THE POLLS ARE WRONG

SO ARE THE ELECTIONS

Poll results for 2016 Republican presidential candidates show that the methods used to collect poll data are inaccurate. In fact:

THE POLLS ARE BIASED


POLING METHOD TrumpCarsonCruzHuckabeeSantorum
PLURALITY VOTING 25%18%23%2%2%
PLURALITY (no Carson) 25%-32%7%6%
INDEPENDENT VOTING* 25%43%45%40%37%
INDEPENDENT (no Carson) 25%-45%40%37%

Here is an example of how the polls are biased today (winner green):


POLING METHOD ABCD
PLURALITY VOTING 30%23%25%22%
PLURALITY VOTING (no candidate B) 30%-37%33%
INDEPENDENT VOTING* 25%68%70%67%
INDEPENDENT VOTING (no candidate B) 25%-70%67%

Here is another example of how the polls are biased (winner green):


Plurality Voting System polls do not agree with public opinion. The bias is that it wrongly requires people to choose only one candidate.

* Assuming that nobody in the same party would vote NO on anyone in that party.


DETECTING BAD POLLING METHODS

POLL
IS IN
THIS
FORM:
Check (√) only one you like:

Candidate A [ ]

Candidate B [ ]

Candidate C [ ]

Check (√) all you like:

Candidate A [ ]

Candidate B [ ]

Candidate C [ ]

Rank these (1st to 3rd):

Candidate A [   ]

Candidate B [   ]

Candidate C [   ]

Pick from each pair:

Candidate A or B [   ]

Candidate A or C [   ]

Candidate B or C [   ]

Pick Yes, No, or Abstain:

Candidate A   [Y] [N] [A]

Candidate B   [Y] [N] [A]

Candidate C   [Y] [N] [A]

KEY
WORDS
SELECT ONLY ONE
CHECK ONLY ONE
SELECT ALL YOU LIKE
CHECK ANY
RANK THESE IN ORDER
SELECT 1ST TO 3RD
PICK FROM EACH PAIR
PICK BEST EACH ROW
DO FOR EACH CHOICE
PICK YES NO ABSTAIN
IS POLL
FAIR?
NO NO NO NO YES
REASON Requires each person to favor only one candidate. Gives some people more statistical weight than others get. Doesn't indicate whether person likes or dislikes some candidates. Doesn't indicate whether person likes or dislikes some candidates. Collects information correctly.
WHAT
TO SAY
TO THE
POLL
TAKER:
Multiple-Choice is not a valid polling method when people can have multiple preferences. This poll is biased, so it is unscientific. Check what you agree with is not a balanced polling method when the items are related. This poll is biased, so it is not scientific. Ranking does not indicate the point where the subject stops liking candidates and starts disliking them. This poll is biased, and is not scientific. Ranking does not indicate the point where the subject stops liking candidates and starts disliking them. This poll is biased, and is not scientific. This poll is unique. It collects information correctly. You guys know what you are doing.
WHAT
TO PUT
ON THE
POLL
FORM:
Leave the answers blank. Write "Invalid Plurality Voting poll question" in the margin. Check what you agree with. Write "Invalid poll question" in the margin. Select only the choices you agree with. Write "Ranking is statistical garbage" in the margin. Select only the choices you agree with. Write "Ranking is statistical garbage" in the margin. Answer the poll truthfully. Make no notations.

