A two-income family today is poorer than a one-income family was in the 1970s

[mathematic]

77. You really should read the CPI documentation

I'm kinda surprised you haven't considering you were tested on it. Here's the link:
http://www.bls.gov/opub/hom/pdf/homch17.pdf

The basket of goods is often not representative of any particular group. It's a composite. A classic example is that the elderly have a different spending pattern than the general population. The CPI-U doesn't claim to represent the changes in prices for them. The BLS does publish the CPI-E which is an inflation index weighted to reflect the expenses of that demographic.

The basket of goods itself is constantly being updated and it's based on the Consumer Expenditures Survey by the BLS. Originally this was once a decade but starting early last decade it's every 2 years.

When no price for item in the CPI sample can be found, the BLS looks for a new item to replace it.
There are three possibilities to determine the price change:
1. The prices are directly compared
2. The value of the difference in quality between the items are estimated, chiefly through:
a. Manufacturer cost data
b. Hedonic regression
3. The new item price is imputed, for cases where the data is missing/uncollectable.

Substitution effects are only applicable for very specific expenditure categories. An example is that one type of steak can be substituted for another type of steak but not for ground beef. The substitution effect is implemented via a geometric formula.