If they say that 50% percent of marriages end in divorce, exactly how do they measure that?

Do they look at all marriages since a certain date, then see how many have been ended naturally (death of a partner), and compare this to the number of divorces? What about marriages that are are still intact? Are they left out of the equation or are they counted as no-divorces?

I cannot answer your question as I lack the knowledge to do so, but I'm responding just to say that this is what I find rather annoying with feminist & other studies. They do not explain how they developed their answers and so we're left to assume full integrity on the part of the campaigner who undertakes the role of collecting the data.
We've all heard, now that it's too late, how some feminist studies purposely 'forgot' to include some important areas in their studies, so it begs the question - who else is copying this manner of data collection?