COMPARING CANDIDATE POLLING METHODS

POLL PROPERTIES2ND EXAMPLE ABOVE (%)
METHODEXAMPLEFAIR?ATTRIBUTES AND FAULTS ABCDWIN
PluralityCheck Only One: Cand 1Cand 2Cand 3: NOWon't count multiple preference. 30232522A
Plurality Runoff Uses above poll, 2nd round (method for polls shown below) NOWon't count multiple preference. 30-70-C
ApprovalCheck Any Number: Cand 1Cand 2Cand 3: NORespondent power inequality. 30697067C
Approval Runoff Uses above poll, 2nd round (method for polls shown below) NORespondent power inequality. -6436-B
METHODEXAMPLEFAIR?ATTRIBUTES AND FAULTS ABCDWIN
RankingCheck 1st Choice: Cand 1Cand 2Cand 3: NO Can't tell likes from dislikes. 30232522(1st)
Check 2nd Choice: Cand 1Cand 2Cand 3: 0353233(2nd)
Check 3rd Choice: Cand 1Cand 2Cand 3: 0323335(3rd)
Check 4th Choice:   70101010(4th)
Borda Count Uses above table, scoring = 3(#1)+2(#2)+1(#3) NOCan't tell likes from dislikes. 90171172167C
Instant RunoffUses above, scoring NOOrder reversal possible. 3070→B→BB
METHODEXAMPLEFAIR?ATTRIBUTES AND FAULTS ABCDWIN
PairingsCheck your favorite: Cand 1Cand 2 NO Can't tell likes from dislikes. ### √ 70√ 70√ 70 > A
Check your favorite: Cand 1Cand 3: 30### √ 5249 > B
Check your favorite: Cand 2Cand 3: 3048### 47 > C
Check your favorite: 30√ 51√ 53###  > D
Condorcet scoring: [Count the √s]   Scoring: 0231C
METHODEXAMPLEFAIR?ATTRIBUTES AND FAULTS ABCDWIN
IndependentCheck None Or One: Cand 1YESNO YES Correctly represents opinion. 30687067YES
Check None Or One: Cand 2YESNO 70303030NO
Check None Or One: Cand 3:YESNO [YES - NO]   Scoring: -40384037C

If a poll wants to simulate a system that uses a runoff, it usually collects data for the first part normally, and also collects either a ranking or a pairing set to simulate the runoff results.

Our elections are predicted by systems that do not correctly represent public opinion.




ELECTIONS USE THE SAME STUPID METHODS

Because the elections use the same methods the polls use, they have the same biases.

Since most elections use the Plurality Voting System, polls using the Plurality Voting System will accurately predict the outcome of the election.

But neither correctly represents public opinion.

People favoring multiple candidates are disenfranchised.

It is wrong thinking to expect each person to like only one candidate.

Our officials are elected by systems that do not correctly represent public opinion.



COMPARING GENERAL POLLING METHODS

POLL PROPERTIES
METHODEXAMPLEDATA TYPEFAIR? ATTRIBUTES AND FAULTS
Plurality Rate GroupedCheck Only One: AgreeNeutral DisagreeCategorical YESNo exact value for respondant
Plurality Rate NumberRate from 1 to 3: 1 (low)2 3 (high)Numerical YESValues equally spaced.
Plurality Which is Better?Check Only One: Choice 1Choice 2 Choice 3Categorical NOWon't count multiple preference.
Plurality Runoff Which is Better Uses above poll, 2nd round (method for polls shown below) Categorical NOWon't count multiple preference.
ApprovalCheck Any Number: Choice 1Choice 2 Choice 3:Categorical NORespondent power inequality.
Approval Runoff Uses above poll, 2nd round (method for polls shown below) Categorical NORespondent power inequality.
METHODEXAMPLEDATA TYPEFAIR? ATTRIBUTES AND FAULTS
RankingCheck 1st Choice: Choice 1Choice 2 Choice 3:Categorical NO Can't tell likes from dislikes.
Check 2nd Choice: Choice 1Choice 2 Choice 3:
Check 3rd Choice: Choice 1Choice 2 Choice 3:
Borda Count Uses above table, scoring = 3(#1)+2(#2)+1(#3) Categorical NOCan't tell likes from dislikes.
Instant RunoffUses above, scoring Categorical NOOrder reversal possible.
METHODEXAMPLEDATA TYPEFAIR? ATTRIBUTES AND FAULTS
PairingsCheck your favorite: Choice 1Choice 2 Categorical NO Can't tell likes from dislikes.
Check your favorite: Choice 1Choice 3:
Check your favorite: Choice 2Choice 3:
Condorcet scoring: Count the pairs the choice won
METHODEXAMPLEDATA TYPEFAIR? ATTRIBUTES AND FAULTS
Rating (Range 9)Rate 1 (low) to 9 (high): Choice 1Enter #:_______ Numerical YES Correctly represents opinion.
Tie scores likely
Rate 1 (low) to 9 (high):: Choice 2Enter #:_______
Rate 1 (low) to 9 (high):: Choice 3:Enter #:_______
METHODEXAMPLEDATA TYPEFAIR? ATTRIBUTES AND FAULTS
IndependentCheck None Or One: Choice 1YESNO Categorical YES Correctly represents opinion.
Check None Or One: Choice 2YESNO
Check None Or One: Choice 3:YESNO
Independent scoring: [YES - NO]   Scoring

Links: