The National Bureau of Economic Research has the NCHS's vital stat data for marriages and divorce on their site!
NCHS's Marriage and Divorce Data 1968-1995
http://www.nber.org/data/marrdivo.html
"Marriage microdata includes variables such as date of marriage, state residency, education, previous marital status, number of marriages, and ages of bride and groom. Divorce microdata include variables such as number or children under 18, custody, month and year of marriage, marriage duration, number of marriages, age, race, and state residency of husband and wife."
For a time series, I guess would need to count the entries. Say add up all the marriages in a given year, then divide by the population. That means I would need population numbers as well. Unless, I could just use the number of marriages. Would that be acceptable?
[NEAS: Identify the ideal data. Marriage rates per capita are needed to compare different states. The U.S. census publishes populations by state every ten years. In a single state, or for the U.S. overall, per capita figures are easier to compute.
Few student projects use ideal data; you are always limited by the data that are available. The student project is an application of the course concepts to real data. In the write-up, explain what data are available and how other factors (that are not easily measured) might distort the results. Then do the student project with the data you have.
Most data can be read into Excel, though you may need to spend some time reformatting the spreadsheet.]