Embedded or nested-case research design is especially well suited for descriptive studies, which are focused on understanding the general situation surrounding a phenomenon, and uncovering possible correlations between that phenomenon and others. For example; the aim of a descriptive study on smoking might be to discover whether there are any habits, environmental factors, or personal characteristics that correlate to smoking. It could simultaneously answer multiple questions such as: Do smokers watch more movies than nonsmokers?; Are smokers more likely to own goldfish than nonsmokers?. The aim of descriptive studies is to find out as much as possible about the circumstances of a situation, and embedded studies aid in that by basically allowing researchers to look at more than one independent variable at a time. Correlations gleaned from descriptive studies are typically studied further in more focused, controlled, clinical studies in hopes of determining causation.
Because researchers can use one group of subjects (or cohort) to study multiple variables and possible correlations, time and money spent on subject recruitment can be significantly reduced. Even time spent on the very process of data collection may be reduced, since all the necessary tests for a given time period may be given to the subject within one visit.
If you, as a researcher, focus your study on only one cohort, you do not have to worry about the many confounding variables that could affect results from the general population. The smaller sample size can work for you, especially if the group you study is well defined. For example; you could conduct a descriptive smoking study on current patients in St. Ann's Hospital in Springville, Idaho. The control group for your study can be the nonsmoking current patients at that same hospital. In this way, confounding differences between the study group and control group can be significantly reduced, compared to a study where; for example, the study group is hospitalized and the control group is not.
Unfortunately, with such a well-defined sample, particularly if the sample size is small, the results of the study may not apply to any group outside the one studied. For example; correlations to smoking behavior for current patients of St. Ann's in Idaho may be very different than the correlations to smoking behavior for the entire population of the U.S. The results may even be different than the results you find from current patients at St. Ann's in Miami. Differences between the populations may be due to cultural influences, available resources, local environmental factors, or any number of other reasons. Whatever the reason, it is important to remember that any variable not specifically accounted for cannot be included in your sample.