They don't release their exact statistics, but Conversion Rate Experts are pretty authoritative on the subject of scrolling vs. fold.

http://www.conversion-rate-experts.com/scrolling-tips/

  In split tests, long pages often beat shorter pages.

  As a rule of thumb, your page should contain at least as 
  many words as you’d use when selling your product or
  service face to face.  

  The winning page we designed for SEOmoz was six times 
  longer than the control, which it outperformed by 52%.

  Whenever someone tells you that they’d never buy from a 
  long page, remind them how long Amazon’s pages are.

  Long pages are effective, but only if your users know that
  they can scroll and are given compelling reasons to do so.

http://www.conversion-rate-experts.com/crazy-egg-case-study/

  The redesigned page is about 20 times as long as the
  control... new page outperformed the control by 30%.

http://www.marketingexperiments.com/improving-website-conver...

  In our initial micro-test, long copy outperformed short 
  copy by 40.54%. Click-through traffic sent to the short
  copy page was unprofitable (-14% ROI), while traffic sent
  to the long copy page produced an ROI of 21%.

http://conversionxl.com/4-cases-where-short-home-pages-outpe...

  Short copy performs better when there is low perceived
  risk, low cost, and low commitment... when the customer 
  has an emotional, impulsive, and “want-oriented” 
  motivation. When you want people to take action that 
  doesn’t cost them dollars (email signups, free trial,
  clickthrough)

http://blog.clicktale.com/2007/10/05/clicktale-scrolling-res...

  Visitors are equally likely to scan the entire page no
  matter the page size.

  Below 540 pixels, both visitor attention and page exposure
  decline exponentially.