Monday, December 5, 2011

Precision Competition

In a recent post I raised the point about the spurious degree of precision that is often encountered with reported regression results. So, here's a challenge for you - how many decimal places (or maybe significant digits) are appropriate when reporting OLS regression results?