An old rule of poll interpretation was that incumbents who poll below 50 percent were going to lose. The idea was that 100 percent knew them, and 51 percent plus weren’t voting for them. That rule has proved to be less predictive in recent cycles: Incumbents polling in the high 40s win fairly often, while those polling in the low 40s very occasionally (two instances in the six Senate cycles between 2002 and 2012). What does the rule look like this year in Senate races?
The answer is that Democratic incumbents in seriously contested races tended to get about what they were polling. The few (two) Republican incumbents in such races did better. Here’s the data, first for Democrats and then Republicans: the incumbent’s last percentage in the RealClearPolitics average of recent polls (in the case of Louisiana for the Nov. 4 primary, not the Dec. 6 runoff), the incumbent’s percentage of the total vote as currently tabulated and the amount the incumbent exceeded or fell short of that percentage.
• Alaska — Mark Begich		43.8	45.3	+1.5
• Arkansas — Mark Pryor 	 41.2	39.5	-1.7
• Colorado — Mark Udall 	 44.0	45.2	+1.2
• Louisiana — Mary Landrieu 	 40.2	42.1	+1.9
• New Hampshire — Jeanne Shaheen 	 48.8	51.9	+3.1
• North Carolina — Kay Hagan 	 44.1	47.3	+3.2
• Virginia— Mark Warner 	 48.5	49.1	+ .6
• Kansas — Pat Roberts 	 42.6	53.2	+10.6
• Kentucky — Mitch McConnell 	 49.0	56.2	+ 7.2
Clearly the rule seems to hold on the Democratic side. Democratic incumbents Shaheen and Warner polling just under 50 were able to win, narrowly. Democratic incumbent Hagan, though leading in polling through most of the cycle, was under 45 and did not manage to win. When you’re an incumbent and the tide is running against your party, you better be polling pretty close to 50 if you hope to win. And maybe there’s a little premium for female candidates: Shaheen and Hagan ran farther ahead of their polling than male Democratic incumbents in other states.
On the Republican side, both Roberts and McConnell are Senate veterans in their 70s, and Roberts had the additional problem (heavily aired in a closely contested primary) of not having a real residence in the state for a long time. Yet both managed to run ahead of their polling. Hypothesis: a lot of voters engaged in anti-incumbent grumbling and declined to tell pollsters they’d vote for the incumbent. But in states that voted 60 and 61 percent for Mitt Romney, when it came time to vote, they pulled the lever for the Republican.
