Seasonally Adjusted Data

After reading a blog from the vice president of the Atlanta Federal Reserve's Research Department, I have decided to try to fully understand seasonal adjustments. Before I dive into the technical aspects of it, I've been first working on how to conceptualize it. My thoughts were that if I were able to understand the basic intuition behind seasonal adjustments that I would see exactly how it can skew data reported under atypical situations (like a recession).

The first Thing to look at is how the data gets adjusted and what that adjusted data represents. Seasonally adjusted data is designed to show the reader exactly how the variable in question changes due to changes in all other variables accept the seasonal effect. A seasonal effect is characterized by a change in some variable due entirely to the change in time, usually months. A good example is looking at the monthly sales figures for Macy's . It is no surprise that during the end of November through the December those figures are going to be extremely large relative to the rest of the year due to increased shopping for the Christmas season. If we were to report the unadjusted sales figures for December we can pretty much bet that they would always be a positive number and that they will always be larger then every other month. However, say your an investor and you want to know how Macy's is doing relative to the same period last year, can you simply look at last years December sales figures and compare then with this years? Absolutely, but now you would need 2 different data points to tell you one thing. Also this is only going to give you the change between 2 years, what if last year they had record high sales? This years data point will be negative unless they have another record year. But lets say that they do very well relative to every year accept last years record year, the variable we will get for this yearly change will not represent the large picture of how Macy's is preforming overall.
How about relating Decembers sales figures to the inflation adjusted average of all other Decembers? This is basically what seasonal adjustment does. It allows us to report a figure, not a percentage change, that represents the performance of a variable (in this case Macy's December sales) relative to structural and economic effects without the "noise" that is created specifically by time.

Questions for further research:
1) How do these adjustments work during atypical times?
2) If we base our analysis of the economy on seasonally adjusted data, what is the affect it can have on consumer confidence? On the economy as a whole?
3) What model does the government agencys that are widely sited in popular media use to compute seasonally adjusted data?
4) Do we need to make an adjustment to the way they are computed in an attempt to take into account larger cyclical changes like recessions, or does this already take place within the widly used models?