2020 Nevada Caucuses: How did my final polling averages fare?

Given the extremely volatile polling for the 2020 Democratic presidential nomination now that voting has commenced, I will not provide global monthly updates for the next few months. Instead, I will focus on the first handful of primaries and caucuses: Iowa on February 3, New Hampshire on February 11, Nevada on February 22, South Carolina on February 20, the 14 Super Tuesday contests on March 3, and so forth.

Here is my updated weighting scheme:

  • Polls conducted entirely or partially after February 3, 2020, but before February 12, 2020 are weighted 2.00 or 1.00+fraction[1] times, respectively, higher than polls conducted entirely before February 4, 2020.
  • Polls conducted entirely or partially after February 11, 2020, but before February 23, 2020 are weighted 3.00 or 2.00+fraction times, respectively, higher than polls conducted entirely before February 12, 2020.
  • Polls conducted entirely or partially after February 22, 2020, but before March 1, 2020 are weighted 4.00 or 3.00+fraction times, respectively, higher than polls conducted entirely before February 23, 2020.

Table 1 below lists my final Nevada Caucuses WAPA (weighted-adjusted polling average) for the eight declared Democratic presidential candidates, calculated four different ways:

  • Using all 22 polls conducted since January 1, 2019
  • Using only the 19 polls released since the 1st Democratic debate on June 26, 2019
  • Using only the 10 polls released since the 5th Democratic debate on November 19, 2019
  • Using only the 7 polls released since the 7th Democratic debate on January 14, 2020

Feb 2020 lighthouse

Table 1: Final Nevada Caucuses WAPA for declared 2020 Democratic presidential nomination candidates

Candidate All Polls Since 1st Debate Since 5th Debate Since 7th Debate
Sanders 24.6 24.8 26.4 27.8
Biden 19.6 19.0 16.8 15.5
Warren 13.4 13.4 11.4 11.2
Buttigieg 10.9 11.0 12.6 13.7
Steyer 9.8 9.8 12.2 12.9
Klobuchar 6.5 6.3 8.3 9.4
Gabbard 1.4 1.3 1.5 1.5
Bloomberg 0.1 0.1 0.1 0.0
DK/Other 13.7 14.4 10.6 7.9

Based solely on these numbers, one could reasonably draw the following conclusions:

  • United States Senator (“Senator”) from Vermont Bernie Sanders; former South Bend, Indiana Mayor Pete Buttigieg; Minnesota Senator Amy Klobuchar and billionaire activist Tom Steyer were rising in the polls heading into the Nevada Caucuses.
  • Former Vice President Joe Biden and Massachusetts Senator Elizabeth Warren were declining in the polls.
  • No other candidate was moving in the polls in either direction.

Unlike results from the Iowa Democratic Caucuses conducted 19 days earlier, the following three unofficial, final caucus tabulations were reported by the Nevada Democratic Party within two days (Table 2), though not without some griping:

  1. Initial headcount of support for each Democratic candidate (“Initial tally”)
  2. Post-realignment headcount of support for each Democratic candidate (“Final tally”)
  3. Allocation of “county convention delegates,” or CCD’s, the only measure previously reported [ Note: I mistakenly referred to these as “state delegate equivalents” in the earlier post.]

Table 2: Final Nevada Democratic Caucuses results, February 3, 2020

Candidate Initial Tally Final Tally SDE’s
Sanders 34.0 40.5 46.8
Biden 17.6 18.9 20.2
Warren 12.8 11.5 9.7
Buttigieg 15.4 17.3 14.3
Steyer 9.1 4.1 4.7
Klobuchar 9.6 7.3 4.2
Gabbard 0.3 0.0 0.1
Bloomberg 0.0 0.0 0.0

The following three tables list the arithmetic differences between each candidate’s final Nevada Caucuses WAPA and each of the three reported measures; positive values indicate better performance in the Caucuses than in the polls.

Table 3: Arithmetic difference between Initial Nevada Caucuses % of vote and Nevada Caucuses WAPA

Candidate All Polls Since 1st Debate Since 5th Debate Since 7th Debate Mean

Difference

Sanders 9.4 9.2 7.6 6.2 8.1
Biden -2.0 -1.4 0.8 2.1 -0.2
Warren -0.6 -0.6 1.4 1.6 0.5
Buttigieg 4.5 4.4 2.8 1.7 3.3
Steyer -0.7 -0.7 -3.1 -3.8 -2.1
Klobuchar 3.1 3.3 1.3 0.2 2.0
Gabbard -1.1 -1.0 -1.2 -1.2 -1.1
Bloomberg -0.1 -0.1 -0.1 0.0 -0.1

Initial tally. If the Nevada Caucuses were instead the Nevada Primary, this would have been the only vote reported. On this measure Sanders, Buttigieg and Klobuchar outperformed their final WAPA percentages by 8.1, 3.3 and 2.0 percentage points (“points”), respectively. And for these three candidates, the closer in time the polls were to the Caucuses, the more “accurate” the WAPA. By contrast, Steyer performed an average of 2.1 points worse in the initial tally than his WAPA, with his percentages becoming less accurate over time. United House of Representatives Member from Hawaii Tulsi Gabbard also performed worse than her WAPA, with no noticeable effect of polling period. Finally, WAPA for Biden and Warren were very close, on average, to their initial tallies, with a minimal impact of polling period.

