I read a lot of books. I probably average a book every two weeks during the year. I tend to read almost all non-fiction, as I have found that I don’t have the attention span for fiction. As a result, I tend to read a lot of technology, business and management books.
Unfortunately, I am starting to scale back the types of books that I read in these genres. What I’m finding is that since many of these books focus on showing that some idea or trait proves to be successful, they tend to all fall prey to the same basic flaw.
Typically used to describe the mutual fund sector, survivorship bias occurs when failed companies are excluded from studies due to the fact that they no longer exist. This is the case with most mutual fund portfolios. In terms other studies based on finding common properties of successful ideas or companies, survivorship bias simply means focusing those that are still around and hence, successful.
For example, let’s say that I’m writing a book about how Aeron chairs contribute to running a successful company. Being a good researcher, I go find a dozen or so companies to see what types of chairs they use. Right there, I’ve made a critical flaw: by finding companies that are still in business, I’m ignoring all of those companies who are already out of business. They could have all used Aeron chairs. Because of this, it would be easy to dismiss my findings.
The latest book that i read that falls prey to this is Made To Stick by Chip & Dan Heath. In Made To Stick, the authors put forth a simply acronym, SUCCES, that represents the 6 essential qualities that an idea needs to have in order for it to stick. They prove that successful ideas exhibit these qualities, but unfortunately, they don’t really make the case that ideas that don’t have these qualities consistently fail. Without the latter, the former is kind of incomplete.
I guess i’ll have to work hard to find books that present well researched and complete ideas.
As for how survivorship bias relates to the financial industry, that’s the topic of another post.