However, Smith et al. suggested that offering metacognitive reading instruction without modeling may be worse than offering no reading instruction at all. 

This goes on to describe a study with three groups of developmental college readers one group was taught metacognitive strategies, one group had the strategies modeled as well as taught, and a control group received no metacognitive instruction. The modeling group significantly outperformed the other two groups. The odd result was that the control group outperformed the group that had received only strategy instruction with modeling. 

This was surprising even though I have learned in other articles and discussions that offering strategies for reading alone has not worked in the past. My thought was that any intervention is better than nothing. 
The researchers speculated that students who might otherwise develop their own strategies through trial and error were confused by the strategy instruction. This underlines the importance of modeling and the need to communicate caution about teaching the strategies without modeling. This makes a case for appropriate intervention based on the framework, with no shortcuts.