PPC polls during the past month

Absentees?

I don't about some of the districts you polled, but I was doing AV chase for one candidate and at least 34% of the vote in my district was absentee.

If a large amount of the vote is starting to trickle in on June 26th through July 7th or so, how does that affect polling?

I know Durant was surging with the later vote, but AV's killed him and took him out of the running.  

"He who would trade liberty for some temporary security, deserves neither liberty nor security" - Benjamin Franklin

---

by: Republican Michigander @ Thu Aug 09, 2012 at 14:24:45 PM UTC

Early AVs should make polling easier.

AVs aren't the problem.  The biggest problem is a campaign that only starts rolling a few days or week before the election, leaving the polling data behind.

In last week's election, 1.5 million votes were cast, of which 425000 (28%) were absentee.  Because the return rate was slightly less than 3% per day, only about 80000 ballots were received by local Clerks through July 15 - 5% of the total vote.

In this election, the absentee ballots seemed to come in later than usual, and roughly 15% were never returned at all.

---

by: Grebner @ Thu Aug 09, 2012 at 22:23:21 PM UTC

[ Parent ]

What's the difference?

How was it that you could call the Democratic gubernatorial nomination race in 2010 with such precision, but have to disclaim this polling?

Among the Trees

---

by: Eric B. @ Sat Aug 11, 2012 at 13:58:35 PM UTC

Why the difference?

I'm not sure, but I have some ideas.

For the statewide gubernatorial primary, I could follow the story line of the campaign, just by reading the papers and talking to political friends.  So I knew how the polling data fit into the narrative.  That is, I knew when Dillon had peaked, and when Bernero was starting to make his move.  I knew that Snyder's lead was large but fragile because I knew that his four opponents were all far to his right, so if their votes ever coalesced, it could overwhelm Snyder's smaller number of moderate votes.  In short, I knew the story.

But for many of these races, all I know is the poll result.  Nerat and Gray are just names to me; I have no idea how their politics differ, or what kind of a campaign they ran.  In other words, I didn't have context.

I don't think any of the results was terrible.  If we knew the inside story, I bet each one of them would be illuminating.

In Ingham, the Drain race was static, and so the polling didn't suffer from trying to track a dynamic race.  So were the three state representative primaries, whose data was never published so I don't include it in the chart, and which we also pegged within a few percentage points.  In other words, PPC did a lot of good work this summer, but not everything worked out.

The Republican Senate race was obviously a story of Hoekstra holding an almost unchallenged dominance until the last two weeks of the campaign, when Durant finally started unloading his money into media.  Knowing that, PPC's results were probably right on target in tracking the race, although they appear "inaccurate"  if you simply compare them to the final vote totals.

Maybe I'd put it this way:  knowing the local story, but not having any polling data, leaves you open to misjudging the situation because of subjectivity or depending on friends who may all share a blind spot.   Let's say that relying on insider gossip is worth one point in some measure of knowledge.

Collecting data in a poll gives you objective information, but unless you are part of the political scene, there is the danger you won't really understand what the data MEANS.  Let's say that's also worth one point.

If you have the good fortune to both know the polling data and also the local story, each kind of knowledge increases the value of the other, and reduces the dangers inherent.  If you add one point for local knowledge to one point for a poll, you get four points worth of predictive accuracy.

---

by: Grebner @ Sat Aug 11, 2012 at 17:15:16 PM UTC

[ Parent ]