Analysis

We discuss the final six rows in order.

Averages in the light blue row show that Rick Perry beats Carole Strayhorn by a small margin. Perry also wins (and more convincingly too) if NO OPINION scores are treated as zero.

The green row represents the results after normalization ("stretch" the scores so that the favorite gets a 10, the least favorite gets a 0, and the rest are scaled accordingly), showing that voters would have switched Perry's and Strayhorn's respective placings by normalizing their responses. Strayhorn wins.

The purple row represents the following strategy: if a candidate is preferred to Perry, he gets a 10; otherwise he gets a 0; except if Perry was the voter's favorite, then he gets a 10; otherwise a 0; and any candidates a voter prefers equally to Perry get the same score as Perry. Strayhorn wins. The same strategy applied to Strayhorn causes her to place a distant fourth: Perry, 4; Friedman, 3.49; Bell, 3.13; Strayhorn, 1.72.

The final red row gives median scores with NO OPINION and INVALID scores ignored, and [in square brackets] with them regarded as 0. In Bell's case his two bimedians were 4 and 5, denoted "4-5."

Note that James Werner could not be a valid winner (even if his average were better), under certain quorum rules.

Ranked-ballot voting methods

In the below analysis we regard unranked candidates as ranked below all ranked candidates – which is a common convention which however is unjustified for us, since the unranked candidates actually were "no-opinion" votes.

All of the following rank-order ballot voting methods (based on the 36 votes above) would have declared Perry the winner:
Every Condorcet method, Baldwin, Black, Borda, Copeland, Dodgson, Nanson, Raynaud, Schulze, Simpson, Small, Tideman, IRV.

The results of head-to-head contests were all compatible with the following social ordering: Perry>Strayhorn>Bell>Friedman>Werner.

If Perry were removed from the race (and from all ballots) then all of those voting methods would have declared Strayhorn the winner, except that Borda would have given it to Bell.

However: If we instead (which is more justified) do not regard unranked candidates as below ranked ones – we just regard them as literally unranked – then Strayhorn beat Perry 15 to 14, beat Bell 15 to 10, and also beat everybody else, causing Strayhorn to be the Condorcet winner.

Comparison with official (plurality voting) results

Based on the Edison-Mitofsky exit polls and our range-voting exit poll, it appears that plurality voting "got lucky" in this election and returned the correct winner—Perry. Perhaps, though, the plurality 2nd-place finisher Bell was not the "correct" second-place finisher – that should have been Strayhorn. (But we cannot have any great confidence in that because we only got 36 respondents and only from one precinct in Texas.) It is also possible that Strayhorn would have won a head-to-head race against Perry (which is what happened in our poll).

(Note, for voters who tied two candidates at the top, we here gave half a vote to each.)

We thank David Cary for correcting our mis-entry of ballot #19.

Comparison with press reports about other exit polls: Edison Media Research and Mitofsky International conducted an exit poll of 2077 Texas voters in 25 precincts; where these 2077 also include 700 phone interviews of absentee and early voters. Here are some claims in Associated Press articles about their results. (Margin of error is ±3% and larger for results about subsets.) Perry would have had a 5% edge over Bell in a two-way race. About two-fifths of those who voted for Strayhorn and Friedman would have backed Bell in a two-way race. Bell won about half – the largest share – of voters who oppose the war and disapprove of Bush's performance. A slight majority of Texas voters approved of President Bush's performance and the war; nationally, sentiment was different, with nearly six in 10 voters disapproving of both Bush's performance and the war. Perry also got nearly two-thirds of the votes of "conservatives," those who support the war in Iraq, and those who approve of President Bush's performance. By a slim margin, voters disapproved of Perry's job performance. Fortunately for him, those voters were split among three main challengers; Bell took nearly half the anti-Perry vote, but it wasn't enough. Strayhorn and Friedman together got just as many anti-Perry votes as Bell. More than half the voters surveyed said illegal immigrants should be given a chance to apply for legal status in the United States. Only about one-third said they should be deported. One-third of those surveyed said a fence along the U.S.-Mexico border would do "some" good in reducing illegal immigration. One-quarter said a fence would do no good at all, and nearly as many said it wouldn't help much. Nearly half of the Strayhorn voters and more than one-third the Friedman voters said they wouldn't have voted in a Perry-Bell election. Perry won about three-fourths of the GOP vote despite a challenge from Strayhorn, a longtime Republican (Comptroller) running as an independent. (Strayhorn was also the mother of President Bush's former press secreatary Scott McClellan.) Bell won the votes of nearly two-thirds of Democrats, but he barely won a majority among liberals, and carried only about one-third of Hispanics, who traditionally have backed Democrats. Friedman, an author who ran as the anti-politician, got only about the same percentage of independent voters as traditional politicians Perry, Bell and Strayhorn. But Friedman may have brought new voters to the polls. About one-fourth of his supporters said they would have stayed home if the election had been strictly between Republican Perry and Democrat Bell.

Thanks Beaumont!
The folks at RangeVoting.org

Return to main page