So I’ve spent the last few days mulling the potential meanings of various post-convention bounces, which you can catch up on here, here, and here. But today, I got an invitation to a Labor Day BBQ and the terrible news dawned on me: We might not get a very good idea of Romney’s bounce.
In 2008, just six national surveys were conducted between the DNC and the RNC, including the Rasmussen and Gallup tracking polls, both because the window was extremely short, and because pollsters are reluctant to poll over vacation weekends, like Labor Day. There have been fewer polls so far in 2012 than 2008, which might reduce the number of non-tracking polls beneath four.
Complicating matters further are Rasmussen and Gallup’s pre-convention polls, which showed Obama with a lead heading into the RNC. In both instances, Obama’s slight edge represented his best performance in weeks, particularly for Rasmussen. Now perhaps Obama really did make gains over the last week, but other polls haven’t shown similar movement and, if anything, the other polls have shown a tightening race among registered voters since Ryan was selected as Romney’s running mate.
All of this raises the question of the baseline: if Romney takes a 3-point lead in Rasmussen and a 2-point lead in Gallup, should that be considered a bounce? That would represent a 3 or more point improvement compared to the pre-convention polling, but both tallies would be highly consistent with Gallup and Rasmussen’s medium-term numbers. Put differently: If Romney’s up 5 points in Rasmussen on Tuesday, should that be interpreted as a 6 or a 3 point bounce? I’m not sure, but I lean toward the former.
So I’m getting geared up for a disappointing post-RNC bouncefest. There’s a real chance we won’t have a satisfactory set of polling data to get a clear picture of Romney’s bounce, especially since the pre-convention tracking polls deviated from their medium-term trends. Folks geared up to judge Romney's bounce should be cautious about over-interpreting a limited set of polls conducted over a holiday weekend.