Department of
History & Political Science



Follow me
on Twitter for
updates to this
forecast

 

Forecasting the 2014 Senate Elections

DeSart Senate Forecast - 2014


ELECTION DAY UPDATE November 4, 2014 - The final update of my Senate forecast projects that the Republicans have a 76.3% probability of winning majority control of the United States Senate today.

There are no changes in the projected winners in any of the races with this update, however there has been a slight move towards the Republicans in the most competitive races in the past week, which accounts for the slight increase in the probability that the Republicans will win the majority. The model's final projected win probabilities are presented below.

Projected
Democratic Wins
Projected
Republican Wins
Δ
State D Win
probability
R D Win
probability
R State
RHODE
ISLAND
Reed > 99.9% Zaccaria - 100% Sessions ALABAMA  
HAWAII
(special)
Schatz > 99.9% Cavasso Hardy > 99.9% Enzi WYOMING  
MASSACHUSETTS Markey 99.6% Herr Silverstein > 99.9% Inhofe OKLAHOMA  
DELAWARE Coons 99.1% Wade Johnson > 99.9% Lankford OKLAHOMA
(special)
 
OREGON Merkley 98.8% Wehby Dickerson > 99.9% Scott SOUTH CAROLINA
(special)
 
NEW JERSEY Booker 97.4% Bell Mitchell > 99.9% Reich IDAHO  
ILLINOIS Durbin 97.1% Oberweis Bellows > 99.9% Collins MAINE  
MICHIGAN Peters 95.5% Land Domina > 99.9% Sasse NEBRASKA  
MINNESOTA Franken 92.5% McFadden Childers 99.9% Cochran MISSISSIPPI  
NEW MEXICO Udall 92.2% Weh Alameel 99.6% Cornyn TEXAS  
VIRGINIA Warner 85.7% Gillespie Ball 99.5% Alexander TENNESSEE  
NORTH CAROLINA Hagan 54.0% Tillis Hutto 98.5% Graham SOUTH CAROLINA  
NEW HAMPSHIRE Shaheen 52.7% Brown Weiland 97.0% Rounds SOUTH DAKOTA REP
GAIN
Tennant 96.9% Capito WEST VIRGINIA REP
GAIN
Curtis 91.1% Daines MONTANA REP
GAIN
Orman* 75.3% Roberts KANSAS  
Grimes 75.3% McConnell KENTUCKY  
Landrieu 71.9% Cassidy LOUISIANA REP
GAIN
Pryor 62.3% Cotton ARKANSAS REP
GAIN
Begich 62.1% Sullivan ALASKA REP
GAIN
Nunn 60.5% Purdue GEORGIA  
Braley 59.4% Ernst IOWA REP
GAIN
Udall 54.3% Gardner COLORADO REP
GAIN
NET REPUBLICAN GAIN +8
*Greg Orman is an independent candidate, and the leading challenger to the Republican incumbent, Pat Roberts,
after the Democratic candidate, Chad Taylor, dropped out of the race in September.


Running Monte Carlo simulations based upon these win probabilities generates the frequency distribution of possible instutional outcomes below. As you an see, the most likely outcome is the Republicans holding a 52-48 majority in the Senate when the 114th Congress convenes in January.


RACES TO WATCH


Assuming that the Republicans will pick up the seats in South Dakota, West Virginia, and Montana, which the model suggests they should do by comfortable margins, that means they only need to pick up an additional 3 seats in order to gain the majority. For the Democrats to have a chance at holding on to the majority, they will need to win at least 6 of the following 8 races:
  • Alaska
  • Arkansas
  • Colorado
  • Georgia
  • Iowa
  • Louisiana
  • New Hampshire
  • North Carolina
The challenge for the Democrats is that the model gives the edge to the Republicans in 6 of these races.




October 28, 2014 - My October Senate forecast model projects that the Republicans now have a 73.6% probability of winning majority control of the United States Senate next Tuesday.

The model now projects that Republican candidates will win Democratic-held seats in 8 states, adding Arkansas to the list of GOP pick-ups and making the prospect of a Republican majority even more likely. In addition, the model now classifies New Hampshire and North Carolina slightly more marginal, but still in the Democratic column. Alaska, Georgia, and Iowa did become somewhat more competitive according to the model, but they are each still projected to be won by the Republican candidate.

Projected
Democratic Wins
Projected
Republican Wins
Δ
State D Win
probability
R D Win
probability
R State
RHODE
ISLAND
Reed 100% Zaccaria - 100% Sessions ALABAMA  
HAWAII
(special)
Schatz > 99.9% Cavasso Hardy > 99.9% Enzi WYOMING  
MASSACHUSETTS Markey 99.8% Herr Silverstein > 99.9% Inhofe OKLAHOMA  
DELAWARE Coons 99.1% Wade Johnson > 99.9% Lankford OKLAHOMA
(special)
 
OREGON Merkley 98.4% Wehby Bellows > 99.9% Collins MAINE  
NEW JERSEY Booker 97.9% Bell Dickerson > 99.9% Scott SOUTH CAROLINA
(special)
 
ILLINOIS Durbin 97.1% Oberweis Mitchell > 99.9% Reich IDAHO  
MICHIGAN Peters 94.0% Land Domina > 99.9% Sasse NEBRASKA  
MINNESOTA Franken 92.6% McFadden Childers 99.9% Cochran MISSISSIPPI  
NEW MEXICO Udall 92.2% Weh Alameel 99.6% Cornyn TEXAS  
VIRGINIA Warner 85.5% Gillespie Ball 99.5% Alexander TENNESSEE  
NORTH CAROLINA Hagan 55.0% Tillis Hutto 98.6% Graham SOUTH CAROLINA  
NEW HAMPSHIRE Shaheen 53.2% Brown Weiland 97.3% Rounds SOUTH DAKOTA REP
GAIN
Tennant 96.9% Capito WEST VIRGINIA REP
GAIN
Curtis 93.3% Daines MONTANA REP
GAIN
Landrieu 75.3% Cassidy LOUISIANA REP
GAIN
Orman* 74.2% Roberts KANSAS  
Grimes 72.6% McConnell KENTUCKY  
Begich 61.9% Sullivan ALASKA REP
GAIN
Braley 58.9% Ernst IOWA REP
GAIN
Nunn 55.0% Purdue GEORGIA  
Udall 54.2% Gardner COLORADO REP
GAIN
Pryor 53.2% Cotton ARKANSAS REP
GAIN
NET REPUBLICAN GAIN +8
*Greg Orman is an independent candidate, and the leading challenger to the Republican incumbent, Pat Roberts,
after the Democratic candidate, Chad Taylor, dropped out of the race in September.



The new probability distribution for the projected institutional outcomes is presented below. The projected results suggest that Senate Republicans will likely hold a 52-48 seat majority in the 114th Congress.





October 20, 2014 - My October Senate forecast model projects that the Republicans now have a 67.1% probability of winning majority control of the United States Senate in two weeks.

The model projects that Republican candidates will win Democratic-held seats in 7 states, which is more than enough to give them a majority. The September forecast projected they would pick-up 6 seats (Alaska, Iowa, Louisiana, Montana, South Dakota, and West Virginia) to win the majority. This update of the forecast now projects that Cory Gardner will defeat the Democratic incumbent, Mark Udall, in Colorado. In addition, the model also suggests that another Democratic incumbent, Mark Pryor, is also quite vulnerable. The model gives him a slight edge over Republican challenger Tom Cotton, but with a win probability of only 50.9%.

The only question marks for Republican-held seats are in Georgia and Kentucky, where the model projects that the Republican candidates will win but with win probabilities of less than 70%. The seat currently held by retiring Republican Senator Saxby Chambliss of Georgia is the most vulnerable for the GOP, with David Purdue projected to win over Democrat Michelle Nunn with only a win probability of 55.3%. Pat Roberts of Kansas, also remains somewhat vulnerable to independent candidate Greg Orman, who has stated that if he wins he will caucus with whichever party holds the majority. So even if Roberts loses, it is likely that the seat will, in effect, remain in "Republican" hands.

Below is a race-by-race rundown of the projected win probabilities generated by the model:
PROJECTED DEMOCRATIC WINS PROJECTED REPUBLICAN WINS
StateDemocratic
Candidate
Win ProbabilityRepublican
Candidate
StateDemocratic
Candidate
Win ProbabilityRepublican
Candidate
HAWAIISchatzCavassoALABAMA - Sessions
RHODE ISLANDReedZacharriaWYOMINGHardyEnzi
DELAWARECoonsWadeOKLAHOMA
(special)
JohnsonLankford
MASSACHUSETTSMarkeyHerrOKLAHOMASilversteinInhofe
NEW MEXICOUdallWehMAINEAllenCollins
NEW JERSEYBookerBellNEBRASKADominaSasse
ILLINOISDurbinOberweisIDAHOMitchellRisch
MINNESOTAFrankenMcFaddenSOUTH CAROLINA
(special)
DickersonScott
OREGONMerkleyWehbyTEXASAlameelCornyn
MICHIGANPetersLandTENNESSEEBallAlexander
VIRGINIAWarnerGillespieSOUTH CAROLINAHuttoGraham
NORTH CAROLINAHaganTillisMISSISSIPPIChildersCochran
NEW HAMPSHIREShaheenBrownMONTANACurtisDaines
ARKANSASPryorCottonSOUTH DAKOTAWeilandRounds
ALASKABegichSullivan
WEST VIRGINIATennantCapito
LOUISIANALandrieuCassidy
KANSASOrmanRoberts
KENTUCKYGrimesMcConnell
IOWABraleyErnst
COLORADOUdallGardner
GEORGIANunnPurdue
NET GAIN: Republican +7


Based on these win probabilities, the model generates the probability distribution of projected institutional outcomes presented on the right. As you can see, the Republicans hold at least 51 seats in a majority of the scenarios. The prospect of a 50-50 split with Vice President Biden's tie-breaking vote being the deciding factor in determining procedural control is much less influential than it was in giving the Democrats an edge as it did in the September forecast.

One assumption I have made, which gives the Republicans a slight edge, concerns how Kansas' Senate seat should be allocated in the event Greg Orman defeats Pat Roberts. As an Independent, his choice of which party with whom he would caucus could have an impact on the institutional outcome. For purposes of this analysis, I have made the assumption, should Orman win, that he would caucus with whichever party wins the majority. More importantly, I have made the assumption that in the event that Orman wins and that his one vote would make the difference between a 50-50 split and a 51-49 Republican majority, he would choose to caucus with the Republicans. Without that assumption, the Republicans' probability of winning a majority shrinks to 61.7%

One other interesting thing to note about this is that while it seems likely now that the Republicans will gain a majority in the Senate, there is a slight possibility that Mitch McConnell would not be the Majority Leader. McConnell is favored by the model to win re-election, but not by an overwhelming margin. In fact, the model lists him as the most vulnerable Republican incumbent with a win probability of just over 65%. The Democrat's hopes of retaining the majority, according to the model, rest on them holding onto Mark Pryor's seat in a very close toss-up race in Arkansas, and possibly picking off the Republican held seats in Georgia and Kentucky. Either that, or late surges by Bill Braley in Iowa and/or Mark Udall in Colorado appear to be the Democrats' best hopes to holding on to their Senate majority.



October 6, 2014 -  My Senate forecast model, which I presented at the 2012 Annual Meeting of the Western Political Science Association, projects that the Democrats have a 52.8% probability of holding on to their majority control of the United States Senate this year.

While the point estimates generated by the model suggest that the Republicans may gain the six seats they need in order to retake control of the Senate, their fate is essentially tied to a handful of very close races that are classified, for all intents and purposes, as toss-up races. The Republicans are projected to make likely gains in Montana, South Dakota, and West Virginia, where Democratic incumbents have all chosen not to run for reelection. The model also suggests that there is a better than 65% probability Republican challengers will defeat Democratic incumbents Mary Landrieu in Louisiana and Mark Begich in Alaska. If these 5 races go as projected, and all other seats were held by their respective parties, that would result in a 50-50 split in the Senate. Given Vice President Biden's constitutional prerogative to cast tie-breaking votes in the Senate, this would still effectively result in procedural control being in the hands of the Democrats.

There are four races that the model essentially categorizes as true "toss-up" races (win probabilities of less than 60%): Arkansas, Colorado, Iowa, and Kansas. Democrats currently hold three of these four, with Pat Roberts of Kansas being the only Republican incumbent running in what would be considered a toss-up race. The model thus suggests that the balance of power in the Senate rests largely on these races, and the Democrats would need to win three of those four to maintain their majority. In Kansas, the model projects that Roberts has a 58.2% probability of holding his seat. Assuming that happens (which is by no means a sure thing), that means that the Democrats will have to sweep the three remaining toss-up races. Simply based on the point estimates alone, the model suggests that this will not happen, projecting that the Republican candidate Joni Ernst has a slightly better than a coin-flip chance of winning the Iowa seat being vacated by retiring Democratic incumbent Tom Harkin. This would be the gain that the Republicans would need to break the tie, put them over the top, and give them an outright majority.


                  Below is a table of the model's projected outcomes of each of the 36 races, along with their associated win probabilities:
STATE Democratic
candidate
Win Probability
(color denotes projected winning party)
Republican
candidate
Predicted
Outcome
HAWAII Schatz Cavasso Democratic
hold
RHODE ISLAND Reed Zaccaria Democratic
hold
MASSACHUSETTS Markey Herr Democratic
hold
DELAWARE Coons Wade Democratic
hold
ILLINOIS Durbin Oberweis Democratic
hold
NEW MEXICO Udall Weh Democratic
hold
NEW JERSEY Booker Bell Democratic
hold
OREGON Merkley Wehby Democratic
hold
VIRGINIA Warner Gillespie Democratic
hold
MINNESOTA Franken McFadden Democratic
hold
MICHIGAN Peters Land Democratic
hold
NORTH CAROLINA Hagan Tillis Democratic
hold
NEW HAMPSHIRE Shaheen Brown Democratic
hold
COLORADO Udall Gardner Democratic
hold
ARKANSAS Pryor Cotton Democratic
hold
IOWA Braley Ernst Republican
gain
KANSAS Orman* Roberts Republican
hold
LOUISIANA Landrieu Cassidy Republican
gain
ALASKA Begich Sullivan Republican
gain
KENTUCKY Grimes McConnell Republican
hold
GEORGIA Nunn Purdue Republican
hold
SOUTH DAKOTA Weiland Rounds Republican
gain
MONTANA Curtis Daines Republican
gain
WEST VIRGINIA Tennant Capito Republican
gain
MISSISSIPPI Childers Cochran Republican
hold
SOUTH CAROLINA Hutto Graham Republican
hold
TENNESSEE Ball Alexander Republican
hold
TEXAS Alameel Cornyn Republican
hold
SOUTH CAROLINA
(special)
Dickerson Scott Republican
hold
NEBRASKA Domina Sasse Republican
hold
MAINE Allen Collins Republican
hold
OKLAHOMA Silverstein Inhofe Republican
hold
OKLAHOMA
(special)
Johnson Lankford Republican
hold
IDAHO Mitchell Risch Republican
hold
WYOMING Hardy Enzi Republican
hold
ALABAMA - Sessions Republican
hold
Projected net change +6
Republican
*Democratic candidate Chad Taylor dropped out of the race for the Kansas Senate seat held by Republican incumbent Pat Roberts, leaving Independent candidate Greg Orman as the leading challenger.
Should Orman win, his decision of whether to caucus with the Democrats or the Republicans is very much an open question at this point. He has, however, noted that he would likely caucus with
whichever party holds the majority after the election.



Below is a probability distribution of the potential institutional outcomes based on the win probabilities of each of the races as presented above. It shows that the ultimate question of which party will hold majority control is a virtual coin-flip, with the slight edge going to the Democrats, largely because of the fact that Vice-President Biden's tie-breaking vote could tilt the balance in the Democrat's favor in the event of a 50-50 split, just as Vice President Cheney's vote gave the Republicans procedural control in spite of an evenly divided Senate following the 2000 election.



Of course, much could depend on what Independent candidate Greg Orman does in the event he defeats Pat Roberts in Kansas. For purposes of this analysis, I have made the assumption that he would caucus with the Democrats. However, there is one particular set of circumstances where his decision could have a significant impact, and fundamentally alter these probabilities. Assume, for example, that Orman does defeat Roberts and that the combination of other outcomes gives the Republicans 50 seats. In this situation, Orman would have a decision to make: He could caucus with the Democrats, create a tie, and that would leave the tie-breaking vote in the hands of Vice President Biden to give the Democrats procedural control. That is the assumption that I have made here.

However, it is also possible in that scenario that he could prevent a tie by caucusing with the Republicans. This would effectively give the Republicans a 51-49 majority. My analysis of the win probabilities above suggests that this scenario has around a 10.2% probability of happening. If Orman were to do this, it would fundamentally alter the projected institutional outcomes presented above. Instead, it would create the distribution of outcomes presented below.



From this, you can see that this scenario would significantly improve the Republicans' chances of gaining a majority, raising their prospects from a probability of 47.2% to 57.4%.

In the final analysis, what these data reveal is that the ultimate status of the US Senate following the election is still very much in doubt, and will ultimately rest on the outcomes in a small handful of states.