# jemdoc: menu{MENU}{turk.html}, showsource
= Voting Research - Mechanical Turk Poll
[http://www.princeton.edu/~cuff/ Paul Cuff] - [http://www.princeton.edu/~kulkarni/ Sanjeev Kulkarni] - Mark Wang -  John Sturm


*[theory.html Click here to read about the purpose of our research (voting theory).]*


== Summary of Results (Condorcet order)

This table summarizes the Mechanical Turk Poll results, with voters separated by party.  The first row lists the winner, followed by the runner-up, and so forth.

~~~
{}{table}{Survey Result}
*All voters*|*Democrat voters*|*Democrat leaning Independents*|*Independent voters*|*Republican leaning Independents*|*Republican voters*||
Obama		| Obama	| Obama	| Obama	| Paul	| Romney||
Clinton	| Clinton	| Clinton	| Paul	| Romney	| Paul||
Paul		| Paul	| Paul	| Clinton	| Gingrich|Santorum||
Romney	| Huntsman	| Huntsman	| Romney	| Huntsman\*| Gingrich	||
Huntsman	| Romney	| Romney	| Gingrich	| Perry\*	| Perry||
Gingrich	| Johnson	| Johnson	| Huntsman\*| Santorum\*| Bachmann	||
Santorum	| Santorum	| Gingrich\*| Johnson\*| Johnson	| Huntsman	||
Perry		| Gingrich	| Wrights\*| Perry\*	| Clinton	| Clinton\*	||
Johnson	| Perry	| Perry	| Santorum\*| Bachmann| Johnson\*	||
Bachmann	| Wrights	| Santorum	| Wrights\*| Obama	| Wrights\*	||
Wrights	| Bachmann	| Bachmann	| Bachmann	| Wrights	| Obama
~~~

(\* indicates a tie or cycle, meaning the order within the \*ed group can be disputed from the data)


== Survey Explanation

Amazon's Mechanical Turk System is becoming a popular way to administer surveys.  We used this system in February, 2012, to obtain responses to a survey about the US presidential election.  The result was 472 completed surveys.  *Most of the survey participants were Democrats or Independents leaning Democrat.*  Please see our [method.html methodology page] to understand how we collected the survey responses and to see the demographic breakdown of the participants.

The survey asks for participants to give a score (between 0 and 100) to each of 11 choices for president of the United States of America, to indicate who they would pick if the choice were up to them.  The choices include declared candidates as well as other such as Hillary Clinton.  With this detailed information about preferences, we can compare who would win the election under several different voting systems.

Our main objective is to see if there is a Condorcet winner---a winner who would win even if any of the other candidates dropped out of the race.  It turns out there is one, in this data and in pretty much all the data we look at.  In this data, *Barack Obama is the winner*, with Hillary Clinton as second choice, according to the Condorcet method.  An interesting finding is that *Ron Paul is the top choice among Republican candidates*, with Mitt Romney next behind him.  The fact that a Condorcet winner exists has profound implications about how we could dramatically improve our voting system, as discussed on our [theory.html voting theory] page.

Since the survey data was gathered throughout the month of February, and we had a large enough number of participants, we can split the data between the beginning and end of the month to look for a shift in the opinions.  Although we won't expand on this analysis here, one observation was noteworthy.  Mitt Romney made double digit progress on narrowing the gap to those ahead of him in the Condorcet order:  Barack Obama; Hillary Clinton; and Ron Paul.  In fact, he beats Ron Paul by one vote in the later part of the month, after he had been trailing Ron Paul by 19\% of the vote in the beginning of the month.


= Comparison of Voting Systems

The results of five different voting methods are shown in these tables, one in each column.  The first row lists the winner, followed by the runner-up, and so forth.

*Please see the first few sections of the [google.html Google Ads Poll page] or the [theory.html voting theory page] for an explanation of the different voting systems.*

Notice how the plurality vote and instant run-off put Ron Paul in second place behind Barack Obama, even though Hillary Clinton is preferred to Ron Paul in this survey sample.  In fact, Clinton loses quite early in the instant run-off system, even though she is the second choice in the Condorcet order.  This highlights the vulnerability of these systems, that competitive candidates can be eliminated early because of splitting of votes.  Hillary Clinton is losing votes to Barack Obama.

~~~
{}{table}{Survey Result}
*Condorcet*|*Plurality*|*Instant Run-off*|*Range*|*Borda*||
Obama		| Obama	| Obama	| Obama	| Obama	||
Clinton	| Paul	| Paul	| Clinton	| Clinton	||
Paul		| Clinton	| Romney	| Paul	| Paul	||
Romney	| Romney	| Clinton	| Romney	| Romney	||
Huntsman	| Santorum	| Santorum	| Huntsman	| Huntsman	||
Gingrich	| Gingrich	| Perry	| Gingrich	| Gingrich	||
Santorum	| Perry	| Gingrich	| Santorum	| Santorum	||
Perry		| Huntsman	| Huntsman	| Perry	| Perry	||
Johnson	| Bachmann	| Bachmann	| Johnson	| Johnson	||
Bachmann	| Johnson	| Johnson	| Bachmann	| Bachmann	||
Wrights	| Wrights	| Wrights	| Wrights	| Wrights
~~~



