While I am not always a fan of experimental economics (or finance in this case), this one is interesting and it gets to whether people diversify rationally or not.
SSRN-Irrational Diversification: An Experimental Examination of the Diversification Heuristic by Thierry Post, Guido Baltussen
Short version: people do an OK job, but not optimal and how the diversification topic is framed, influences how people diversify. In other words, they tend to ignore how assets work with rest of the portfolio and focus more on how they work in isolation.
From their conclusion:
* "we conviningly reject the simple 1/n rule. Only a few subjects select an even allocation across all lotteries. A large majority of sugjects focus on a subset of the lotteries. Further, the subsets chosen are consistent with the idea that the subjects exclude the individual choice alternatives that are unfavourable when held in isolation"
"emphasizing the diversification benefits...leads to better choices."