Formulate the hypothesis you intend to test in your study. Define one or more variables of social conditions or behavior and an associated, or correlated, effect that you expect to find to be caused by different levels or combinations of your variables. For example, you might expect higher rates of literacy in some socioeconomic classes or in certain kinds of neighborhoods compared to others.
Learn what other scholars in your subfield have already found out about your subject population (such as the people in particular neighborhoods) or as effects of the same variables. Gather this review of the literature to demonstrate that your hypothesis has not already been tested --- or has not been properly tested --- and to show related findings.
Determine whether a suitable database of information for testing your hypothesis already exists. One of the most widely used U.S. databases is the U.S. Census, collected every 10 years. Some of the other studies you've reviewed may mention databases they used or created, and they may be available to you under certain conditions. If you will collect new data, describe how you intend to go about it --- who will actually conduct your surveys or records searches, what kind of questionnaire or other instrument they will use or whether you will use any instruments already validated by other research.
Define how you intend to sample your subject population, whether from an existing database or to collect new data. It's rarely feasible to use every record in an existing data set and even harder to collect a large new database. Sampling is the process by which you decide which parts of the available population you will study. Your study might base sampling for new data collection on the inventory of residential addresses assembled for the most recent census. The basic principle is to make sure the sample fairly represents the total population by giving every member of the total population an equal chance of being selected for the study.
Describe the methods you will use to analyze your data. What statistical measures will you use to test your hypothesis?
Develop a detailed budget covering the costs for use of existing databases or new data collection and for analysis of the data. Set a time budget as well.