This study was conducted in 2006 by Jennifer M. Urban and Lauren Quilter, surveying the effects of Section 512 of the Digital Millennium Copyright Act on the Internet.  The two used an empirical approach to look at the notice and takedown landscape, and collected data about the number and type of notices that were sent in recent years.  Google provided all the notices the company had received between 2002 and 2005 (constituting the majority of the data), with non-trivial supplements coming from the Chilling Effects Clearinghouse.  The researchers were careful to point out a variety of issues with the data set, including a potential bias in the Chilling Effect notices, since these were self-reported cases.  The Google information also is flawed to a degree, since notices sent to a search engine like Google are not necessarily emblematic of the entire notice and takedown climate.  This is displayed by a discrepancy between the data and common perception, with music and movie companies accounting for few of the takedown notices, since they find it more useful sending takedown notices to non-search engines.  Acknowledging the need for additional data and further research, the study concluded that there a large number of claims had serious substantive questions.  While anticipating some notices to be unjustifiable, the high number of problematic notices that were found was “particularly troubling.”  Since the researchers used a high threshold of what would be considered questionable (choosing to use cases where fair use only could likely be used as a proper defense) the results are even more severe than first appear.  Even so, enough claims were made without sufficient justification or sometimes without any at all (claims regarding material which are not subject to copyright) for the study to conclude that the “implications for expression on the Internet of this extrajudicial process appear, from our limited data, significant.”
    This is going to be very helpful in my paper, since it will be one of the few but important statistical analyses I use.  Many of the other works are theoretical expositions by professors and academics, citing specific cases and expanding out the reasoning to apply to more generic cases.  However, this study uses nearly 1,000 data points to arrive at its significant conclusions that will aid me in my argument.  Most importantly, I will reference the high rate of improper claims, representing the low barrier to entry to submit even a fraudulent claim, and its negative impact on free speech on the internet.