Saturday, March 23, 2013

Small Correction & Clarification Related to Earlier Post RE Matthew Johnson's Paper

For any of you following the rather detailed discussion about the impacts of relationship education services as tested in the large, federal studies here in the U.S., I just made the following correction to that long blog post on 3-11-2013.

The correction is this:

Correction/Clarification added 3-23-12:  In looking deeper into the technical report for BSF at 36 months, Mathematica did test for moderation by African American identification in the overall BSF pooled sample at 36 months, as they did at 15 months. But they did not test for this moderation by site at 36 months. Matt Johnson had implied that this moderation became null at 36 months when it had been significant at 15 months. As applied to the full sample, Johnson is correct. As applied to the Oklahoma site, my assertion that this was not tested is correct (Rob Wood, personal communication, 7-25-2012).  As has been clear here and elsewhere, I have focused on the Oklahoma site within the BSF study because of their much greater delivery of services to the intended participants compared to the other 7 sites. At 15 months, moderation by African American identification was tested and found to be very strong in the Oklahoma site, with African American couples benefitting the most. But on the question as to if this moderation maintained or not in the only site with evidence of an important lasting impact at 36 months (and if it holds or not for the family stability finding), we do not know the answer because, to the best of my knowledge, it has not yet been tested.

The entire post with links to all of the articles has been updated with the clarifying note above, and remains at this download link for readers here who are deeply interested in this subject. There are no other changes to the document.