The overperformance of Sanders and Buttigieg aside, these polls were remarkably accurate. This is reassuring given the dearth and quality of recent Nevada Caucuses polling I detailed in the previous post.

Table 4: Arithmetic difference between Final Nevada Caucuses % of vote and Nevada Caucuses WAPA

Candidate All Polls Since 1st Debate Since 5th Debate Since 7th Debate Mean

Difference

Sanders 15.9 15.7 14.1 12.7 14.6
Biden -0.7 -0.1 2.1 3.4 1.2
Warren -1.9 -1.9 0.1 0.3 -0.8
Buttigieg 6.4 6.3 4.7 3.6 5.3
Steyer -5.7 -5.7 -8.1 -8.8 -7.1
Klobuchar 0.8 1.0 -1.0 -2.1 -0.4
Gabbard -1.4 -1.3 -1.5 -1.5 -1.4
Bloomberg -0.1 -0.1 -0.1 0.0 -0.1

Final tally. Only three candidates improved their vote totals after supporters of non-viable candidates shifted to a viable candidate (15% of attendees at a precinct caucus):

  • Sanders (+5,423 supporters; +6.5 points)
  • Buttigieg (+1,496; +1.9)
  • Biden (755; +1.3)

These three candidates also performed better in the final tally than their WAPA, on average; Sanders, in particular, outperformed these WAPA by a remarkable 14.6 points! As with the initial tally, WAPA using more recent polls was most predictive for Sanders and Buttigieg, with the opposite true for Biden. At the same time, Steyer lost 5,383 supporters—and 5.0 points—between initial and final tallies, as did Klobuchar (-2,724; 2.3) and Warren (-1,735, 1.3). For Steyer and Klobuchar, older polls were more predictive, with the opposite true for Warren.

Table 5: Arithmetic difference between Nevada Caucuses CCD % and Nevada Caucuses WAPA

Candidate All Polls Since 1st Debate Since 5th Debate Since 7th Debate Mean

Difference

Sanders 22.2 22.0 20.4 19.0 20.9
Biden 0.6 1.2 3.4 4.7 2.5
Warren -3.7 -3.7 -1.7 -1.5 -2.7
Buttigieg 3.4 3.3 1.7 0.6 2.3
Steyer -5.1 -5.1 -7.5 -8.2 -6.5
Klobuchar -2.3 -2.1 -4.1 -5.2 -3.4
Gabbard -1.3 -1.2 -1.4 -1.4 -1.3
Bloomberg -0.1 -0.1 -0.1 0.0 -0.1

CCDs. The same pattern holds for CCDs as for final vote tally.

  • Outperforming their WAPA were Sanders, by an extraordinary 20.9 points, along with Biden and Buttigieg
  • More recent WAPA were more predictive for Sanders and Buttigieg, and less so for Biden
  • Underperforming their WAPA were Steyer, by fully 6.5 points, along with Klobuchar, Warren and Gabbard, with the difference increasing with more recent polls for Steyer and Klobuchar
  • More recent WAPA were more predictive fore Warren; there was no polling period effect for Gabbard.

The bottom line. To evaluate these comparisons globally, I used the sum of the squared differences (“SSE”) between each WAPA value and the results value. Table 6 lists the SSE for each comparison; higher values indicate lower predictive power.

Table 6: SSE of differences between Nevada Caucuses Resuults and Nevada Caucuses WAPA

Polling period Initial Tally Final Tally CCDs
All Polls 124.0 332.1 551.5
Since 1st Debate 118.4 324.1 541.9
Since 5th Debate 80.8 293.7 508.5
Since 7th Debate 64.1 269.7 481.9

WAPA was most predictive of the initial tally, not surprising given that poll respondents are asked which candidate they planned to support upon arriving at the caucus site, and not about second or third choices. WAPA was more predictive of the final raw tally of supporters than of the distribution of CCDs, though neither was especially predictive; I attribute this to Sanders’ astonishing ability both to add supporters during realignment and to acquire at least 15% of the final tally in nearly every precinct caucus.

For each reported measure, WAPA was more predictive the closer the polls were to the Caucuses; this surprised me, given the candidate-specific differences detailed above. One explanation is that including older polls, however low-weighted, masks late polling movement of the kind that occurred to Sanders and Buttigieg.

For now, however, I will continue to report multiple versions of WAPA, if only to see if this pattern holds for later contests.

Now, on to South Carolina!

Until next time…

[1] Percentage of days the poll was being conducted were after the most recent primary or caucuses

4 thoughts on “2020 Nevada Caucuses: How did my final polling averages fare?

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s