What Do The Election Percentages and Numbers Tell? Maybe Nothing

The Assembly chair has said he will appoint an investigator to review the April 3, 2012 Anchorage Municipal election.  A number of people have gathered data.  None of it proves anything.  But there are lots of anomalies, which taken altogether call out for a thorough investigator to determine whether these anomalies are just coincidences or whether they add up to something seriously wrong with the election.

Elections are important because voting is the fundamental basis of our democracy, which is based on the idea that the public determines our political leaders through elections.  The power comes from the people.  We can argue about whether voters are duped by election propaganda and many other ways elections can go wrong.  But most fundamental is that every person has the right to vote and every vote is counted right.

As I looked at the percentages for the various propositions after the election I was struck by wide margins in each case.  It seemed to me that often school bonds and parks and public transit either lost or won by narrow margins.  But not this year.  So I went back to the Muni website and pulled out the data for several categories of city wide bond propositions.  The data come from the Municipal website.  While I tried to be careful, there is always the possibility of mistakes.

The numbers in the Scribd table show bond elections for the last 6 Anchorage elections. (2012 numbers are from the 4/20 final count.)  All of the bonds for 2012 were at the highest percentage "for" votes in the last six Municipal elections. The school bonds tied 2007 and 2008.  The others were record highs.  Two areas that are most likely to be voted down are public transit and general parks and recreation. (The 2008 vote was for swimming pools and there was a very strong parent backed campaign to pass it.  But when parks and bike trail upgrades are in the ballot, it's likely to barely pass or fail.  Even when it's bundled with emergency medical services as it's been often.)

Why would a surge of conservative voters (there to vote no on Prop. 5) vote in record percentages for bonds that normally win narrowly at best or not at all?  Such voters generally vote to keep property taxes down.  It's an anomaly the investigator should check into. 

A larger percentage of people voted for public transit (in with emergency vehicles) than ever before
Anchorage Prop Vote History 2006-2012

(You can play with the Table above here or go to Scribd and download it.)

I'd note the election data are posted by year and some of the results pages show some Propositions more than once - it appears that they had different numbers of precincts in.  I choose the ones with 100%.  You'll also note that transit is usually packaged with emergency vehicles and it's difficult to figure out it includes public transit funds.

You can see all the past election results here.
Since the results pages are vague, I also checked the sample ballots to see what people were voting for. 


The table below looks at the total vote for the 2012 election,

• for each Muni-wide bond issue and
• each Muni-wide office (mayor and school board members), and 
• compares them to the total votes for or against Prop 5, and 
• then compares them to the total number of voters.

Muni-Wide	Votes	Prop 5 Total	Fewer than Prop 5	Total Vote	Fewer than total
Schools	68,376	70,431	2,055	71,099	2,723
Roads/Drains	69,523	70,431	908	71,099	1,576
Med/Transit	69,890	70,431	741	71,099	1,409
Parks	68,246	70,431	1,185	71,099	1,853
Prop 5	70,431	70,431	0	71,099	668
Mayor	69,655	70,431	776	71,099	1,444
School Seat E	55,226	70,431	15,205	71,099	15,873
School Seat F	53,615	70,431	16,616	71,099	17.484
School Seat G	55,921	70,431	14,510	71,099	15,178

HTML Tables

[UPDATE 8:30: I fixed a typo in the Prop 5 column changing 70451 to 70431.  Fortunately it was correct in the chart I did the calculating so the next column should still be ok.]]

You can see no issue or person on the ballot had more people voting than Prop. 5 (to make it illegal to discriminate against GLBT folks).  The difference ranges from 741 to 2, 055 among the bond propositions and 766  to 16,616  among the candidates running.

Despite the fact that Prop 5 got the most total votes (for or against) there were another 688 people who didn't vote on that proposition.

I don't know what this means, if anything.  It makes sense that Prop 5 attracted more voters than any other issue.  It was the most emotional and the one subject to campaigns for and against. 

I do think that whoever does the investigation ought to look at these figures, probably with the help of a statistician, and determine if they are consistent with past elections. Or if the extra numbers and the high percentages in favor of bond issues Conservatives normally vote against could be the result of tampering with the voting machines.

I would also note that the total number of voters - 71,099 is the highest count going back to 2006.  But it's not something the election planners shouldn't have anticipated.  The 2006 election - also a mayoral race - had 70,859 ballots cast.  That's only 240 fewer ballots than 2012.  If the Deputy Clerk was going to consider more than one previous mayoral race to figure out how many voters to expect, she would have had to have seen this figure.

[UPDATE 9:30 am:  I'm presenting numbers here as objectively as I can.  What they mean is much harder to ascertain.  A good example of this data interpretation problem is in a New York Times article today that looks at attempts to get data to determine the US income gap.  It's  not easy to get clean numbers that reflect what you want to know.  But once you get the numbers it's easy to twist them to make them support your position.  Well, easy if your audience isn't asking hard questions.  In any case, the election numbers are pretty easy to get.  They raise questions that require more investigation.  Are the high percentages for increasing bond indebtedness really high enough to raise eyebrows?  If so, is it just an anomaly or is there something wrong in the vote count?  We can ask similar questions about the overall vote counts for each ballot issue.  Given so many signs from different aspects of the election, the investigation coming up is clearly justified.]