Now that most of the dust has settled, we take a look back at our predictions about the overall state climate and specific legislative and congressional races. We’ll take each of our five predictions in order.

1. Republicans sweep statewide races, including the U.S. Senate race, which we project U.S. Sen. Ted Cruz (R) will win by 5 points, and the governor’s race, which we project Gov. Greg Abbott (R) will win by 17 points.

Republicans did indeed sweep the statewide races, and Republican candidates have won every statewide race since 1994. We missed on the margins, which were closer than we projected in every race. Cruz defeated U.S. Rep. Beto O’Rourke (D-El Paso), 51%-48%, and Abbott defeated former Dallas Co. Sheriff Lupe Valdez (D), 56%-42%. These differed from our projections by 2 and 3 points, respectively.

We projected Cruz to defeat O’Rourke, 51%-46%, with Libertarian nominee Neal Dikeman receiving the remaining 3%. We were spot on with Cruz’s number. Dikeman finished with less than 1% of the vote, 2 percentage points below our estimate. O’Rourke received 2 percentage points more than our estimate.

We projected Abbott to defeat Valdez, 57%-40%, with Libertarian nominee Mark Tippetts receiving the remaining 3%. We overshot Abbott’s and Tippett’s support by 1 percentage point each, and underestimated Valdez’s share by 2 percentage points.

We were similarly off in our other statewide race projections, coming closest in SC4, where we projected Justice John Devine (R) to defeat R.K. Sandill, 55%-45%. He won, 54%-46%.

2. The average statewide Democratic candidate will receive 44.4% head-to-head against the average statewide Republican candidate.

Calculated head-to-head, the average statewide Democratic candidate received 46.4% of the vote against the average statewide Republican candidate, a difference of 2.0% from our estimate.

This estimate is the key figure for our model that projects results in each legislative and congressional district. For this election, we modeled 55 districts that appeared in the Top 30 of at least one metric we used to evaluate seats potentially in play. The model’s average error was 2.8 percentage points. Had we accurately predicted the average statewide Democratic head-to-head percentage, its average error would have been 0.8%.

3. HD47 (Workman), HD105 (R. Anderson), HD114 open, HD115 (Rinaldi), HD136 (Dale) and CD7 (Culberson) will flip to Democrat from Republican.

They did. Four of these seats – HD115 (56.7%), HD114 (55.6%), HD136 (54.9%) and HD105 (54.7%) – finished in the top four with respect to the vote percentage received by the Democratic nominee head-to-head against the Republican. CD7 ranked 9th, and HD47 ranked 10th.

4. No Democratic seats will flip to Republican.

None were even close. As we projected, Rep. Victoria Neave (D-Dallas) had the smallest head-to-head margin of anyone defending a Democrat-held seat. She won by 14.1 points.

5. Every other seat will remain in the hands of the party who currently controls it. That said, if the vote breaks the right way in the right places, we would not be surprised to see one or more of SD16, open HD52, HD108, open HD113, HD134, CD23 or CD32 flip to the Democrats.

Broadly, we got most of this right, but we incorrectly predicted seven House, two Senate and one congressional seat would be won by Republicans.

We incorrectly overruled the model in SD10 (Burton), HD45 open and HD113 open, which the model showed flipping even with the 2-point underestimation of the average statewide Democratic candidate.

  • We predicted a narrow Burton win because the strongest early voting turnout occurred in historically Republican precincts. Beverly Powell (D) was ahead, 51%-49%, in the early vote and won Election Day, 53%-47%. Clearly in the district, and across Tarrant Co. generally, Democrats’ performances in those traditionally Republican precincts improved over recent years.
  • We predicted Ken Strange (R) would keep HD45 in the Republican column because the district, while bluer than the state as a whole, was not trending in any discernible direction. We also over-respected Strange’s fundraising advantage. A very strong showing by Democrats in Hays Co. generally contributed greatly here.
  • We predicted Jonathan Boos (R) would hold HD113 primarily because his opponent, Rhetta Bowers (D), had underperformed the Democratic statewide slate by 3.7% two years ago. She also faced a financial disadvantage that recent history suggested made her victory unlikely.

