Comment by sbierwagen

  Another health factor at work might be the unprocessed nature of the food they 
  consume: as Trichopoulou observed, because islanders eat greens from their gardens 
  and fields, they consume fewer pesticides and more nutrients. She estimated that the 
  Ikarian diet, compared with the standard American diet, might yield up to four 
  additional years of life expectancy. 

Oh boy, 4 years.

Eureka. You've done it. You've found the secret to eternal life.

  She also pointed out a preliminary study of Ikarian men between 65 and 100 
  that included the fact that 80 percent of them claimed to have sex regularly,
  and a quarter of that self-reported group said they were doing so with 
  “good duration” and “achievement.” 

If you had asked me the same question in middle school, I would have said that I was having all sorts of sex too.

  Although unemployment is high — perhaps as high as 40 percent — most
  everyone has access to a family garden and livestock, Parikos told me.
  People who work might have several jobs. Someone involved in tourism,
  for example, might also be a painter or an 
  electrician or have a store. 

This is a common pattern in the very poor. When you don't have any savings, you can't afford any interruptions in your income stream. Consequently, they're forced to be jacks of all trade, master of none. This makes it impossible to specialize in any one field, and lack of income means they can't make any capital investments.

  "When everyone knows everyone else’s business, you get a feeling of 
  connection and security. The lack of privacy is actually good, because
  it puts a check on people who don’t want to be caught or who do 
  something to embarrass their family."

I'd hate to be homosexual here, or be a member of any kind of minority group.

Anyway, the effect is almost certainly due to a "small study effect" sampling bias. In a tiny population (one island) a handful of exceptions (164 people over 90) massively affects the average lifespan. There's no randomization or blinding at all: observational studies like these are exquisitely sensitive to methodology errors, especially when examining small effects.