Essay:Analyses of misleading polls

The following is a list of analyses mainly related to the 2020 congressional elections as well as the concurrent presidential election.

August 23, 2020

This essay is an original work by Liberaltears. Please comment only on the talk page.

This poll was conducted by Public Policy Polling for the 2020 U.S. Senate election in Iowa between Republican incumbent Joni Ernst and Democrat challenger Theresa Greenfield, having been administered in mid-August 2020; it had a moderately low sample size of 729 respondents and a margin of error at ±4.1%. While the poll is mostly inaccurate, specifics need to be thoroughly analyzed. According to the poll, Ernst is trailing Greenfield by three points.

Voter registration data

According to official Iowa party registration data listed here, there was a total of 1,996,153 registered voters in November 2016 amidst the 2016 U.S. presidential election, with 662,167 (33%) being Republicans, 629,081 (31.5%) being Democrats, and 694,005 (35%) being Independents. For August 2020, out of 2,012,553 active registered voters, 683,888 (34%) are Republicans, 681,389 (34%) are Democrats, and 633,420 (31.5%) are Independents.

Voter turnout calculations

According to CNN exit polls for the 2016 presidential and Senate elections in Iowa, Republican and independent voters were roughly each split between 34.5% of the makeup (34% for Republicans and 35% for Independents for the presidential race, vice versa for the Senate race), while Democrats made up 31%. In the presidential and Senate elections respectively, it was reported that totals of 1,566,031 votes and 1,539,351 votes were cast. And to calculate for turnout:

Presidential election:
Republican:  (1,566,031 × 34%)/683,888 = 78%
Democrat:    (1,566,031 × 31%)/629,081 = 77%
Independent: (1,566,031 × 35%)/694,005 = 79%

Senate election:
Republican:  (1,539,351 × 35%)/683,888 = 79%
Democrat:    (1,539,351 × 31%)/629,081 = 76%
Independent: (1,539,351 × 34%)/694,005 = 75%

This means that for the voter turnout rates in 2016, around 78% of registered Republicans voted, along with 76% of registered Democrats and 77% of registered Independents.

Calculations

The following table shows the preferences between Ernst and Greenfield by political party, as the poll states:

	Total	Republican	Democrat	Independent
Joni Ernst	45%	82%	7%	45%
Theresa Greenfield	48%	6%	91%	46%
Total	93%	88%	98%	91%

The top of the page also states for the sample makeup that Republicans and Democrats are split with 34% each, while Independent voters make up 32% of the respondents. Based on previous calculations regarding turnout numbers, the following would be expected in the 2020 presidential and Senate elections in Iowa, assuming that the rates are similar to 2016, which may not be the case:

Republican:  (683,888 × 78%) =   533,433
Democrat:    (681,389 × 76%) =   517,856
Independent: (633,420 × 77%) =   487,733
————————————————————————————————————————
                               1,539,022

Simplified numbers:
Republican:  533,433/1,539,022 = 35%
Democrat:    517,856/1,539,022 = 34%
Independent: 487,733/1,539,022 = 32%

After simplifying, the numbers are nearly identical to the ones on the poll. However, there are important details to note: the company from which the poll was administered admitted at one point to be a Democrat-aligned and frequently uses low samples that are skewed towards the left, especially among independent voters. Furthermore, the calculated numbers are based on the presumption that turnout in 2020 would be the same as in 2016. While Ernst may potentially under-perform compared to Trump due to the fact that Greenfield may likely enthuse voters more than the senile, cognitively dysfunctional Biden will, especially among Independents, the incumbent senator has generally been popular in her home state, as most polling and election ratings covering the 2020 Senate elections have mostly "concluded" that some Republican incumbents somehow suddenly lose popularity several months before the general elections.

August 19, 2020

This essay is an original work by Liberaltears. Please comment only on the talk page.

This poll by the Public Policy Polling conducted in mid-August 2020 for the 2020 U.S. Senate election in Maine should set back the hopes of liberals desperate in trying to flipping RINO Susan Collins' seat blue. While the poll comes off as inaccurate due to being administered by a left-leaning company, having a sample size of only 571 respondents and a ±4.1% overall margin of error, crucial data listed in the tables for the poll statistics suggests that the incumbent Sen. Collins has a greater chance of winning re-election than many may expect.

