By Robert Preuhs and Arina Rakytianska

It’s state legislative election prediction time once again!  Two years ago, in our inaugural post, we presented three sets of predictions based on models incorporating basic fundamentals, a Blue Wave and a modified Blue Wave that accounted for incumbency.  While none of the models correctly predicted all the State Legislative races in 2018, the Blue Wave model did predict a 40/25 split favoring the Democrats in the House and a 19/16 split in the same direction in the Senate (which includes seats that were not up for re-election).  Not bad given that there is currently a 41/24 and 19/16 split in the two chambers, respectively.

This year, we are going to pit two models against each other.  Similar to our predictions two years ago, the goal is parsimony.  That is, we aim to correctly predict the most outcomes with as few factors as we can.  Thus, we’ll limit the factors to two—the Democratic percentage point advantage in partisan registration among active voters as of October 1, 2020 (what we call the Democratic Advantage, or DA); and a Momentum indicator which is just the change in the DA over the last year (from Oct. 1, 2019 to Oct. 1, 2020).  Just like last time, the models are pretty easy to replicate and really basic in terms of the indicators we use.  The point, of course, is that while campaigns matter to some extent, State Legislative races and individual candidate success are driven primarily by the political fundamentals—the partisan composition of the district and an account of a general direction of that composition.

The two models also follow really simple rules which do not rely on polls, campaign events, messaging, news, advertising dollars, or anything else that are campaign or candidate specific.  We employ just plain old fundamentals and the data can be retrieved by anyone for replication from the Colorado Secretary of State’s Office.

Our first model is the simple Democratic Advantage (DA) Model.  The rule is the predicted winner of the election will be the candidate from the party with the most registered voters within the district. We determine this by subtracting the percent of Active Voters that are registered as Democrats from the percent of Active Voters that are registered as Republicans.  Positive values mean the Democrats have more registered voters and a negative value indicates that the Republicans have more registered voters.  If a district has a positive DA score, the Democratic candidate is predicted to win.  Conversely, in districts with a negative DA score, Republican candidates are predicted to win.

Our second model is the Democratic Advantage + Momentum (DA+M) Model. This model is somewhat of a tie-breaker model for what might be considered toss-up or competitive districts.  Again, the prediction rule is simple.  First, apply the rule of the DA Model, except where the absolute value of the DA is less than 5 points.  If the DA is less than 5 points, the Momentum indicator is the determining factor.  If Momentum is in the same direction as the Democratic Advantage, then the DA Model’s prediction remains intact.  If, however, the Momentum score is in the opposite direction of the Democratic Advantage, and of an absolute value greater than one, the direction of the Momentum score predicts the outcome.

Below are two graphs that present the Democratic Advantage and Momentum scores for all 65 State House districts and the 18 districts where an election is being held for the State Senate.  Under the Democratic Advantage model rules, blue lines predict a Democratic win while red lines predict a Republican win.  Adding momentum into the mix, the DA+M Model relies on the black lines indicating direction of momentum (positive values favor the Democrats) when the Democratic Advantage is less than 5 points in absolute value and the Momentum score is greater than one in absolute value.  Pretty simple.  See the end of the post for a complete list of Districts, data and predictions (R or D).

The models also allow us to make a prediction about the composition of the state legislature following the 2020 elections.

According to the Democratic Advantage Model, the predicted composition of the State House will be 36 Democrats and 29 Republicans.  In the Senate, 8 of the 18 seats up for election will be won by Republicans and the remaining 10 by Democrats.  That would mean a flip of one Senate seat (District 25 would go Democrat) and thus the Democrats would gain one seat and have a 20-15 majority.

The potential for the Democrats to loose five seats in the House likely seems improbable to most observers of Colorado politics, but we’ll let the model stand on its own for now.  Election Day will be the test.

That said, we do have a second model to test.  The DA+M Model accounts for competitive districts and the potential for partisan composition changes favoring one party or the other to put a candidate over the top.

For the DA+M Model in the Senate, two districts fall under the less than 5 point margin rule.  District 27, an open seat in a very competitive district with an extremely small (-.57) Democratic Advantage favoring Republicans, has experienced gains in Democratic registration over the last year (a 2.37 Momentum score).  Thus, the prediction flips as the DA+M model predicts a Democratic win.  District 35, also an open seat but with a slightly larger advantage for the Republicans (about 4.7 points) has momentum favoring the Republicans. Here, the DA+M model predicts a Republican win.  Ultimately, only one Senate prediction changed between the DA and DA+M Models.  The DA+M Model thus predicts a 21 to 14 Democratic majority in the Senate.

For the House, the DA+M model addresses seven races where the Democratic Advantage was below 5 points in absolute value.  With the momentum value changing the direction of the prediction in six relative to the DA model.  Here, the Democrats pick up six of the seven competitive races, while the Republican win predicted by the simple DA Model remains intact.  The composition of the House is thus predicted to be a 41 to 24 Democratic majority with the DA+M model.

For quick reference, here are the seven races affected by the DA+M model (* indicates a change in prediction relative to the DA Model):

59—Democrat Win* (-.85 DA; 1.28 Momentum)

50—Democratic Win* (3.31 DA; -.03 Momentum)

47—Republican Win (-1.09 DA; -1.22 Momentum)

38—Democrat Win* (-3.45 DA; 2.50 Momentum)

37—Democrat Win* (-.07 DA; 2.67 Momentum)

27—Democrat Win* (-.25 DA; 1.41 Momentum)

25—Democrat Win* (-3.26 DA; 1.95 Momentum)

The election will determine the best model.  But on their face, the predictions indicate a continuing trifecta for the Democrats in state government—they will hold both legislative chambers along with a Democratic governor.  And, the Democratic majorities will be almost exactly the same as they were leading up to the election.

We’ll see how well these models fair after the election.  Assuming a grading scheme that resembles a college course’s grading scheme, anything above 94% correct (or 78 of the 83 races) should be considered an “A” particularly since only two variables were used to make these predictions.  Or, if we want to focus solely on competitive districts, eight out of those nine competitive districts predicted correctly should earn an similar grade.  It is a pretty high bar, but we are hopeful that one or both of the models capture the essential factors that determine state legislative election outcomes.

There are a few more data points that are of interest, but not entirely relevant to making predictions.  First, there has been some suggestion in recent media accounts that the GOP has been leading the registration race in key states.  But in Colorado, at least over a year’s timespan, Democrats have clearly led the registration race in Colorado’s legislative districts.  83% of Senate seats up for election and 82% of House districts experienced more growth in Democratic registered voters relative to Republican registered voters over the last year.  The average Momentum score across House Districts is 1.18 and it is 1.24 in the 18 Senate races (1.08 across all Senate Districts).  In short, the national narrative of a GOP momentum advantage just does not hold in Colorado.

The second point that emerges is that there are very few competitive districts in Colorado’s legislative races.  Of the 83 district elections (both House and Senate), only 9 had partisan advantages that were less than 5 percentage points.  That is not to say that there won’t be competitive elections in Colorado (the Blue Wave in 2018 equated to a double-digit shift, on average, for Democratic candidates).  But it does suggest room for a non-partisan redistricting commission to create even more competitive districts during the 2021 round of redistricting.  If they do so, that may make the job of predicting legislative outcomes a bit harder in the coming decade.

So, there you have it—a new round of predictions based on very basic fundamentals.  We doubt the predictions will nail each and every district’s outcome, but part of the point here is to use non-campaign specific data to underscore the importance of the political context in structuring election outcomes.  If we are successful based on the 94% benchmark we established for ourselves, we hope that these models will underscore the importance of registration data in guiding an independent redistricting commission as they consider the newly established criteria of competitiveness for drawing district lines.

Here are the data and predictions for each of the districts:

House Districts

Senate Districts

Direct Link:  https://tinyurl.com/y2tqeru8

New Report from MSU Denver’s Golda Meir Center Documents the Rise and Impact of Gender, Racial and Ethnic Diversity in Colorado’s Legislature

Denver, CO— July 2, 2020 — Today, the Golda Meir Center for Political Leadership releases the report, Diversity in Colorado’s Legislature: Lessons from the 2018 Elections and 2019 Session, which documents the historic numbers of women and people of color running for, and holding office, in the State’s legislative body resulting from Colorado’s Blue Wave election in 2018 and the important impact diverse voices bring to the policy-making process.

“With four years of protests demanding justice, from the Women’s March to the on-going protests against racialized police brutality, documenting why diverse legislatures really matter, and how Colorado’s legislature became diverse, is as important today as in any time in our history,” said Robert R. Preuhs, Ph.D., Professor and Chair of Political Science at MSU Denver and editor of the report that was authored by participants in the Applied Political Research Lab.  “Moreover,” Preuhs added, “the 2020 state legislative primary results point to an equally diverse legislature in 2021, and this report speaks directly to how the 2020 general elections will play out and what the 2021 round of redistricting means for diversity and advocacy for the interests of people of color.”

Key Findings of the Report

• Democratic women ran and won at much higher rates than Republican women in 2018. 42 females ran in primaries for Democratic House seats compared to 32 males. Among Republicans, only 22 females ran compared to 44 males.  In Senate races, an even number of men and women ran for Democratic nominations, while men outnumbered women 5 to 1 among Republicans.  By the general election, these differences were even more pronounced.  Of the 39 females winning seats in 2018, 31 were Democrats (79%).
• Racial gerrymandering can affect the prospects of racially/ethnically diverse legislature. In particular, only in districts with more than 40% Latinx populations were Latinx legisalators the most likely to hold seats relative to Whites and Blacks.
• Democratic candidates of color were the most successful in their bids to win seats in 2018.  Two-thirds or more of all Black or Latinx candidates that ran as Democrats in the primaries eventually won their election.  80% of Latino candidates did so, and 100% of the 9 Latina Democrats running in the primaries eventually won the legislative seat in the general election.
• Females dominated campaign spending. Female Candidates Spent 3.5% More than Male Candidates, Even Though Females Accounted for Only 41% of Candidates. Latina candidates for the House spent more than any other group, on average.  With $85772 as an average.  The highest average was for White female candidates for Senate, spending $178,852 on average. Two Senate districts where Democratic women won elections (Senate 16 and 24) accounted for 31% of all spending in the 17 Senate races.
• Females publicize more events and engage in more Facebook outreach. Averaging 25.7 events compared to 17.2 events for males. 
• Issues, whether emphasized on websites or in sponsored legislation, reflect unique perspectives and underscore the need for legislative diversity. For instance, Latinx legislators mentioned Immigration (30% of them) more than others (12% of whites). Women sponsored all immigration bills, with two-thirds sponsored by Latinas. 40% of all Labor and Employment bills were sponsored by Black legislators or Latina legislators.  Black female legislators comprised 15% of Crime bills while only comprising 5% of the House seats.
• Democratic control led to more women and legislators of color holding leadership positions. Women, Black and Latinx legislators are over-represented in committee leadership positions, and at or above parity in chamber leadership.

###

The Golda Meir Center is a nonpartisan educational project whose purpose is to expand public understanding of the important role of leadership at all levels of political and civic life, from community affairs to transnational relations.

Robert R. Preuhs is Professor and Chair of Political Science at Metropolitan State University of Denver.  He’s research and expertise lie at the intersection of democracy and racial/ethnic politics.  He is co-author of Black-Latino Relations in U.S. National Politics, as well as numerous scholarly articles and book chapters.

Link to the full Report: Diversity in Colorado’s Legislature

By Robert Preuhs

In a previous post, I discussed the clear negative relationship between precinct-level Latinx voting age population (VAP) and turnout in Denver’s 2019 General Election and suggested turnout may be one factor in another mayoral election resulting in a non-Latinx mayor.  This installment turns its attention to the relationship between precinct-level Latinx VAP and candidate support.  As one should expect, Dr. Lisa Calderón tended to do better in precincts with higher Latinx VAPs as the figure below highlights.  But it was incumbent Michael Hancock who dominated precincts with large Latinx VAPs which at least provided part of his overall lead in the general election.  

Calderón’s performance in high Latinx VAP precincts was not stellar.  In fact, the bivariate correlation coefficient between the proportion Latinx VAP and the proportion of a precinct’s vote for Calderón was just .31—the weakest of all relationships between Latinx candidate’s support and Latinx VAP where data are available since 1983 (Pena’s 1983 general and runoff coefficients were .65 and .52, respectively, while Mares’ 2003 runoff coefficient was .87 and Mejia’s 2011 general election coefficient was .56).  The correlation between support for Hancock and Latino VAP was .45, a bit stronger of a relationship than Calderón’s.  Support for Penfield Tate and Jamie Giellis, in contrast, diminished as precinct Latinx VAP increased (coefficients of -.62 and -.56, respectively). (Note that all of these correlations are based on the 346 precincts for which Latinx VAP data from 2016 are available).