== Republican Voters

Now we look only at survey responses from Republicans.  Even though Mitt Romney is the undisputed winner, according to the Condorcet method, plurality voting would erroneously elect Ron Paul in this data sample.

~~~
{}{table}{Survey Result}
*Condorcet*|*Plurality*|*Instant Run-off*|*Range*|*Borda*||
Romney	| Paul	| Romney	| Romney	| Romney	||
Paul		| Romney	| Paul	| Paul	| Paul	||
Santorum	| Perry	| Perry	| Perry	| Perry	||
Gingrich	| Santorum	| Clinton	| Santorum	| Santorum	||
Perry		| Gingrich	| Santorum	| Gingrich	| Gingrich	||
Bachmann	| Obama	| Gingrich	| Bachmann	| Bachmann	||
Huntsman	| Clinton	| Obama	| Huntsman	| Huntsman	||
Clinton\*	| Bachmann	| Bachmann	| Johnson	| Clinton	||
Johnson\*	| Wrights	| Wrights	| Clinton	| Johnson	||
Wrights\*	| Huntsman	| Huntsman	| Wrights	| Wrights	||
Obama		| Johnson	| Johnson	| Obama	| Obama
~~~

(\* indicates a cycle, meaning the order within the \*ed group can be disputed from the data)



== Republican-leaning-Independent voters

Next we look only at survey responses from Independents who are leaning Republican.

~~~
{}{table}{Survey Result}
*Condorcet*|*Plurality*|*Instant Run-off*|*Range*|*Borda*||
Paul		| Paul	| Paul	| Paul	| Paul	||
Romney	| Romney	| Romney	| Romney	| Romney	||
Gingrich	| Obama	| Obama	| Gingrich	| Gingrich	||
Huntsman\*	| Gingrich	| Gingrich	| Santorum	| Santorum	||
Perry	\*	| Clinton	| Clinton	| Huntsman	| Perry	||
Santorum\*	| Santorum	| Santorum	| Perry	| Huntsman	||
Johnson	| Perry	| Perry	| Obama	| Obama	||
Clinton	| Huntsman	| Huntsman	| Johnson	| Clinton	||
Bachmann	| Johnson	| Johnson	| Clinton	| Johnson	||
Obama		| Bachmann	| Bachmann	| Bachmann	| Bachmann	||
Wrights	| Wrights	| Wrights	| Wrights	| Wrights
~~~

(\* indicates a cycle, meaning the order within the \*ed group can be disputed from the data)



== Independent voters

We find another controversy among the survey submissions from Independent voters.  Again, the plurality vote is trying to elect Ron Paul (as occurred in the Republican voter data sample), even though there is an undisputed winner according to the Condorcet method.  Barack Obama would beat Ron Paul in a head-to-head comparison.

~~~
{}{table}{Survey Result}
*Condorcet*|*Plurality*|*Instant Run-off*|*Range*|*Borda*||
Obama		| Paul		| Obama	| Obama	| Obama	||
Paul		| Obama		| Paul	| Paul	| Paul	||
Clinton	| Clinton		| Clinton	| Clinton	| Clinton	||
Romney	| Romney\*		| Romney	| Romney	| Romney	||
Gingrich	| Gingrich\*	| Gingrich	| Huntsman	| Gingrich	||
Huntsman\*	| Huntsman		| Huntsman	| Gingrich	| Perry	||
Johnson\*	| Bachmann		| Bachmann	| Perry	| Huntsman	||
Perry	\*	| Johnson\*	| Johnson\*| Johnson	| Santorum	||
Santorum\*	| Santorum\*	| Santorum\*| Santorum| Johnson	||
Wrights\*	| Perry		| Perry	| Bachmann	| Wrights	||
Bachmann	| Wrights		| Wrights	| Wrights	| Bachmann
~~~

(\* indicates a tie or cycle, meaning the order within the \*ed group can be disputed from the data)



== Democrat-leaning-Independent voters

Now we look only at survey responses from Independents who are leaning Democrat.

~~~
{}{table}{Survey Result}
*Condorcet*|*Plurality*|*Instant Run-off*|*Range*|*Borda*||
Obama		| Obama	| Obama	| Obama	| Obama	||
Clinton	| Clinton	| Clinton	| Clinton	| Clinton	||
Paul		| Paul	| Paul	| Paul	| Paul	||
Huntsman	| Romney	| Romney	| Huntsman	| Romney	||
Romney	| Huntsman	| Huntsman	| Romney	| Huntsman	||
Johnson	| Santorum	| Santorum	| Johnson	| Johnson	||
Gingrich\*	| Bachmann	| Bachmann	| Gingrich	| Gingrich	||
Wrights\*	| Johnson\*| Johnson\*| Wrights	| Santorum	||
Perry		| Gingrich\*| Gingrich\*| Santorum| Perry	||
Santorum	| Wrights\*| Wrights\*| Perry	| Wrights	||
Bachmann	| Perry\*	| Perry\*	| Bachmann	| Bachmann
~~~

(\* indicates a tie)



== Democrat Voters

Finally, we look only at survey responses for Democrats.

~~~
{}{table}{Survey Result}
*Condorcet*|*Plurality*|*Instant Run-off*|*Range*|*Borda*||
Obama		| Obama	| Obama	| Obama	| Obama	||
Clinton	| Clinton	| Clinton	| Clinton	| Clinton	||
Paul		| Paul	| Paul	| Paul	| Paul	||
Huntsman	| Huntsman\*| Huntsman\*| Romney| Romney	||
Romney	| Romney\*	| Romney\*	| Huntsman	| Huntsman	||
Johnson	| Bachmann\*| Bachmann\*| Johnson| Johnson	||
Santorum	| Johnson\*\*| Johnson\*\*| Santorum| Santorum	||
Gingrich	| Santorum\*\*| Santorum\*\*| Gingrich| Gingrich||
Perry		| Gingrich\*\*| Gingrich\*\*| Perry| Perry	||
Wrights	| Perry\*\*| Perry\*\*| Wrights| Wrights	||
Bachmann	| Wrights\*\*| Wrights\*\*| Bachmann| Bachmann
~~~

(\* indicates a tie, and \*\* indicates a different tied group)