New Democratic Louisiana Senate Poll Finds A Tied Race, But With Strong Likelihood of Republican Gains

A recent television ad paid for by by 'Friends of Mary Landrieu' shows citizens of a rural Louisiana town watching Landrieu TV appearances in which she takes it to the Obama Administration on a number of issues. Separating herself from an unpopular president is a must if she hopes to hang on to her seat.

Of the premier 2014 midterm contests, the Louisiana Senate race is *the* most likely to flip into Republican hands, at least according to current Huffington Post Pollster averages. That's why the  new Public Policy Polling survey allowed the Landrieu campaign to breathe at least a slight sigh of relief, especially considering the recent spate of run-off polling. An incumbent who is tied with their opponent at 47% wouldn't typically be received as welcome news for most political campaigns, unless six of the eight polls taken this year found that opponent ahead.

But as far as Democrats are concerned, the good news stops there. Because a closer inspection of the PPP memo and crosstabs suggests little room for improvement for the incumbent:

The likely Landrieu/Cassidy match up for the December runoff is tied at 47. Among those who support Maness or Hollis or are undecided for the November election, 68% move to Cassidy for December compared to only 11% who move toward Landrieu. Even though only 6% of voters are undecided in that match up, they don't set up great for Landrieu- 61% voted for Romney to 20% who voted for Obama, and she has a 14/65 approval rating.

It's July, and if you buy the PPP numbers, a mere 6% of likely voters are undecided. So no matter where they end up, the Louisiana Senate race is bound to be at least remotely close.

As Nate Silver has argued before, at least at the presidential level, there comes a point in the election cycle where job approval ratings become a leading indicator of final result. So let's divvy up that tiny 6% slice of the electorate to the Louisiana Senate candidates, with 14% going to Mary Landrieu (her approval rating among undecided voters), and 65% going to Republican Bill Cassidy (her disapproval rating among undecideds).

Suddenly, the GOP challenger jumps out to a 51-48% lead, which goes to show that despite undecided voters being few and far between, Landrieu is unpopular enough with them to make a difference. The same thing happens when you reallocate undecideds according to how they voted in the 2012 election (61% for Romney, 20% for Obama). What was a 47-47 tie again becomes a 51-48% Cassidy lead.


Beyond the make-up of undecided voters, the PPP survey drops subtle hints of trouble ahead for Landrieu throught their survey, from racial I.D., 2012 presidential vote, and partisan I.D. findings. For example, the Democratic polling organization found a more racially diverse midterm electorate than the one that showed up on election day 2010 (at least according to exit polling), when incumbent Senator David Vitter demolished his opponent, Charlie Melancon, 57-38%. In fact, PPP's racial findings look much more like the 2008 presidential electorate than the 2010 midterm electorate:

Percentages courtesy of CNN exit polling. No exit poll was conducted in 2012.

This matters because Bill Cassidy's lead with white voters is a substantial 64-28%, as is Landrieu's lead among black and 'other' voters (89-7% and 58-41%, respectively).

For what it's worth, the final survey PPP conducted of the 2010 Louisiana Senate race found white voters making up 71% of the electorate, blacks at 25%, the rest at 4%, essentially the same thing found by 2010 exit pollsters.

So what happens to the 47-47% tie between Landrieu and Cassidy, in the event PPP had found Louisianans racially identifying as they did in their final 2010 survey, as well as 2010 exits?

The Republican challenger gains two points from the original poll finding to hit 49%, while the Democratic incumbent falls three points to 44%. What was a tie becomes a more discernible edge for the Republican, while Landrieu slips further and further from the 50% mark.

Something similar happens when considering another one of PPP's crosstabs - the 2012 presidential vote. Among the 664 likely Louisiana voters identified by PPP, 53% say they voted for Mitt Romney in 2012, while 40% say they voted for Obama. The actual 2012 vote distribution gave the win to Romney, 58-41%. What would happen to the PPP result had the response to the '2012 vote' question more closely resembled the actual result?

Keep in mind that of the twenty-five state exit polls conducted in 2010, the "who did you vote for in 2008" finding shifted in the Republicans favor in twenty of them (by an average of 5 points). It shifted in the Democrats favor in just 4 of the 20 states.

 As you can see, Cassidy again emerges with an edge of 49-45%.

Another clue that Landrieu could be maxing out her level of support in the PPP survey comes from their partisan identification finding. Democrats make up 44% of poll respondents, Republicans make up 36%, and Independents account for 20%, giving Democrats a partisan ID advantage of D+8. That's significantly less Republican than exit poll findings from 2010, and even 2008 presidential exits.

Though anything is possible, recent history and midterm dynamics suggest it's unlikely Landrieu will enjoy a Democratic partisan identification surge above 2008 levels, which PPP suggests. In fact, their final 2010 Louisiana Senate survey found Republicans with a net party I.D. advantage of one point. But just for fun, how drastic of a shift would we see in the final result, assuming the partisan make-up of the 2014 Louisiana electorate is identical to 2010 exit polling, all other PPP findings remaining the same?

Landrieu falls to her lowest level of support seen in any of the above experiments, while Cassidy reaches his highest.

In conclusion, I'd like to note one thing: this exercise is not to suggest that PPP's findings are incorrect. It's more an attempt to illustrate how much the racial and partisan composition of the electorate matters to the final outcome, especially when the candidates are both winning 2/3 or more of each of the various groups, as Landrieu and Cassidy are. But more importantly, this exercise was intended to show that while Democrats may be tied in Louisiana, the fundamentals don't look very good, at least not according to this Democratic pollster.