Analysis

Much of the poll appears to echo "conclusions" similarly made by other fake polling. According to the data, President Trump supposedly is unpopular among a majority of voters, etc. The first page states that for the 2020 Senate election in Maine, 44% of respondents intend to vote for Susan Collins, 49% intend to vote for her establishment-backed Democrat opponent Sara Gideon, and that 7% are undecided. Information on the second page states that for sample makeup by political party affiliation, 30% are Republicans, 38% are Democrats, and 32% are independents. The following table shows the preferences by party ID:

	Total	Republican	Democrat	Independent
Susan Collins	44%	87%	5%	48%
Sara Gideon	49%	5%	91%	41%
Total	93%	92%	96%	89%

Another minor though interesting detail to note is Collins' current high popularity among Trump supporters, as the fourth page states that 90% of those who voted for Trump in 2016 intend to back the incumbent senator. The most important detail to note is the middle table on the twelfth page, which shows that Collins leads independent voters by seven points over Gideon. This is in contrast to this previous PPP poll that showed independents evenly split between the two candidates, as well as this poll by administered by a left-wing source that showed Gideon leading Collins among independents by 14 points.

Adjustments for the sample makeup need to be made regarding party ID. In the poll, Democrats outnumbered Republicans by 8 percentage points and outnumbered registered independents by 6 points. According to the Portland Press Herald on August 10, 2020, statistics from the Maine Secretary of State office showed that 27% of registered voters were Republicans, 36% were Democrats, and 32% were Independents. And to adjust the poll by party ID:

Collins: (27% × 87%) + (36% × 5%) + (32% × 48%) = 40.65%
Gideon:  (27% × 5%) + (36% × 91%) + (32% × 41%) = 47.23%

Since the Secretary of State data showed that Republicans were over-weighed in the poll, Gideon's lead increases after adjustments. However, this isn't an accurate method to weigh fake polls! The simple reason for such is that for the state of Maine in the 2016 U.S. presidential election, voter turnout was only 73%. This means that out of all the registered voters statewide, around one-fourths didn't vote in the election. In addition, the statewide turnout percentage fell to only 60% in the 2018 Midterms. Since the 2020 United States Senate election in Maine coincides with the 2020 U.S. presidential election, the turnout would be expected to match that of 2016, if not greater due to potential greater enthusiasm for Trump. Thus, it would be more accurate to adjust the poll by 2016 exit polling for the state of Maine. According to the statistics in this CNN exit poll conducted then, 30% of respondents for it were registered Republicans, 31% were Democrats, and 39% were Independents. In addition, this source shows that the party registration numbers in 2016 in terms of percentages are very close the numbers currently. And to weigh properly:

Collins: (30% × 87%) + (31% × 5%) + (39% × 48%) = 46.37%
Gideon:  (30% × 5%) + (31% × 91%) + (39% × 41%) = 45.70%

After calculating based on the more accurate method of using exit polls, Collins leads Gideon by just under one percentage point. This shows that rather than leaning Democrat, the 2020 U.S. Senate election in Maine slightly tilts towards the incumbent RINO senator's favor, if not a tossup.

July 16, 2020

This essay is an original work by Liberaltears. Please comment only on the talk page.

This poll conducted by CBS should send a panic to the liberal Democrats who have their sights on flipping Arizona blue in 2020. Conducted from July 7, 2020 – July 10, 2020, 1,099 registered Arizona voters were sampled, with the overall margin of error being ±3.8%. The poll "concluded" that Donald Trump is tied with Joseph R. Biden in the 2020 U.S. presidential election and that Republican Sen. Martha McSally was trailing Democrat Mark Kelly by four points; the polling data requires a thorough analysis.

