[buildingsramen]

This is poorly reported. From the article:

* Group A reported a 26% increase in participants staying in housing they rented or owned;

* Group B reported a 35% increase in housing they rented or owned;

* And Group C, which received the smaller sum of money, reported a 20% increase in renting or owning housing.

But these are not actually percentages - these are percentage points!

* Group A went from 8% to 34% - a 4.25x increase

* Group B went from 5% to 40% - an 8x increase

* Group C went from 11% to 30% - a 2.7x increase

Which is a lot more dramatic than saying 20% vs 26% vs 35%.

The study authors might have good reason for reporting it this way, such as many participants dropped out of the study before follow-up, which probably skews toward people with more chaotic lifestyles. So the denominator for each group has changed at follow-up and could at least partially explain why all groups saw improvements.

[itsdrewmiller]

Reporting it in the more dramatic way puts a lot of emphasis on the starting conditions of what were supposed to be randomized groups - I think it would be more misleading the way you propose. If you invert it and report on the % reduction in the bad thing, the numbers would be almost the same as those they report.

The part about people dropping out is complete speculation, right?