The table below highlights the relatively weak relationship between Latinx VAP and support for Calderón as well as the generally strong showing for Hancock across all precincts.  Note that even in precincts with more than 60% Latinx VAP, on average, Hancock had higher levels of support than Calderón.

Another way to look at this is the actual one-on-one outcomes (Hancock support vs. Calderón support).  Citywide, Hancock beat Calderón in 342, or 96%, of the 356 precincts.  Are the 14 precincts where Calderón beat Hancock high Latinx precincts?  Not at all.  Of the 53 precincts with greater than 40% Latinx VAP, 52 preferred Hancock over Calderón.  The exception was a single precinct (Precinct 251), which was one of seventeen precincts with more that 60% Latinx VAP (Hancock beat all the candidates in 14 of the remaining 16).  In short, no matter how one slices the data, the clear pattern of co-ethnic support at the precinct level failed to emerge at anything resembling the same intensity witnessed in previous mayoral contests with a viable Latinx candidate.  While low voter turnout in precincts with high Latinx VAP did not help Calderón (see this analysis), Hancock’s widespread support in these precincts was clearly a dominant factor in his citywide success (albeit a qualified “success” tempered by the fact that a majority of voters preferred an alternative).

Finally, it is instructive to examine the relationships between Latinx VAP and support for each of the candidates.  The figure below shows the bivariate linear relationships.  Those general patterns, with support for Hancock and Calderón increasing across levels of Latinx VAP, while support for Tate and Giellis tended to decrease, remain even after controlling for the number of eligible voters in a precinct, turnout and an indicator of precinct ideology (support for Clinton in 2016).  More precisely, within precincts of similar size, levels of turnout and ideological orientation, the estimate is that a 10 percentage point increase in Latinx VAP was associated with a 1.4 percentage point increase in support for Hancock, a 0.6 percentage point increase in support for Calderón, and 1.2 and 1.0 percentage point decrease for Tate and Giellis, respectively.  (Details of the analysis are presented in the Appendix).

So what does all this say about the role of Latinx voters in the 2019 General Election?  To start, a caveat is in order.  Each of these analyses say very little about individual voter preferences.  Given the precinct-level data, one can only say certain types of precincts tended to support certain candidates.  Yet the analyses are informative in a number of ways.  First, Hancock’s support was fairly widespread, and he won far more precincts with high Latinx VAPs than other candidates.  This should bode well for him in the general election barring relatively low turnout.  Second, due to a combination of factors, including a field of four high profile candidates and an incumbent with fairly high support in Latinx precincts, Calderón did not muster the level of support among high Latinx VAP precincts as other Latinx candidates in previous elections.  The opportunity structure for a Latinx candidate to emerge victorious was simply absent during this election.  If Hancock wins the runoff, and is ultimately term-limited in four years, 2023 should provide a better set of conditions for a Latinx candidate.  Perhaps that is what Paul López was thinking in his run for Clerk and Recorder—an office with citywide visibility that could serve as a foundation to launch a mayoral run in four years.  Time will tell.

Appendix

Models of Precinct-Level Support by Candidate (Sorry for the Stata output, but I am running out of time to reformat into a proper table):

DV:  Proportion of the Vote for Hancock

DV:  Proportion of the Vote for Calderón

DV: Proportion of the Vote for Giellis

DV: Proportion of the Vote for Tate

Sources:

The data used in the analyses come from a variety of sources.

