differential in dissolution rates

03:14 Feb 24, 2009 

No to changing differential..

13:38 Feb 23, 2009 

I also meant to note: one of the tricky things about studying trends in divorce rates is that couples who got married longer ago are more likely to have divorced due to the simple passage of time--they've had more time to get divorced in! So based on the fake data above, you can't necessarily conclude that couples married in 1990 are "less likely" to divorce than couples married in 1970. You'd have to cmpare the 15-year survival rate for each cohort, for example, to draw that conclusion.

To Ms. Finch: I interpret the "even" in the sentence as injecting an element of surprise, not uncertainly.

even (adv.)
c. Used as an intensive to indicate something that is unexpected: declined even to consider the idea.
http://www.thefreedictionary.com/even

13:38 Feb 23, 2009 

by the way, is it possible to use "differnce" instead of "differential" in this context?

13:33 Feb 23, 2009 

I think that "has even reversed" is an unfortunate phrase to use in statistical writing, as it implies uncertainty. Either the difference has declined (moved towards zero), OR it has reversed (gone past zero). So it may be that some studies show a decline while others show an actual reversal.

13:16 Feb 23, 2009 

part of my answer got cut off, and I can't seem to add an additional note either, so I will finish what I was saying here:

married in 1970:
cohabited: 56% divorced
no cohabiation: 46% divorced

married in 1980:
cohabited: 47% divorced
no cohabitation: 42% divorced
(differential between groups is less for this cohort--5%--than the earlier cohort--10%)

married in 1990
cohabited: 35% divorced
no cohabitation: 37% divorced
(differential between the two groups has *reversed*--now it is the cohabiting group that has the lower divorced rate)