Can anyone suggest an interesting follow-on study or an improvement to the one discussed in this article?

This study wasn't as "clean" as it could have been - what variables are possibly confounded?

http://news.*****.com/s/nm/20051020/hl_nm/retirement_dc

The study was based on 3500 employees in Shell Oil Company in Houston, Texas. I think education background, and profession may be potential confounding variables.

At the end of the article they mentioned several studies, I would be interested to find out how they were done. The conclusion makes sense though.

Yes, definitely a wider swath of the population would modify the results somewhat.

I was also thinking about studying retirement age, controlling for "health status." In other words, would retirement age alone tell you anything about longevity?

Yeah that makes sense. Since they are talking about health status why not controlling for it?

I am also thinking about a retrospective study by looking at the age people died and retired, assuming the information is in public records.