Turnout total eligible voters data for 2019 were collected from the Denver County Clerk and Recorder’s Office, Denver Elections Division, retrieved on May 8, 2019 from:  https://www.denvergov.org/content/denvergov/en/denver-elections-divison/data.html

Results data for 2019 were collected from the Denver County Clerk and Recorder’s Office, Denver Elections Division, retrieved on May 20, 2019 from:  https://www.denvergov.org/content/denvergov/en/denver-elections-divison/election-archives-and-maps/elections-data-and-maps.html

Data on the Latino VAP for 2019 precincts, 2016 vote totals and the vote for Clinton by precinct were mapped to 2019 turnout data, and generously provided by Bryan Wilcox-Archuleta from UCLA and Michigan State University.  The data was originally used in an article by Wilcox-Archuleta and John Griffin which can be found here: https://thehill.com/blogs/pundits-blog/presidential-campaign/318751-election-autopsy-latinos-favored-clinton-more-than.  Data can be downloaded at:   https://github.com/b-w-a/2016_latino_vote_replication

Talk of the “Blue Wave” at the national level has subsided a bit given the Republican gains in the U.S. Senate and an historically typical shift in seats in the U.S. House of Representatives in the first midterm election in the President’s first term (the Democrats may gain just over 30 seats in 2018, but keep in mind they lost 63 seats in 2010).  Yet, one interesting phenomenon emerging from the electoral tides of 2018 is a solidification of Red and Blue status across states, as conservative and liberal states each hardened their positions in the geopolitical landscape.  Yet, given the victory of the slate of Democrats’ for Colorado’s statewide offices, and what will be a unified Democratic legislature, the blue wave certainly seemed to hit Colorado.

But the breadth and depth of that wave is worth examining more closely, not only in terms of the high-profile races, or the odd shifts in party control for specific local offices (think Adams County Sheriff).  Instead, below we take a look at all four statewide races, the seven Congressional District (CD) races, and all sixty-five State House and seventeen State Senate races in 2018.  If Colorado followed national trends, we might uncover some variation in the effects of the blue wave as partisan lines across legislative districts harden, with some Republicans picking up support as the geopolitical big sort plays out within the state.  Or, if the Blue Wave narrative rings true statewide, then Democratic gains should be both wide (across most districts) and deep (gains of substantial magnitudes).

Let’s start with something simple and compare the percent of the vote in the previous election and 2018’s vote, as well as the margins in each election for the statewide offices and the seven CDs.  The figures presented highlight the breadth of the Democratic gains.  Figure 1 (above) shows how Democrats in 2018 fared relative to Democrats in the previous election (2014 for the statewide offices, and 2016 for the CDs).  The results are striking—in all of these high-profile races, Democrats in 2018 improved upon their support in the previous election.

Given the data in Figure 1, it is not surprising that Democrats also improved their performance relative to Republican or independent opponents.  The percentage point margins for Democrats in both elections are presented in Figure 2, where positive values indicate a Democratic advantage, while negative values indicate the Republican candidate garnered the highest percentage.  In 2018, Democrats either won instead of lost (shifts from negative to positive margins), won by more (higher positive margins), or lost by less (negative values of smaller magnitude) compared to the previous election.  The breadth of these gains is indicative of a widespread blue wave, with not just in statewide races which may mask geopolitical variation, but across all seven CDs with constituencies ranging from primarily urban to suburban to rural constituencies.

To summarize the Blue Wave for the statewide and CD races, Figure 3 displays the percentage point change in the vote for the Democratic Candidate and the percentage point change in the Democrat’s margin from the previous election.  The gains in the level of support for the Democratic candidate ranged from 3.18 to 11.44 percentage points, with an average increase of 6.24.  In terms of margins, the Democrats increased their winning margin (or decreased the margin in a loss) by 10.25 percentage points on average, with a range from 5.75 in the Governor’s race to 19.39 percentage points in the high-profile CD 6 race won by Democrat Jason Crow.  But even in the Republican-dominated CD 4, Democrats narrowed the gap by 10.4 percentage points.   Overall, a good showing for the Democrats across a range of geopolitical contexts.

But the relatively high-profile nature of statewide and CD races, coupled with their larger voter pools, potentially masks some of the variation in the blue wave across Colorado.  State legislative districts, with smaller and more homogeneous constituencies, may have eluded the blue wave.  In order to see if the blue wave was robust across the state, we turn to State House and Senate races and conduct much the same analyses as above.

To save some space, and since the measure captures the various components of the blue wave, we only focus on the change in the Democratic margins for these races.  Change in margins is measured in the same straightforward manner as above, with the percentage point Democratic margin over the nearest competitor in the previous election subtracted from the Democratic margin in 2018.  Positive values indicate the Democratic candidate did better in 2018, while negative values reflect a loss of relative support.  Figures 4 and 5 present the change in margins for the State House and Senate, respectively.