Presidential election analysis

For the 2020 presidential election, the following data was shown on page 13 of the polling information:

	Total	Republican	Democrat	Independent
Donald Trump	46%	90%	3%	46%
Joe Biden	46%	7%	92%	37%
Total	92%	97%	95%	83%

What should send the liberal Democrats panicking already is that Trump is leading among independent voters by nine percentage points over Biden, thus refuting the notion that independents are massively abandoning the Republican Party in the Trump era. And regarding the sample makeup for this section, the same page stated that, of the 1,087 people sampled for this portion, 355 were registered Republicans, 371 were Democrats, and 354 being Independents. Note that this does not add up to 1,087, as the remaining 7 people are presumed to be registered into other political parties.

Weighing by party ID

Since it's assumed for this poll that the ratios by column (excluding the bottom row) in the table shown above are valid and that the samples were weighed improperly, it makes sense to properly weigh the poll.

Current Arizona voter registration

According to the voter registration data in Arizona as of April 1, 2020 here, 34.89% are registered Republicans, 32.51% are registered Democrats, and 31.80% are either registered Independents or of another third party, excluding the Libertarian Party. Already, it becomes clear that registered Democrats were over-sampled in the poll, as 34.13% (rounded from 371/1087) is above 32.51%. In addition, 32.66% (rounded from 355/1087 for Republicans) is less than 34.89%, and 32.57% (rounded from 354/1087 for Independents) is above 31.80%. To weigh properly:

Trump: (34.89% × 90%) + (32.51% × 3%) + (31.80% × 46%) = 47%
Biden: (34.89% × 7%) + (32.51% × 92%) + (31.80% × 37%) = 44.12%

When weighed properly by hard voter registration numbers, it becomes clear that Trump is leading Biden by just under 3 percentage points.

2016 exit polls

While the previous analysis is helpful, it's not sufficient for a full understanding, as a large percentage of registered voters don't vote in U.S. elections. Thus, it's better to use exit polls from the 2016 U.S. presidential election, namely this CNN exit poll that included party ID as part of its polling. According to the data, 32% of voters then were registered Republicans, 28% were registered Democrats, and 40% were registered Independents. Now, to use this to properly weigh the CBS/YouGov poll:

Trump: (32% × 90%) + (28% × 3%) + (40% × 46%) = 48.04%
Biden: (32% × 7%) + (28% × 92%) + (40% × 37%) = 42.80%

According to this, Trump should be expected to win over Joe Biden by 5 percentage points, despite other fake polls showing him losing to Biden by roughly the same margin. And while this should already send a panic to liberals hoping to flip Arizona blue, it's only one-half of the full analysis, with a concurrent Senate election also requiring a close look.

Senate election analysis

For the 2020 U.S. Senate special election in Arizona concurrent with the presidential election, the following data was shown on page 22 of the polling info:

	Total	Republican	Democrat	Independent
Martha McSally	42%	81%	4%	43%
Mark E. Kelly	46%	10%	87%	39%
Total	88%	91%	91%	82%

Similar as with the polling for preferences for the presidential election by political party registration, McSally is leading among independent voters, contrary to what other fake polls have alleged.

Weighing by political party ID

Similar with the previous section, the assumption goes that the data in the chart above is valid in terms of the ratios by column (again excluding the bottom row).

Current Arizona voter registration by political party

Given the data on Arizona party registration as linked above, the following calculations result when weighing the samples properly:

McSally: (34.89% × 81%) + (32.51% ×  4%) + (31.80% × 43%) = 43.26%
 Kelly:  (34.89% × 10%) + (32.51% × 87%) + (31.80% × 39%) = 44.17%

As seen here, when the party ID is weighed properly, Kelly's lead shrinks from 4 percentage points to just under 1%.

