And of course she's referencing to 2009 General Social Survey: Victimization, where it is written that:
For the eight crime types covered by the 2009 GSS on victimization, the proportion of incidents reported to the police by respondents fell from 34% in 2004 to 31% in 2009.
Except that, I think she's wrong about this being contrary to Stockwell, for if the proportion of incidents reported goes down, that means the number of unreported incidents goes up. Which is what Mr. Day seemed to be suggesting.
For an account of earlier rounds, read this. And, mind you, this still doesn't justify building more prisons as unreported crimes are by definition ones in which the criminal was not brought to justice.