In 77 of the 82 House and Senate races in 2018 (or 94% of the races), the Democrats realized gains in their margins.  On average, those gains amounted to 27.17 and 12.77 percentage point increases from the previous election, respectively; and again, note that these could be gains in winning margins or narrower loses.

But, those five House seats with yellowy-orange negative bars are worth noting.  What’s going on there?  It turns out that four of those districts were uncontested in 2016 while Republicans or independents fielded a candidate in 2018.  In other words, the margin decreased from 100% to something less—the ceiling effect kicked in.  The fifth marginal reduction came from the open-seat HD 50, the seat of Treasurer-elect Dave Young (D) who won by just over 16 points in 2016.  The open-seat meant the loss of the incumbency advantage for the Democrats, resulting in Democratic candidate Rochelle Galindo winning by only 6.5 percentage points (still a respectable victory).  Given these anomalies, and some of the large blue bars that indicate a seat newly uncontested by the GOP (such as HD 32), we ought to re-calculate the average change in margins to only reflect contested seats.  Doing so still results in an average increase in the Democratic margin of 15.61 and 12.75 percentage points in the House and Senate respectively (slightly above the 10.25 percentage point average in the high-profile races discussed above).  And, here again, excluding the uncontested seats held by Democrats in 2016 from the analysis, only in a single House district (HD 50) did the Democratic margin shrink relative to the previous election.

The depth and breadth of the blue wave seems apparent.  From statewide contests to House district races across the state, Democrats in 2018 bettered their performance in the previous election in 88 of the 93 races included in our analysis. The five remaining elections, all resulting in Democratic wins for House seats, are not even suggestive of a GOP ripple.  The take-away is that the blue wave hit throughout Colorado and to a meaningful degree.  Despite many GOP-held seats remaining quite safe, the blue wave clearly washed ashore across the state, from liberal urban core districts to conservative rural districts. Geography and political culture could not hold back the tide.

While Trump will continue to be a factor, what Democrats in the Statehouse do with their newly won power over the next two years will certainly play a role in their chances of maintaining these gains. And regardless, barring a major scandal, war, recession, etc., gains will likely ebb.  The question for 2020 then becomes the magnitude of diminishing margins as they revert to the mean.  Unless, of course, the mean was swept away with Colorado’s blue wave of 2018.

For this first installment of PSC-5280’s perspective on Colorado politics, taking a stab at predicting the state legislative races seems like a reasonably bold introduction to our brand of political analysis.  It fits nicely with our goal of drawing upon insights from political science to explain events in Colorado politics in a straightforward, transparent and accessible manner.  And in making our predictions, we hope to peel away some of the hype and omnipresent spin that coincides with elections as interest groups, pundits, politicos, candidates and funders all point fingers for losses or take credit for the incredible impact they had on a legislative race.

The basic insight we draw upon is fairly simple.  Voters rely on partisan cues (read the “R” or “D” on the ballot) when deciding who to vote for in a state legislative race.  These races, no matter how many yard signs are placed, leaflets are distributed, or doors are knocked, are primarily a function of three factors.  In order of importance, they are: 1) the partisan composition of the district, 2) whether the incumbent is running, and 3) the size of the national “wave” of voter sentiment.  It isn’t rocket science, but it is easy to lose sight of the basics when the hype is nonstop and comes to you 24/7 in the form of your preferred social or web media.

Before we proceed to the predictions, we also want to stress that these predictions provide a way to test some fairly straightforward models of electoral politics.  If one does better than other, we can build on that insight the next time around.  If even these simple models do really well (ie. more than 95% of the seats), then we might show readers that Colorado politics is not so mysterious, corrupt, bought or sold…the election was the simple process of voters expressing their preferences with a ballot.  Whether that happens, of course, depends on how well these predictions, or models, perform (Yes, we’ll revisit their performance after the election, so come on back in a few weeks).