Exit polls from 2016 and 2018

While the previous analysis by voter registration data is very helpful, a better tool would be to use exit polls from previous elections, given the large amount of registered voters who don't vote. One good idea would be to use this CNN exit poll from the 2018 U.S. Senate election in Arizona; according to the sample makeup for the voters then by party ID, 38% were registered Republicans, 32% were registered Democrats, and 31% were registered Independents. Now to properly weigh these numbers:

McSally: (38% × 81%) + (32% ×  4%) + (31% × 43%) = 45.39%
 Kelly:  (38% × 10%) + (32% × 87%) + (31% × 39%) = 43.73%

So when the poll is weighed properly as according to 2018 exit poll numbers, McSally is leading Kelly by 1.66%. However, due to the fact that turnout was lower and different in 2018 (a midterm year) as will compare to 2020 (a presidential election year), it's also important to weigh the poll by 2016 exit poll statistics. Using the same one from CNN linked above in here, the following calculations result:

McSally: (32% × 81%) + (28% ×  4%) + (40% × 43%) = 44.24%
 Kelly:  (32% × 10%) + (28% × 87%) + (40% × 39%) = 43.16%

While the lead shrinks down from 1.66% to 1.08%, it still shows that McSally is leading Kelly in the race, despite liberals claiming that the former will lose by potentially 8 to 10 percentage points.

July 3, 3030

See main article: Essay:An analysis of another misleading poll

June 18, 2020

This essay is an original work by Liberaltears. Please comment only on the talk page.

This poll was published by Civiqs/The Daily Kos in mid-June 2020, claiming in the 2020 U.S. Senate special election in Arizona that current Republican interim senator Martha McSally was trailing Democrat Mark Kelly by 9%. Having a sample size of 1,368 people and a margin of error at ±2.9%, there are specifics to be examined and weighed as needed.

Polling data

The following is a table listing the resulted preferences according to the poll between McSally and Kelly by party affiliation:

	Total	Republican	Democrat	Independent
Martha McSally	42%	85%	3%	36%
Mark Kelly	51%	8%	95%	52%
Total	93%	93%	98%	88%

The poll also states regarding its sample makeup that 35% were Republicans, 34% were Democrats, and 31% were Independents.

Weighing

To properly weigh the data by party affiliation, the assumption must follow that the ratios by row in the two-way table are valid, but that the sample makeup is skewed.

Exit polling data

A strong method of finding party affiliation statistics is to use exit polls from recent elections. Since this is regarding the 2020 U.S. Senate election in Arizona, accurate representative data includes using those from the 2018 Senate election in Arizona, although it is important to note that the latter was during a midterm year (see below). According to the CNN exit poll that had a sample size of 1,091 people, 38% were Republicans, 32% were Democrats, and 31% were Independents. According to the Fox News poll that had a sample size of 4,097 respondents, 47% either identified as Republican or "lean" Republican, 41% identifying either as Democrat or "lean" Democrat, and 12% being Independents.

It's important to note that the Fox News poll has greater accuracy in including the "lean" option when asking about party ID; since many registered Independents include partisans who are dissatisfied with their party leadership, they may have greater tendencies to identify by their political ideology though less so the party associated with it. This notion is backed up via the statistics of a previous high-sampled NBC/Marist poll, which found that, in its sample makeup, 22% of respondents identified as "Independent" when there was no "lean" or any likewise option in party affiliation, though only 11% responded as "Just Independents" when a "soft"/"strong" option was included.

Using 2018 CNN exit polling

McSally:  (38% × 85%) + (32% × 3%) + (31% × 36%) = 44.42%
Kelly:    (38% × 8%) + (32% × 95%) + (31% × 52%) = 49.56%

Using 2018 Fox News exit polling

McSally:  (47% × 85%) + (41% × 3%) + (12% × 36%) = 45.5%
Kelly:    (47% × 8%) + (41% × 95%) + (12% × 52%) = 48.95%

Conclusion

When the party ID is weighed properly, it becomes clear that Kelly's lead over McSally diminishes. Where the Civiqs/Daily Kos poll originally put Kelly at a 9% lead, weighing by 2018 CNN exit polling data reduces it to 5.14%, and weighing by Fox News exit polling data conducted in the same year further reduces it to only a 3.45% lead.

Additional note

While it's much more accurate to weigh the poll accordingly to 2018 exit polling data, an important side note is that such doesn't well enough represent the 2020 U.S. Senate election in Arizona, as the latter coincides with a presidential election where regional turnout is greater (and different) compared to the 2018 midterms, which had seen a blue-ish wave following weak GOP establishment leadership during the 115th United States Congress.