# Essay:An analysis of another misleading poll

This essay is an original work by Liberaltears. Please comment only on the talk page.
(see here for mathematical calculations)

This poll by Redfield & Wilton was published on June 25, 2020 covering the 2020 U.S. presidential election and Senate elections in key swing states. For the 2020 U.S. Senate special election in Arizona, the poll "concluded" that Republican interim senator Martha McSally was trailing likely Democrat opponent Mark Kelly by 15 percentage points. Having a sample size of only 865 respondents, there are specifics that need to be analyzed; two-way tables are given for sample makeups in categories for gender, age, region, 2016 vote in the presidential election, and intended vote in the 2020 U.S. presidential election. Of these, the gender and region data will be thoroughly assessed.

## Analyzing gender sample makeup/conclusions

(see here for mathematical calculations regarding this section)

The following is a table that shows preferences between Sen. McSally and Kelly in the Senate election by gender:

Male Female
Martha McSally 34% 34%
Mark Kelly 52% 46%
Other 2% 4%
Don't Know 13% 16%
Total 101%[1] 100%

While the sample makeup is mostly weighed properly (exactly 50% male and 50% female), the real problem here is that the poll is asserting that men are somehow going to overall vote to the left of women by 6 percentage points (in net terms), a mostly absurd notion. In Arizona (as shown in exit polls), men voted to the right of women by a net amount of 17 points in the 2016 presidential election, 11 points in the 2016 Senate election, 4 to 19 points in the 2018 Senate election, and 6 to 15 points in the 2018 gubernatorial election. Furthermore, the entire sample appears to be very left-leaning, making the poll conclusions highly questionable.

## Analyzing region sample makeup/conclusions

(see here for mathematical calculations related to this section)

The following is a table assessing respondents' preferences between McSally and Kelly by their region in Arizona:

Maricopa County Pima County Other Counties
Martha McSally 31% 40% 37%
Mark Kelly 51% 53% 40%
Other 2% 2% 6%
Don't Know 15% 4% 18%
Total 99%[1] 99%[1] 101%[1]

In finding the sample makeup by region, the "Don't Know" section can be ignored. This would then leave three individual rows and columns. With such, a three-variable system can be set up and solved, with the calculated results for sample makeup roughly being 62.15% of respondents from Maricopa County, 14.09% from Pima County, and 24.59% from all other counties (note that this does not add up to exactly 100%[1]). While this does seem to be weighted properly (for the most part), the real problem is that, similar to the gender makeup, the conclusions are absurd. According to the poll, Maricopa County will supposedly vote to the left of Pima County by a net of 7 percentage points, a mostly absurd notion. Pima County had voted to the left of Maricopa County by around 11.4% in the 2018 Senate election, 16.8% in the 2016 presidential election, 18.8% in the 2016 Senate election, 17.8% in the 2012 presidential election, 20.2% in the 2012 Senate election, etc. While it's true that many of the neocons in Maricopa County who supported establishment RINO John McCain and currently oppose Donald Trump are likely to turn against McSally (as McCain donors are), it's unlikely that such numbers will occur. In addition, it's unlikely that such numbers will be the case for the rest of the Arizona counties (that lean Republican), as the entirety of sample appears to be skewed to the left.

## Analyzing 2016 presidential vote sample makeup/conclusions

(see here for mathematical calculations related to this section)

The following table is an assessment of the respondents' preferences between McSally and Kelly by their vote in the 2016 United States presidential election:

Donald Trump Hillary Clinton Other
(Third party)
Didn't Vote
Martha McSally 71% 4% 14% 13%
Mark E. Kelly 15% 86% 46% 43%
Other 2% 2% 17% 6%
Don't Know 11% 8% 23% 37%
Total 99%[1] 100% 100% 99%[1]

Since there are four individuals rows (excluding the "Total") and columns, a four-variable system can be set up and solved, with the calculated results being around 42.36% of respondents having voted for Trump in 2016, 40.78% for Clinton, 2.65% going third party, and 14.78% having not voted then. While the percent ratio between Trump and Clinton is mostly weighed properly, it's important to note that the polling sample for "Other" may be left-leaning and unreliable; since many in the group may have been normally Republican-leaning Mormon voters who possibly voted third party in 2016 (as was especially present in Utah) and there most likely isn't going to be a major third party candidate in the 2020 presidential election, it's unlikely that a much larger percentage of them will vote Democrat.

## Analyzing intended 2020 presidential vote sample makeup/conclusions

(see here for mathematical calculations related to this section)

The following is a table showing respondents' preferences between Martha McSally and Mark Kelly by their intended vote in the 2020 U.S. presidential election:

Donald Trump Joseph Robinette Biden Don't Know Won't Vote
Martha McSally 76% 3% 20% 46%
Mark Kelly 11% 90% 22% 36%
Other 1% 1% 8% 7%
Don't Know 12% 5% 51% 11%
Total 100% 99%[1] 101%[1] 100%

Similar to the previous section, as there are four individual rows (again excluding the "Total") and columns, a four-variable system can be set up and solved to find the sample makeup, with the resulting figures being ~28.65% intending to vote for Trump in 2020, ~40.467% planning to vote for Biden, ~12.7776 answering "Don't Know", and ~18.38% saying that they won't vote in the presidential election. Since it's very unlikely that Biden will win over Trump in Arizona by over 10 points in the general, it becomes very obvious that the entire sample is left-leaning. Furthermore, the conclusions highlighted in this section contradict a previous poll by the same company in late May 2020 which "concluded" that Biden was leading Trump by only 4 points.

## Analyzing age group sample makeup/conclusions

Analysis of the category for sample makeup by age group is being skipped; see here for details.

## Final points

Since the poll overall has appeared to have weighed several groups accurately though selected a left-leaning sample, there would be no practical use to further weigh the poll. Furthermore, because the entire sample is left-leaning, the poll should be distrusted.

## Notes

1. Since the percentages in the two-way tables are much more likely than not rounded from long decimals, the added numbers may not equal exactly 100%.