The first theory is simple.  Partisans vote for their co-partisan candidates.  Duh.  But it is powerful and a fairly simple theory to implement in making a prediction.  We simply used voter registration data from the Secretary of State’s Office as of October 1, 2018 to calculate the percentage of each districts’ registered active voters that were affiliated with the Democratic and Republican Parties.  Take the difference between percentage of voters that are Democratic and the percentage that are Republican, and you have what we call the Democratic Advantage.  Making the prediction is based on a simple decision rule.  If the Democratic Advantage is positive, we predict a Democrat wins the election.  If it is negative, we predict a Republican wins the election.

The figures below order each House and Senate District by their Democratic Advantage.  Blue Bars are our predicted Democratic wins.  Red Bars are Republican wins.  And the overall prediction is….drum roll…. Democrats hold the State House of Representatives with a 37 to 28 seat majority, picking up one seat.  For the State Senate, with only 17 seats up for election, Democrats should win 8 seats and Republicans should win 9 seats, with Democrats taking control with an 18 to 17 majority.  Easy.

So, that was simple, but let’s complicate it a bit more.  Incumbency is powerful, and incumbents tend to win (mostly because they run in districts with more of their co-partisans, but that should be obvious if you buy the above predictions).  To account for the incumbency advantage, we add a simple rule:  If an incumbent is running that was elected in the previous election, but the Democratic Advantage indicates the other party should win, we switch the prediction based on the Democratic Advantage for that seat to the incumbent’s party.*  In all, the incumbency advantage model requires we change our predictions for one House district to a Democratic win (HD 59), one Senate District to a Democratic win (SD 5) and two Senate districts to Republican wins (SD 16 and SD 24).  Thus, the incumbency advantage model predicts a Democratic House majority of 38 to 27, and the Republicans holding on to the Senate with an 18 to 17 majority.

Still simple, but what about the “Blue Wave?”  Okay, so national politics could play a role as well. The Trump Administration has fired up the Democrats, but it has also rallied the Republicans.  So, we take a fairly blind stab at the magnitude of the blue wave here and call it a 5% advantage for the Democrats.   While we are somewhat guessing here, it also seems likely that if a national wave hits Colorado’s legislative races, the advantage of incumbency will be swept out along with some Republicans that would otherwise win in a more normal election.  Therefore, we stick with a fairly simple process: Add 5% to the Democratic Advantage and call it the Blue Wave Advantage. Positive values of this Blue Wave Advantage for each district predict a Democratic win, while negative values predict a Republican win. (You can do this on your own by adding 5% to the values in the figures above.)  The Blue Wave model predicts a Democratic House majority of 40 to 25, and a Democratic Senate majority of 19 to 16, with Democrats winning 10 of the 17 seats up for election.

As an aside, if we do account for incumbency with the Blue Wave Model with the same rule as we did above, the prediction is a 39 to 26 Democratic House majority (HD 37 to Republican) but a Republican Senate majority of 18 to 17 (SD 16 and SD 25 to Republican).

The models are simple and the prediction process was transparent.  No smoke and mirrors, expensive polling, or political consultants required, just basic theory-driven political analysis that reflects an age-old axiom in electoral politics:  Know your district. Moreover, the models are based on easily accessible data.**  We’ll see how powerful they are after Election Day.  But as a reminder, the table below summarizes the predictions for easy comparison.

So, there you have it, our first installment of PSC-5280. We hope the take-away is a straightforward reminder of the potential power of understanding Colorado politics through a political science perspective.  Come back in a few weeks to check on our evaluation of these predictions, where we got it right, and where we went wrong.  The election could, after all, differ from all past elections in the history of our state.

*Using the previous election win may normally be a bit stringent.  But we call this the “Lebsock Rule” after former State Representative Steve Lebsock who switched his party affiliation from Democrat to Republican hours prior to being expelled from the House for sexual harassment.  This allowed the Republican Party to appoint a replacement in a district with an 11-point Democratic Advantage.  It is just too odd to not deal with in our predictions.

**If you are interested in replicating these analyses, or perhaps building upon them, you can find the data on the party affiliation of registered voters at the Colorado Secretary of State’s Website and check on who is running and their incumbency status at Ballotpedia.com.

Contact: Prof. Robert Preuhs, Department of Political Science, MSU Denver, rpreuhs[at]msudenver.edu.

Website: http://sites.msudenver.edu/PSC-5280/