I agree that without getting into the details you can be misled. "Cities" are defined many ways: city, county (same as city in SF but not in LA or SD), metropolitan division, metropolitan statistical area and combined metropolitan statistical area depending on your data source. Frankly it is a big mess and often you are not told which definition is being used. Not only that the definitions change over time! I had to deal with this data for many years - not simple at all.

Which one is most appropriate depends on your goal. La Jolla or Mission Hills are probably a lot more like SF than the SD MSA so likely perform more like SF but will still be tied to SD.

Unfortunately sales only occur infrequently and tend to dry up completely in recessions. This along with lack of data as well as different definitions make tracking detailed property types like 4br 3ba Spanish in La Jolla difficult and unreliable if not impossible. And in SF (city & county) that property type is very rare (also SF does not publish style data like some area - I think SD might). Best you can usuallydo is by bedrooms in a zip but only if the market is pretty active.

My original point was a simple one: even though real estate moves together across the country with the business cycle over time area can still diverge significantly. This has happened in SF vs LA/SD and in California as a whole vs the rest of the US. To see this look at the index levels over longer periods. Maybe go with a big areas like MSA/CMSA so you have lots of data and get both mature inner city and fringe suburbs - this is what Case-Shiller public indexes are like.