Can I clarify a couple of things? Lynne Murray says that of course SH tenants did worse, because those occupying social housing are either poor or vulnerable. But the point about this research is that it takes into account people's characteristics, and finds that even when you take all of the characteristics for which data exist, a person living in social housing as a child does worse on a huge range of measures than a person who has otherwise identical characteristics and does not live in social housing as a child. As far as the researchers can tell, the difference cannot be explained by overcrowding, or by neighbourhood effects. The figures suggest that it is social housing per se.
Clearly in an ideal world we would "know" the answer. But in social sciences we can never "know" the answer - how can we? All we have are correlations, and we have to take those correlations seriously.