How polling firms are like mutual funds
October 21, 2016 Leave a comment
Ever research a mutual fund to buy? Just maybe, you’ve seen “past performance is not an indicator of future results.” Yes, some mutual funds really are better than others, but there’s a hell of a lot of randomness going on in mutual fund performance. I read John Bogle’s (guru of index fund investing) book on the matter way back in graduate school and he explains nicely how when you have literally thousands of funds, just by sheer statistical chance, a certain number of funds are going to beat the market for a 5 or 10 year run. Now, maybe some of those funds have great stock pickers, but it is just as likely that they were basically on a lucky run. Buy a fund because it’s got a good record over 5 years, it is far more likely to regress to the mean than to keep over-performing.
Why a long paragraph on mutual funds? Because IBD/TIPP has just released a poll that has Trump up by 1 (here it is in the Pollster average). And, apparently, they were among the most accurate pollsters in 2012. Now, some pollsters really do use more careful and effective methods than others (i.e., cell phones and landlines with multiple callbacks, etc.), but even those using state of the art procedures are going to get some pretty different results. And, statistically-speaking, that’s just going to happen. More likely that IBD is just somehow better than Selzter and Monmouth (to name a couple of 538 A+ pollsters), or that they quasi-randomly had a final election sample that was right on the nose in 2012? I think you know the answer. Short version: never get too excited (or worried) over a single poll! Shorter version: stick with the aggregate. Oh, yeah, and index funds.