A social whole -- the city of Chicago, for example -- is a densely various empirical reality. At virtually every level of scale there is variance with respect to social characteristics -- income, health status, ethnic or social identity, political adherence and preference, age, race, or occupation. Neighborhoods differ from each other -- but equally, we find variance within neighborhoods as well.
Given this fact of radical non-homogeneity of social characteristics, what is involved in arriving at knowledge about such a reality?
It is evident that we are forced to arrive at generalized descriptions, at some level of scale or granularity. It is neither feasible nor explanatory to provide a "fully" detailed description of a population, individual by individual. Instead, our challenge is to arrive at some ways of segmenting the population into groups that will prove felicitous in revealing causal connections among attributes or circumstances. Groups may be defined with unlimited range: geographically, occupationally, racially or ethnically, educationally, politically, ... We can then observe and measure the distribution and means of various characteristics across these groups (attitudes towards the Patriot Act, for example) and we can consider whether there are meaningful differences across groups with respect to these characteristics. Finally, we can try to find causal explanations of these differences. Are Arab-Americans more distrustful of national security laws than Asian-Americans? Are poor people more prone to asthma than affluent people? Are doctors more favorable to higher taxes than skilled-trades workers? What factors might causally explain these differences?
The point here is that social knowledge requires recognition of the inherent heterogeneity of social phenomena and a fertile effort to find ways of segmenting this heterogeneous reality that shed light on social causation and patterns of behavior. And, importantly, it is important to recognize that any level of granularity of analysis could be further partitioned and more fully described--sometimes with important insight.
There is no "fundamental" or "optimal" level of analysis and description that captures the whole of Chicago. Instead, anthropology; sociology, and political science can continually pursue the upward and downward research journey of discovering meaningful group-level patterns or regularities, and pressing into a deeper understanding of the diversity of the phenomena under study.