We correctly overruled the model in HD134 (S. Davis) and, at least at this point, CD23 (Hurd).

Four seats – SD16 (Huffines), HD52 open, HD113 open and CD32 (Sessions) – flipped out of the seven that we said would not surprise us should a Democrat prevail. Three of them are in Dallas Co., where Democrats largely swamped the Republicans up and down the ballot, and the other was in Williamson Co., where Democrats turned out in historic numbers.

As we discussed, SD16 was a bit of a blind spot. In 2014, Sen. Don Huffines (R-Dallas) was unopposed in the general election, so we had no sense as to how we would fare relative to the rest of the Republican slate. “A very strong showing by Dallas Democrats could make the difference,” we said. Indeed. HD52 was also a bit of blind spot, because neither candidate had a history from which we could draw any conclusions. In CD32, we gave U.S. Rep. Pete Sessions (R-Dallas) too much credit for his past performances relative to the Republican slate and did not give enough credence to a recent polling showing Allred ahead by 4 (He won by 6.).

The model showed the Democratic challenger coming up short in HD102 even though it was the second-highest ranked district on our “index” of metrics we used to evaluate the potential of districts to flip. Rep. Linda Koop (R-Dallas) outperformed the Republican the statewide slate by 3.3% in 2016, and opponent Ana-Maria Ramos (D) did not meet the fundraising threshold we found was (typically) necessary to flip a seat. However, we did say we that Ramos’s “chances of winning depend on an influx of straight-party Democratic voters.” We expect to find that is exactly what happened.

That leaves the surprises: HD65 (Simmons), HD132 (Schofield) and HD135 (Elkins). We had these three seats rated as “likely Republican,” and the model predicted the Democratic challengers would fall below 46%. Nonetheless, we listed these three seats as “possible ‘sleepers’” along with HD138, which Rep. Dwayne Bohac (R-Houston) holds a lead of fewer than 100 votes, so we at least can say we were not blindsided by these results.

These seats ranked 25th, 29th and 15th on our “index,” respectively.

Elkins’s district, has been evolving quickly. It was the 12th bluest Republican-held district, relative to the state, in 2016. It had the 7th-highest midterm shift, which measured the change in Democratic performance between 2002 and 2014, and the 6th-highest shift in straight-party advantage toward the Democrats of any Republican-held district. It was also one of the most diverse districts held by a Republican.

Schofield’s district, too, is among the most diverse in terms of the voting-age population. Like HD135, his district’s shift toward the Democrats between 2002 and 2014 ranked highly, but the straight-party shift was muted by healthy growth in straight-party Republican voters. We expect the precinct-level data will show both of them were swamped by straight-party Democratic voters this year.

Simmons’s district also rated high when it came to the shift toward the Democrats between 2002 and 2014 but was middle of the pack, at best, in every other metric save one. It was the 16th bluest, relative to the state, of any Republican-held district in 2016.

Overall, 23 Republican-held seats were bluer than the state in 2016. All 16 seats that flipped last week were ranked in the top 21 of that metric. The survivors, at least for now, are U.S. Rep. Will Hurd (2nd), Rep. Sarah Davis (6th), Rep. Angie Chen Button (9th), Rep. Dwayne Bohac (15th) and Brad Buckley (19th) in open HD54. Thirteen of the top 14 Republican-held districts exhibiting the largest mid-term election shifts toward the Democrats (2002 to 2014) flipped. The lone survivor was Davis.

All of the districts that flipped have been shifting gradually more Democratic since they were drawn. Some of the 2018 results reflected a continuation of those historic trends. Others indicated a leap forward, equivalent to several election cycles, in just two years. We will begin diving into precinct-level numbers after the counties release the official results in the next few weeks.

Overall, we correctly predicted the winners of 96% of statewide, legislative, congressional and Board of Education seats, a little below our historical average of 98%. We’ll try harder next time.

©2018 Texas Election Source LLC