Tuesday, September 9, 2014

Why Sen. Pat Roberts Isn't Out Of This Yet - A Dive Into The New SurveyUSA Kansas Senate Crosstabs

Pictured above, former Republican and Democrat, now Independent candidate for Senate in Kansas, Greg Orman.

Survey USA was first out of the gate yesterday afternoon, releasing a new poll of the Kansas Senate battle taken entirely *after* the shocking announcement last week that Democrat Chad Taylor would be exiting the race. Cutting to the chase - things look pretty bad for Republicans, at least on the surface.

Long-time incumbent Pat Roberts is polling at just 36% with the general electorate - that, after having won 60% in 2008, 83% in 2002, and 62% in 1996. Republican turned Democrat turned Independent Greg Orman comes in at 37%. A stunning 10% say they will vote for Chad Taylor regardless of the fact he has dropped out of the race. 17% are either undecided or will vote for Libertarian Randall Batson.

Given Kansas Democrats singular desire to kick Roberts out of D.C., it's easy to look at the 37-36-10% split and just say "hey, give that 10% for Taylor to Orman, and you have a 47-36% Independent candidate lead over the incumbent Republican." But why stop the assumptions there? Especially considering the fact that the devil is in the details.

First of all, the odds that 10% of Kansas likely voters will actually wind up voting for a non-candidate like Taylor are unlikely. So to see how the Taylor vote could break-up down the road, consider the partisan make-up of the 10% of Kansas likely voters that say they will vote for Chad Taylor. 21% of Democrats support Taylor. 12% of Independents do. And just 3% of Republicans support Taylor. Now suppose that two-thirds of the Taylor-supporting Democrats decide to abandon the non-candidate in favor of Greg Orman between now and November 4, 2014. The remaining one-third of Taylor supporting Democrats stay with Taylor in this scenario because...well...some Democrats are bound to vote for the guy with a (D) beside his name. In the meantime, reallocate two-thirds of the Taylor-supporting Republicans to the Roberts column, and leave the remaining third with Taylor. Lastly, reallocate two-thirds of the Taylor-supporting Independents evenly between Roberts and Orman, with the remaining one-third staying loyal to Taylor. How would the Survey USA result have looked under such a feasible scenario, all other findings remaining the same?

So you see, it's not quite as simple as just slapping all 10% of Taylor voters up on Orman's board. Orman would lead in such a hypothetical where Taylor voters side overwhelmingly with Orman, but by 4 points overall. And remember, the scenario above also assumes Roberts' 59% among Republican voters increases to 61% as a result of taking two-thirds (2%) of Republicans that claim they will support Democrat Chad Taylor. The scenario further assumes that 2/3 of Independent Taylor supporters will split evenly for Roberts and Orman, with Taylor maintaining the other third. In the end of this scenario, Taylor's support among the general electorate is around 3%, which sounds more likely as awareness grows about his non-candidacy in the 8 weeks remaining until election day.

If you think that Orman being limited to 52% of the Democratic vote, with Taylor still attracting 21% of the Democrats, sounds unlikely to stand up over the next eight weeks, then certainly Pat Roberts sub-par standing with Republicans seems equally implausible, as well as Greg Orman's strong showing with Republicans. To be specific, Senator Pat Roberts is only attracting 59% of members of his own party, and is bleeding 26% to Greg Orman. In a two-man race, the Republican is no doubt going to hammer the Independent for being a Democrat in sheep's clothing. Two months of that should take a toll. Let's assume that 10% of Orman's current Republican support jumps ship and joins team Roberts by November, boosting Robert's Republican support to 71%. Meanwhile, Orman's Republican support slumps from 26% to 16%. Here's how the Survey USA poll would have looked:

Roberts reclaims the lead if you believe Republican Taylor supporters and Republican Orman supporters will return home to Roberts in large part - even after having reallocated the overwhelming majority of Democratic Taylor supporters to Greg Orman's column.

Now, what should we do with those Batson numbers? The little-known Libertarian candidate is polling at 6% (impressive for third party candidates in general, though nearly common place for several elections this year and last). But if 2014 follows the regular rules, Batson will likely win less than the 6% he currently polls at. So suppose we reallocate half of Democratic Batson supporters to Orman's column, while leaving the other half with Batson. Orman's percentage among Democratic likely voters would suddenly be at ~70%. Let's do the same with Batson's Republican supporters - half of them will be reallocated to Pat Roberts column, the other half will stay with Batson. As for Batson's Independent supporters, half of them are reallocated evenly to Roberts and Orman, the other half stay in Batson's corner. In such a scenario, the Survey USA poll would have looked like this:

The last group we've yet to hypothetically deal with is undecided voters. 9% of both Democrats and Republicans were undecided. 18% of Independents are undecided. For simplicity sake, and to give the benefit of the doubt to the Democrats, let's say all 9% of undecided Democrats support Orman in the end. This would elevate his level of Democratic support to 79% (remember, he started at 52%). Regarding the undecided Republicans, lets allocate them evenly between Roberts and Orman (again, purposely being overly beneficial to Orman). This pushes Robert's percentage among Republicans up to 77% (remember, we started at 59%). Finally, lets allocate two-thirds of the 18% of undecided Independents to Orman, and one-third of the undecided Independents to Roberts.  Even then, under this fairly rosy scenario for Greg Orman, Pat Roberts still comes out on top. But only barely:

The news isn't good, by any stretch, for Kansas Republicans. Outside of the Senate race, Brownback is polling pretty terribly for a long-time, once popular Kansas statewide politician. But it isn't over yet for Senator Pat Roberts. 36% is a daunting place to be for any incumbent under any electoral scenario. But even considering some of the rosier situations for Independent Greg Orman, he manages a two point deficit at best.

Who knows how the Kansas Senate race will shake out? And the examples above are just a small glance at a few likely scenarios. If anyone is interested, feel free to show me some likely scenarios in which Orman overtakes Roberts (dealing away with high Taylor vote, high Batson vote, and high number of undecides).


  1. "consider the partisan make-up of the 10% of Kansas likely voters that say they will vote for Chad Taylor. 21% of them are Democrats. 12% are Independents, and 3% are Republicans."


    "In the meantime, reallocate two-thirds of the Taylor-supporting Republicans to the Roberts column, and leave the remaining third with Taylor."

    So, among Republicans who have already rejected the Republican in favor of a Democrat, none will end up with Orman?? Does't it make more sense to assign more of these voters to Orman than to Roberts? Or, at least to split them?

  2. That first section has been edited to make sense (it did not previously, as I assume you were pointing out with 'huh??' Thanks for the catch.

    Regarding your second point - as I noted in my last hypothetical, I'm making lots of assumptions that are favorable to Greg Orman. But you raise a good point. In an effort to be overly favorable to Orman, lets give him ALL of the Taylor-supporting Republicans, and none to Roberts. Even after that, it's a 46-46% tie in the final scenario (46.0 Roberts, 45.7% Orman, to be exact).


Note: Only a member of this blog may post a comment.