If an index isn't representative of what active managers in the class hold, then active managers will outperform or underperform the index, depending on whether the bias helps or hurts them.
The article indicates that this is the case with preferred stocks; active preferred-stock managers prefer stocks which are not callable. When interest rates fall, callable preferred stocks get called, and non-callables have an advantage; when they rise, nothing gets called, and callables have an advantage.
You can see a similar phenomenon in the comparison of Vanguard's bond indexes to their M* peer groups; the peer group is "corporate" funds, while the index has more government bonds. Thus Total Bond Market Index Admiral shares were 9th percentile in 2007 and 2008, when Treasuries outperformed corporates, and 90th percentile in 2009, when corporates did much better. This says nothing about bond-picking abilities; an all-corporate index would have shown the reverse pattern. For example, Intermediate-Term Investment-Grade (index-like but not an index) was below average in 2008 but not any other year in the last ten.