Any statistic that admits a margin of error of 100% is absurd.
First off, there is nothing that remotely relates to a 100% margin of error. The statement simply states that if a copuple cohabitates prior to marriage, the likelyhood of divorce is 50-100% greater. This can in no way be confused with a "margin of error".
Here is an example (I did not read the article, but am addressing only the faulty logic of your statement). If marriage partners who did not cohabitate prior to marriage end up divorcing 30% of the time, then a 100% increase (a 60% divorce rate) is certainly possible.