descriptive study is a case, which studies one subject. The researcher observes the subject in different settings and records observations. Hypothesis testing studies may be analytical, which determines why or how things are associated, or predictive, which tries to establish cause-effect rules with broader applicability. The data collection methodology depends on the study and may include observations, interviewing, case studies, focus groups and surveys. For example, if you are researching managerial behavior, the descriptive method may be best suited because you must observe how managers act in different situations. (See Reference 2)
Design the sampling parameters. Statistical sampling is used in research because it is cost effective. For example, if you are trying to gauge voting intentions for an upcoming election, you cannot possibly ask the entire population. Rather, you are going to survey a random but representative cross-section. Identify the population, which is the group that is of interest to your research. Then, determine the sample size that will yield useful results and select the sample. Nationwide voting intention surveys, for example, often have a sample size of 1,000, and the sample is selected to be representative of the population. (See Reference 3)