According to Dr. Janet Smith's lecture "Contraception Why Not" the Family of the Americas Foundation performed a study regarding rates of divorce and the use of Natural Family Planning techniques (as opposed to oral contraception, sterilization, etc...).
The statistic of 5% divorce rate for couples using NFP is what I commonly hear and read.
Perhaps this is why divorce rates for NFP users are between 1/10 and 1/25 of the overall divorce rate in the United States in the 90s.4 Indeed, a study conducted by the Family of Americas Foundation found only 16 women ever divorced among 505 NFP users, a rate of 3.6%!5
According to the text of the study, this is the first study of it's kind. Was it so conclusive that it was also the last study of its kind? Does Natural Family Planning really dramatically reduce the chance of a couple to get divorced?
Here's what the USCCB (United States Council of Catholic Bishops says the correct way to do the study would be:
if a researcher wanted to determine whether there is a difference in divorce rates among married couples who use NFP versus couples who use oral contraception, the study would be designed to track first time NFP and first time oral contraception users over a 10 to 20 year period. But what if a researcher did not have the time or money to track couples over a long period of time but still wanted to investigate the same issue? A "cross sectional" longitudinal study could be conducted instead. To do this study a large sample of married couples from NFP centers that have records from the 1970s and early 1980s would be needed. Next, a comparison of that data with a large sample of married couples who attended family planning clinics to obtain oral contraception in the same time period would have to be done. Once large enough samples are obtained from each grouping, an interview (by phone, in person or through mail) of the participants would have to be conducted in order to determine which of the couples divorced and the frequency of doing so.
Is that sufficient, does the Family of the Americas Foundation even come close?
Divorce rates among Catholics (who make up the majority of the study) have been shown to be similar to other sects. An easy assumption is that this study just proves that faithful Catholics are far less likely to wind up divorced, but that should have no bearing on the question. But if correlation doesn't imply causation, does that assumption even have any merit?