What sticks or Does advertising really work?

Slate reviews the latest book on marketing - What sticks and finds its data-based approach a refreshing change from the usual use-anecdotes-to-get-the-point-across approach favored by similar books. Some interesting data-based findings from the review -

• A message repeated once on 3 different mediums is more effective than if repeated 3 times on the same medium. AKA the surround sound effect.
  
• Time of day matters (though it depends on the product being plugged, food advertisements around mealtimes for example).
• Show your logo throughout the advertisement and not just at the end.
  
• Intrusive, even irritating pop-up ads work much better than expected.

I've always found the anecdotal approach problematic and my usual argument to an anecdote is - Your sample size is too small. On a side note, I keep noticing books / articles / blogs written by women tend to be more anecdotal in nature (I'm going to wonder why in a separate post). Its important to differentiate between using anecdotes as data and anecdotes as illustrations or examples. Who was it who said that all statistics really start out as anecdotes?